A Professional Journal Published by the International Federation of Technical Analysts

25
Inside This Issue

2 Linton Price Targets 48 Using Renko Charts for Noise Reduction and Directional
Insights in the US Equity Market
16 Prediction is Very Difficult: Especially if It’s About the
Financial Markets! 70 The Similarities in Various Markets and Timeframes,
Through Quantitative Comparison Methods, and Its
26 Trend-Adaptation of Moving Averages (TAMA) Application to Trading Systems
36 How the Deltachart Order Flow and Divergence Delta 92 Chicken and Egg: Should You Use the VIX to Time the SPX, or
Candles Work Together to Forecast the Price Movement on Use the SPX to Time the VIX?
High Volatile Market

"You can't put a limit on

anything. The more you
dream, the farther you get."
—Michael Phelps

ISSN 2409-0271
Letter From the Editor
By Dr. Rolf Wetzer, CFTe, MFTA

Dear IFTA colleagues

2024 is an unusual year for IFTA in many ways. For the first time in our history, we have
published two issues of the Journal.
I would therefore like to thank everyone who made this possible. First and foremost, of
course, are the authors. This year, we have a colorful mix of articles from colleagues, MFTA
papers, an educational section from Italy, and our book review from Australia. Many thanks
again this year to the NAAIM and Susan Truesdale, who has allowed us to publish articles
from her own collection for many years. In this issue, this is the contribution of Rob Hanna,
winner of the NAAIM Founders Award. A major contribution to the production of the Journal
comes from Linda Bernetich's production team. Year after year, they produce a publication
with quality that never fails to impress. Finally, I would like to thank Regina Meani and Mohamed El Saiid for their
input in our team.
Our conference will take place this year from October 4th to October 6th
in China, organized by CIDTAA. As always, you can learn something new,
catch up with old friends or make new ones, or simply enjoy the venue.
It is not a novelty, but it is still unusual for our conference to be hosted
by a developing society from the IFTA family. The last time this happened
was 13 years ago in Sarajevo. I would also like to point out that the first
IFTA event in China also took place in 2011. At that time, President Adam
Sorab organized a congress in Beijing in cooperation with official Chinese
authorities. At that time, the entire CFTe Syllabus was presented in
compact form by IFTA experts. This was reported in detail in the two IFTA
Updates, 2011 Vol. 18, Issues 2 and 3. Here is a reminder of the event.
I hope you enjoy reading the Journal and look forward to the many small
discussions that regularly arise from this publication.

Best regards,
Dr. Rolf Wetzer, CFTe, MFTA
 IFTA JOURNAL 2025 EDITION

Letter From the Editor

EDITORIAL
By Dr. Rolf Wetzer, CFTe, MFTA ..........................................................................................Inside front cover

Dr. Rolf Wetzer, CFTe, MFTA

Editor and Chair of the Editorial Committee Articles
Rolf.Wetzer@ifta.org
Linton Price Targets
Mohamed El Saiid, MFTA, CFTe By David Linton, MFTA....................................................................................................................................2
melsaiid@hc-si.com
Prediction is Very Difficult: Especially if It’s About the Financial Markets!
Regina Meani, CFTe
regina.meani@gmail.com
By Davide Pandini, PhD, MFTA, CFTe, CMT, CSTA......................................................................................16
Trend-Adaptation of Moving Averages (TAMA)
By Dr. Patrick Winter ....................................................................................................................................26
Send your queries about advertising
information and rates to admin@ifta.org
MFTA Paper
How the Deltachart Order Flow and Divergence Delta Candles Work Together to Forecast the Price
Movement on High Volatilate Market
By Nashwan Mohammed Al-Thawr, MFTA.................................................................................................36
Using Renko Charts for Noise Reduction and Directional Insights in the US Equity Market
By Loïc Bellina, CFTe, MFTA......................................................................................................................... 48
The Similarities in Various Markets and Timeframes, Through Quantitative Comparison Methods,
and Its Application to Trading Systems
By Nobuaki Kakimoto, CFTe, MFTA.............................................................................................................70

NAAIM Paper
Chicken and Egg: Should You Use the VIX to Time the SPX, or Use the SPX to Time the VIX?
By Rob Hanna................................................................................................................................................92

Book Review
Technical Analysis for the Trading Professional, 2nd Edition by Constance M. Brown
Review by John Gajewski.............................................................................................................................108

Author Profiles................................................................................................................................................ 109

IFTA Staff............................................................................................................................................................111

IFTA Board of Directors...................................................................................................................................111

The IFTA Journal is published annually by the International Federation of Technical Analysts, 1300 Piccard Dr., Suite LL 14 Rockville, MD 20850 USA.
© 2025 International Federation of Technical Analysts. All rights reserved. No part of this publication may be reproduced or transmitted in any
form or by any means, electronic or mechanical, including photocopying for public or private use, or by any information storage or retrieval system,
without prior permission of the publisher.

IFTA.ORG PAGE 1
 IFTA JOURNAL 2025 EDITION

Linton Price Targets David Linton, MFTA

Updata Ltd.
London, United Kingdom
A groundbreaking new way of projecting price targets david@updata.co.uk
www.updata.co.uk
and when they will be met in the future.

Point and figure charts have largely fallen out of favour Cohen also introduced the idea of 45-degree trend lines drawn
in recent decades with the birth of personal computing and from column high and low points as well as unambiguous buy and
electronic data services. Few software systems calculate them sell signals If a column moved one X above the top of the previous
correctly, and the technique is seen as outdated and difficult for column of X’s this was a breakout signal called a Double Top Buy
the newcomer to technical analysis to understand. Linton Price (not to be confused with a standard double top pattern that we
Targets takes the point and figure methodology for producing normally know as a reversal pattern in technical analysis). A
vertical count targets and applies them to time-based charts move of one O below the bottom the previous column of O’s is a
that are much more widely used for technical analysis. Double Bottom Sell. These objective Buy and Sell signals, being
less sensitive than normal (and often only temporary) breaks
History of resistance and support, provided traders with more reliable
Point and figure charts were devised over 100 years ago as entry and exit points for trades.
a necessary shorthand for manually recording prices as they Cohen also devised two methods, the Horizonal Count and
emerged on a ticker tape. Drawing real time price charts by hand the Vertical Count, for projecting price objectives from these
for the most heavily traded stocks proved too difficult, so the new 3-box charts. The upside Vertical Count takes the length of
point and figure method of only recording prices above or below the thrust off a low (number of X’s) and projects an upside price
round levels was born. If prices did not change outside the ‘box’ target of twice that thrust from the top of the column. The target
on the grid, the point and figure chart did not change. When is said to be activated with the Double Top Buy signal on the next
prices moved in line with the direction of previous recorded X column. The downside Vertical Count takes the length of the
prices, a new cross would be drawn in the next box in that thrust off a high (number of O’s) and projects a downside price
column. If the trend in prices changed direction, the price would target of twice that thrust from the bottom of the column. Here,
be recorded in that box in the next column to the right. These the target is activated with the Double Bottom Sell signal on the
charts became known as 1-box point and figure charts. next O column.
In 1947, A. W. Cohen published his book on the Three-Point Figure 1 shows how this rules-based approach appears on a
Reversal Method of Point and Figure. This provided further typical 3-box point and figure chart. This is a daily $1 x 3-box
filtering by introducing an asymmetrical filter whereby prices chart whereby the box size unit is $1. The sensitivity of the chart
had to reverse by at least three boxes in order to move to the next can be increased by making the box size smaller (say a 50c box)
column. This new 3-box method introduced the idea of drawing and by using intra-day price points. The original point and figure
‘X’s for a column of rising prices and ‘O’s for a column of falling charts were constructed with every price tick. This level of a long
prices, which could be further differentiated with different tick price history may not be readily available today, such that
colours. This chart construction means the x-axis of a point and one minute, hourly or daily data may be used.
figure chart is not time, but instead column reversals.

PAGE 2 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 1: 3-Box point and figure chart of Apple, Inc.

The Problem with Point and Figure targets seeks to address is this main shortcoming by placing
Charts Today point and figure style price targets on time-based charts and
Point and figure charts slowly fell out of favour with the projecting them into the future.
birth of modern computing and telecommunications. The need
for this shorthand method of recording price information was Deconstructing Point and Figure to a
superseded with the technological ability to store and retrieve Time-Based Chart
large volumes of real-time and historical price data. Point and To place Point and figure price targets on a time-based chart,
figure charts are also hard to computerise and very few software we first need to relate the conditions that produce the vertical
systems are able to produce them on a computer screen properly. count targets. Figure 2(a) shows a typical Point and figure Double
For the newcomer, point and figure is hard to understand and Top Buy pattern with a vertical upside target generated from a
does not appear to offer additional value to other technical low point in price. A price low in point and figure terms is where
analysis techniques. the base of the column of O’s is lower than the previous column
One of the biggest advantages of point and figure charts is the of O’s. Figure 2(b) shows a schematic diagram of how the pattern
ability to project vertical price objectives. But because there is no in Figure 2(a) might appear on a time-based line chart. Figure 2(c)
time axis on a Point and figure chart, there is no way of knowing shows how a point and figure Double Bottom Sell pattern may
when a vertical price objective may be reached. Point and figure look as a line chart.
price targets have no time scale. The idea that Linton Price

IFTA.ORG PAGE 3
 IFTA JOURNAL 2025 EDITION

Figure 2: How point and figure patterns might look in simple line chart form.

Vertical Targets are only generated with uninterrupted moves Projecting Price
off a high or a low point in prices. A pullback of at least 3 boxes The price projection following the point and figure 3-Box
locks the thrust column and therefore the price target. A move method is relatively straightforward. The standard projection
of at least one box above (in the case of an upside target off a used is twice the original move from the top of the initial thrust
low) or one box below (downside off a high) ‘activates’ the price level. This derives from the 3-Box construction devised by
target. Here the buyers and sellers respectively are confirmed. Cohen, whereby the initial thrust count is a third of the overall
Conversely a move below the base of an upside target column, or price count projection. But there is no reason to limit the Target
above the top of a downside column ‘negates’ the vertical target. Price Factor to the value to 2. A value of 1 could be used in the
In this case, the buyers and sellers have been superseded by case of consolidation patten where the move out of the pattern
subsequent events. is roughly equivalent to the move into the pattern. A value of
1.618 could be used for Fibonacci Retracements or Extensions or
a value of 2 x log, can be used to deal with increasing box (unit)
sizes as price changes. Figure 3 shows an example of a target
factor of 1.0 on the DAX Index.

Figure 3: Using a target projection factor of 1 for the move into and out of a consolidation phase in prices.

PAGE 4 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Projecting Time
Projecting a potential price target with is relatively straight forward. Determining a time in the future when such a price target will be
met is more of a challenge. This has been seen as one of the major drawbacks of point and figure charts for decades. Because there is no
time axis on a Point and figure chart, there is no saying when a count projection target will be met.

Figure 4: The key elements for Price Target construction.

Figure 4 shows the key elements for predicting future prices. The Target Price Level is purely formulaic. For the Time to Target, we
need to consider potential methodologies such as:

1. Price to Time Ratio – t units of price for every x units of time – ie $1 every 2 days
2. Thrust Angle Factor – a factor x the initial trust angle for the target angle
3. Time to Activation Factor – time to target is x the time taken for a target to activate
4. Follow the Price – track prices as the progress to target and adjust time to target accordingly
5. Historical Average Slope – historical average price time average for last n targets

Considering the Price to Time Ratio method, Figure 5 shows a chart of the price targets for the US stock Applied Materials with a
Unit size of $1. The targets are projected Log Scale 2x the initial thrust. From this chart we see that the target prices are reached later
than the projection predicted. This means that we need to consider a lesser slope. Figure 6 shows the same chart with the slope now
adjusted to $1 every three days. This chart shows that recent targets for Applied Materials have been approximately met with this slope.
Therefore, this is a better slope to use in this instance.

IFTA.ORG PAGE 5
 IFTA JOURNAL 2025 EDITION

Figure 5: Applied Materials with price targets with a unit size $1 – target projection slope $1 every two days.

Figure 6: Applied Materials with Price targets with a unit size $1 – target projection slope $1 every three days.

PAGE 6 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 7: Applied Materials (unit size $1) – target projection slope 1/2 initial thrust slope.

The second method of projecting price targets assumes the time that a price target will be reached is directly related to the speed of
the initial thrust, which generates the target. Figure 7 shows the same security as in the previous examples but using this method with
an angle of slope which is half the initial thrust angle. The factor can also be altered with this method to best fit the data. In the previous
examples (Figures 5 & 6) we see the slope of each of the targets is constant. Using the Thrust Angle Factor method, different buying and
selling thrust angles produces different target slopes.
A third possible projection method assumes that the longer a price target takes to activate, the longer it takes for a target to be
reached. The argument goes that the pullback from the initial thrust is more of a consolidation phase rather than a sharp reaction and
therefore, the potential overall move will take longer. Figure 8 shows this method. Again, we see that, due to the varying times of price
targets to activate, the slopes of the targets are not uniform as in Method 1 which uses a consistent price to time slope.

Figure 8: Applied Materials (unit size $1) – target projection x times the time taken for target to activate.

IFTA.ORG PAGE 7
 IFTA JOURNAL 2025 EDITION

Figure 9: Applied Materials (unit size $1) – target projection readjusts with new price information.

A fourth method for predicting when in the future that a price target might be met adjusts the slope of the targets from the activation
point as new price information arrives. With multiple targets activated at different points on the chart, this method also produces price
targets of different slopes. Because targets are readjusted with every new price, it is best to set this method to ignore the last x bars in
order to spot any divergence from the targets. Figure 9 shows this methodology.
Figure 10 shows a method where the average slope of price over time is taken for the previous n targets that are achieved and used as
the slope for projecting targets into the future. While the slopes for upward and downward targets can be separately adjusted with the
previous methods mentioned, this method automatically calculates the different slope speeds of upside and downside targets.

Figure 10: Applied Materials (unit size $1) – target projection based on the average slope of the last x targets.

PAGE 8 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Multiple Price Targets

As with Point and figure count targets, multiple price targets point to the same price or price level increases the likelihood of price
targets being met. This is known as ‘clustering’. Now with the ability to project price targets to a future date on a chart, it is not only
possible to see clustering of the price of multiple targets, but also clustering of times targets may be met. This can lead to a ‘cluster zone’,
an area of price and time in the future that multiple targets may be met. Figure 11 shows an example of this.

Figure 11: Applied Materials (unit size $1) – target zone of future price and time of multiple targets.

Achievement and Non-Achievement of Price Targets and Prevailing Trend

Point and figure targets are approximate and are more often than not, not met precisely. They are regularly not achieved or exceeded,
but this provides valuable information in itself. Upside price targets that are achieved or exceeded shows bullish confirmation, whereas
these targets not being achieved indicates a degree of bearishness. Conversely, downside price targets achieved or exceeded is bearish
confirmation and such targets not achieved is an indication of inherent bullishness.
Unsurprisingly, price targets are normally achieved or exceeded in line with the prevailing trend. Upside price targets should be given
more weight in uptrends, while downside ones may only serve as a temporary moment for caution, because they are counter-trend.
Downside Targets will carry more weight in downtrends. It is also often the case that the last target in line with the prevailing trend is
never met as the trend changes and a new set of targets in the opposite direction are generated with the new reversal of trend. Active
price targets in both directions areoften an early sign of this. This is particularly true with multiple targets in the new trend direction
verses one lone target in the previous trend direction. This lone target is likely to be negated, clearly signalling the new trend direction is
taking hold.

Activation and Negation of Price Targets

An upside price target is only activated when prices rise a further than a full price unit above the top of the initial uninterrupted
buying thrust in prices from a low. A low is defined by a price level at least one full price unit below a previous recent low. The pullback
downwards of at least three price units ‘locks’ the initial thrust that generates the upside price target. Here the bulls buying from the
bottom have been confirmed.
A downside price target is only activated when prices fall further than a full price unit below the bottom of the initial uninterrupted
selling thrust in prices from a high. A high is defined by a price level at least one full price unit above a previous recent high. The pullback
upwards of at least three price units ‘locks’ the initial thrust that generates the downside price target. Here the bears selling from the
top have been confirmed.
A target is valid once the column is locked with the pullback of at least three units, but it should not be considered as active until the
price breaks through the activation level. An unactivated target serves as advance notice that a target is in place and will become active
once the activation price level is broken.

IFTA.ORG PAGE 9
 IFTA JOURNAL 2025 EDITION

An upside price target is negated if prices fall below the bottom of the initial uninterrupted buying thrust in prices. In this instance
the bulls have been beaten by the bears. Conversely, a downside price target is negated if prices rise above the top of the initial
uninterrupted Selling thrust in prices. Here the bears selling from the top have been beaten by the bulls.
It is important to note the difference between a target that is activated first and then negated and a target that was never activated
and negated first. Research shows that normally more than half of all negated targets were never activated and wouldn’t have been
taken. Taking the prevailing trend into account further reduces the number of negated targets that would have been taken at the
activation point. Figure 12 shows the difference between targets activated then negated and not activated and then negated.

Figure 12: Showing 11 targets activated then negated versus 13 negated never activated.

Evaluating a Target as Price Progress consider that the target has been ‘de-activated’. If we fall further
Because Linton Price targets can be evaluated with subsequent below the low of the pullback low point, this previous support
new price information with the passage of time, it becomes level also failed to hold and this is providing us with an early
possible to see more easily, than on a point and figure chart, when warning that the target is quite possibly ‘failing.’ If prices are
a target might be failing. The ideas of activation, negation, and moving towards the target as expected, we can say the target is
achievement of price targets are understood in point and figure ‘in train.’ This is particularly appropriate for multiple targets that
charting and apply similarly here to time-based charts. But the run parallel using the first price/time slope prediction method
ability to now see prices diverging from the target path presents where the targets look like ‘train tracks.’ Figure 13 shows these
us with some potential new states of a target. target states for an upside price target and Figure 14 shows them
In the case of an upside target, if prices fall away or wander for the reverse case of a downside price target.
sideways from a target path this alerts us to the fact that the
prices on their way to the target may be ‘exhausting’. If we fall or
wander back below the target activation level, this implies the
previous resistance level off the thrust high has not managed to
become a new support level for the price. Consequently, we may

PAGE 10 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 13: Target states for an upside price target.

Figure 14: Target states for a downside price target.

IFTA.ORG PAGE 11
 IFTA JOURNAL 2025 EDITION

Improbable Targets
Occasionally an improbable target a long way from the price will be generated. This is particularly true using a log scale projection.
Beware of a target that points to a very large change in price. This is especially true of a lone target. It is also quite likely that the unit size
has been set too small where a bigger unit size may not produce a target at all. Figure 15a shows an improbable target and a check of the
point and figure chart for the same instrument with matching unit and box sizes, 15b, shows the box size used is much too small.

Figure 15a: Example of an improbable Target. Figure 15b.

Longer term charts based charts it becomes necessary to look at weekly or monthly
Point and figure charts have always meant to be constructed data. Figure 16 shows a daily point and figure chart of gold over
with tick data. The point and figure methodology reduces this the past ten years on the left. The chart on the right is a monthly
down to just the ticks that create a new box on the chart. Long chart of gold over the past ten years. Here a daily chart would
tick data price histories are typically expensive and hard to come be very noisy with thirty times more data points. Monthly time
by. This can also be an overwhelming amount data to store and series analysis can also be applied, in this case an Ichimoku cloud
analyse, particularly in the case of very liquid instruments such chart.
as a major currency pair. For intraday charts, one minute data We also see in Figure 16 that long term price upside targets
will normally suffice. But these histories may not be long enough are generated that are not on the daily chart. This is because
either and it may be necessary to use a 60-minute chart. daily the movements will not provide the same uninterrupted
It is also possible to construct point and figure charts using buying thrusts as with the monthly data. The daily pullbacks
high/low data or even open-high-low-close data making some are effectively ignored when using monthly data. The other
assumptions based on a rising or falling candle, on which advantage is the unit size is now months so we can say that the
came first, the high or the low. The targets will be impacted target slope equates to 1% of price every month for a 1 to 1 slope
accordingly. for example. Using weekly or monthly data to construct the price
When it comes to longer term charts such as weekly or targets is a significant departure from the traditional point and
monthly charts it is unlikely that these time frames would figure charting method.
be used for point and figure charts. The construction method
already filters the data. But when it comes to long-term time-

PAGE 12 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 16: Daily point and figure chart of gold and a monthly Ichimoku chart with price targets.

Time-Based Charts Are Easier to Understand Than Point and Figure Charts
In recent years, the vast majority of people carrying out technical analysis of charts do not use the point and figure charts. This is
partly because very few software systems draw them correctly and do not calculate the price targets. Newcomers to technical analysis
find point and figure charts hard to understand. Figure 17a shows the Linton Price Target for Apple, Inc. and Figure 17b shows the
corresponding point and figure chart. Most technical analysts would prefer the first chart over the second.

Figure 17a: Linton Price Target for Apple, Inc. Figure 17b.

Combining With Other Techniques

Using point and figure charts has also often meant the need to switch between different chart types for the same instrument.
Time-based charts allow for a vast set of technical analysis time-series based techniques to be married with Linton Price Targets. For
instance, Figure 18 shows a 60-minute chart of silver with Bollinger Bands and the price targets on the same chart. Having different sets
of analysis on the same chart can increase the power of the analysis without having to swap between different chart types.

IFTA.ORG PAGE 13
 IFTA JOURNAL 2025 EDITION

Figure 18: Bollinger Bands with Linton Price targets for 60-minute chart of silver.

Conclusion
Linton Price Targets builds on the technical analysis body of
knowledge developed over the past 100 years by bringing an old,
largely lost, technique into the modern age.
The main advantages of Linton Price Targets are:

• The ability to have price targets on time-based charts.

• It is now possible to ascertain when in the future a price
target may be met.
• With the passage of time, it becomes clearer if a target track
is being followed.
• The targets can be applied to longer-term time-based charts.
• Time-series based analysis techniques can be used on the
same chart as the targets.
• The targets are much easier to understand for the newcomer
to technical analysis.

PAGE 14 IFTA.ORG
 IFTA JOURNAL 2025 EDITION
Official Trading Partner of the ATAA

Master the market

forces with City Index
Forex | Indices | Crypto | Commodities | Share CFDs

All trading involves risk.

IFTA.ORG PAGE 15
 IFTA JOURNAL 2025 EDITION

Prediction is Very Difficult: Especially if It’s

About the Financial Markets! Davide Pandini, Ph.D., MFTA, CFTe, CMT, CSTA
Ronco Briantino, Monza e Brianza, Italy
davide.pandini@hotmail.com
+393483973247
A rigorous evaluation of statistical forecasting
methods and their comparative effectiveness.

"… Prediction is very difficult, especially if it’s about the future! …" practitioners, and academics dedicated to advancing the science
Niels Bohr (1885-1962), Danish physicist, Nobel Prize in Physics of market prediction within the realm of Technical Analysis.
(1922)
2. Forecasting Models
Forecasting models consist of time series prediction by
1. Introduction to the First Part identifying the trend, seasonal and cyclical patterns in the
In the ever-evolving landscape of financial markets, financial markets [1][2][3]. The trend component in time
accurately forecasting asset prices and market trends remains series forecasting models captures the long-term movement
a critical challenge for analysts, traders, and investors. This of financial data. In the context of financial markets, this
comprehensive essay, now divided into four parts, presents represents the direction in which an asset's price, or a market
a detailed exploration of statistical forecasting methods, index, is heading. Identifying and understanding trends is
rigorously evaluating their effectiveness and comparative critical for investors and traders as it offers insights into
advantages in the financial sector. This first part introduces whether a market is in an upward, downward, or sideways
fundamental concepts of forecasting models, providing a trajectory. Seasonality is the recurring fluctuations in financial
general introduction to those models and their components, data driven by calendar-based or periodic effects. These
describing the financial asset and time series data utilized, patterns often follow a consistent annual, quarterly, or monthly
and explaining the segmentation of the historical backtesting schedule. For instance, retail stocks may experience increased
period into training and testing ranges. It also discusses model demand during the holiday season, leading to predictable
selection criteria and accuracy metrics and introduces simple seasonal spikes in their prices. In finance, identifying and
forecasting methods as a baseline for understanding more accounting for seasonality is crucial for understanding when
sophisticated techniques. certain assets tend to perform better or worse. Cyclical patterns
This initial focus on simple forecasting methods sets the in financial time series data reflect the medium to long-term
stage for understanding more complex models, making it fluctuations resulting from economic cycles. These cycles
crucial for financial technical analysis as it lays the groundwork encompass periods of economic expansion and contraction,
for accurate and reliable predictions, essential for informed which can last several years. In the financial context,
decision-making in the financial sector. Subsequent parts recognizing these cycles helps investors and policymakers
will delve deeper into more advanced methods. Part Two navigate volatile markets, adjust strategies, and anticipate
will explore exponential smoothing methods, offering a market turning points. In this work we will consider only trends
comprehensive analysis and discussing procedures for optimal and seasonality patterns.
parameter selection. Part Three will focus on ARIMA models
and their variants, providing detailed discussions and practical Additive Forecasting Models
implementations of non-seasonal and seasonal ARIMA models. Additive models are time series forecasting models used to
Finally, Part Four will address advanced forecasting methods, predict future values of a time series by considering the additive
incorporating volatility modeling through GARCH models and combination of its various components. The observed value y t
tackling the challenges of non-Gaussian financial data. can be expressed as:
By beginning with the simple methods analyzed in this paper,
readers can build a strong understanding before progressing
to more complex models, ensuring a comprehensive and
structured exploration of statistical financial forecasting. This
segmented approach ensures that each part provides a focused where the three components are:
examination of forecasting methods, paving the way for reliable 1. Trend Component tt: The trend component represents the
and insightful financial predictions and decisions. underlying long-term direction or pattern in the time series
This extensive analysis will be featured in the IFTA Journal, data. It accounts for the gradual increase or decrease in
starting with Part One in the 2025 issue and continuing with values over time;
the subsequent parts in following issues. The publication of this 2. Seasonal Component st: The seasonal component captures
work in the leading journal of technical analysis underscores the the recurrent, short-term patterns that repeat at regular
significance and relevance of these topics to the field of financial intervals, such as daily, weekly, monthly, or quarterly. These
forecasting. It aims to reach a broad audience of professionals, patterns are often associated with calendar-based or periodic

PAGE 16 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

effects, like holidays or seasonal demand; • Useful for Relative Proportions: Multiplicative models are
3. Error Component et: The error component, also known as appropriate to capture relative proportions or growth rates
the residual or noise, represents the random fluctuations within the components.
or unexplained variations in the time series that are not
accounted for by either the trend or seasonal components. It The limitations of the multiplicative models can be
is typically assumed to be normally distributed with a mean summarized as follows:
of zero.
The advantages of additive models can be summarized as • Complex Interpretation: Multiplicative models may be less
follows: straightforward to interpret than additive models, as they
involve multiplications rather than additions;
• Interpretability: Additive models allow for a clear • Challenging Forecasting: These models can be more
interpretation of the individual components (trend, challenging to forecast with, especially when dealing with
seasonality, and error) and their impact on the time series; long forecasting horizons.
• Ease of Implementation: These models are relatively The trend component, tt, and the seasonal component, st,
straightforward to implement and do not require complex are multiplied together, implying that the seasonal effect
mathematical techniques; becomes proportionally larger or smaller as the trend increases
• Useful for Linear Patterns: Additive models are well-suited for or decreases. Multiplicative seasonality is appropriate when
time series data with linear or additive patterns. the magnitude of the seasonal fluctuations changes with the
trend, and it is often used when the seasonal patterns exhibit
The limitations of additive models are: relative growth or decay over time. In summary, multiplicative
forecasting models are valuable for predicting time series data
• Not Suitable for Non-Linear Patterns: When the relationships when the trend, seasonality, and error components are believed
between the components are not additive, these models may to have a multiplicative relationship, and they are better suited
result in poor forecasts; for non-linear patterns and time-varying variance.
• Inadequate for Time-Varying Variance: Additive models Forecasting model patterns have the following graphical
assume constant variance, which may not hold true for time representations. A multiplicative trend occurs when the rate
series with changing volatility. In such cases, models like of change in a time series is not constant but instead varies
GARCH might be more appropriate; in proportion to the current level of the data, as it is shown in
• Limited Forecasting Horizon: Additive models may not Figure 1.
perform well when forecasting too far into the future, as they
do not account for the compounding effect of the components. Figure 1. Liner vs multiplicative trend

Therefore, additive models can be particularly useful when

the impact of these components on the time series is considered
additive or linear.

2.2 Multiplicative Forecasting Models

Multiplicative models are time series forecasting models
used to predict future values of a time series by considering
the multiplicative combination of its various components, as
represented by the following expression:

The advantages of the multiplicative models are: A dampened linear trend combines elements of both linear
and multiplicative trends. It represents a linear trend that
• Suitable for Non-Linear Patterns: Multiplicative models gradually decreases or dampens over time. In other words, the
are well-suited for time series data with non-linear or rate of change is linear, but it diminishes as time progresses, as
multiplicative patterns, where the impact of the components illustrated in Figure 2.
is proportional rather than additive;
• Accurate for Time-Varying Variance: These models are better
at handling time series with changing variance or volatility,
making them useful for financial data, which often exhibits
such behavior;

IFTA.ORG PAGE 17
 IFTA JOURNAL 2025 EDITION

Figure 2. Linear vs damped trend

made to the closing prices of a financial asset, usually a stock
or an index, to account for various corporate actions such as
dividends, stock splits, and rights offerings. These adjustments
are made to provide a more accurate representation of the
asset's value over time and to ensure that historical price data
are consistent and comparable. By using the adjusted close
prices, financial analysts and investors can accurately analyze
the historical performance of a stock over time, even in the
presence of corporate actions that might otherwise distort the
raw price data. These adjustments help maintain consistency
in price trends and facilitate meaningful technical analysis,
charting, and performance comparison.
The SPY adjusted close prices time series is divided into a
training range for optimal parameters estimation or fine tuning,
and a testing range for forecasting accuracy and robustness
evaluation. The training range covers 3,424 observations, from
In additive seasonality patterns, the seasonal component is December 1st, 2004, to July 9th, 2018, while the testing range
added to the overall trend and error components of the time includes 1,273 observations, from July 10th, 2018, till July 31st,
series. This means that the seasonal fluctuations are relatively 2023. The training and testing range charts are shown in Figure
constant and do not change in proportion to the trend or the 4 and Figure 5 respectively.
level of the data. Additive seasonality is commonly observed
when the seasonal effect is fairly consistent from one season Figure 4. SPY ETF training range (Dec 01, 2004 – Jul 09,
to the next. In a multiplicative seasonality pattern, the 2018)
seasonal component is multiplied by the overall trend and error
components of the time series. This means that the seasonal
fluctuations are proportional to the trend or the level of the
data. Multiplicative seasonality is commonly observed when
the seasonal effect grows or shrinks with the overall trend, as
shown in Figure 3.

Figure 3. Additive vs multiplicative seasonality

Figure 5. SPY ETF testing range (Jul 10, 2018 – Jul 31,
2023)

3. Forecasting Models Data

The forecasting models’ data used in this work is based on
S&P 500 State Street Exchange Traded Fund ETF daily prices
(ticker: SPY) spanning a time period of about nineteen business
calendar years (from December 1st, 2004, to July 31st, 2023:
4,697 observations). The historical data time series used in this
work was downloaded from Yahoo! Finance [4]. The "adjusted
close prices” have been considered: they refer to a modification

PAGE 18 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

The choice of the time window for testing a forecasting model accurate forecasts in various contexts (i.e., on an unknown
or a trading strategy has always been of interest and concern to testing set) and has the capability to generalize to new data.
the technical analysts to assess its robustness and accuracy, since
different periods and sizes of the window can lead to different In this work, to evaluate the accuracy and robustness of the
results and conclusions. The study presented in [6] assessed the forecasting models, we use the testing set, which is a critical
robustness of the performance of a strategy given the window component for assessing not only the accuracy, but also the
size of the backtesting period. This study shows the impact that robustness of a model's predictions.
the chosen window can have on the results and as such, the
authors argue that the window should not be arbitrarily selected. 4.1 Robustness evaluation using the testing set
In [7] it was demonstrated that an active market timing strategy When evaluating the robustness of a forecasting model, we
outperforms the passive buy-and-hold strategy during bear examine how well the model performs under various challenging
markets and vice versa during bull markets. To account for these conditions, such as changes in data distribution and structure,
results, the study in [7] concluded that the look-back period should the presence of outliers, or shifts in the underlying patterns.
include bear and bull markets to observe both these market The testing set, which consists of new and unseen data for the
conditions. Therefore, the forecasting models analyzed in this model, provides an ideal platform to assess how the model
work were tested across an historical data length covering the responds to changing scenarios. By testing the model in diverse
past nineteen years (from December 2004 to July 2023), because situations represented in the testing set, we can determine
it includes multiple bull and bear markets, some of which were whether the model's accuracy and performance remain stable
quite significant, like the global financial crisis (2007-2008) and or degrade significantly. If the model's performance on the
the more recent COVID-19 sell-off (March 2020), and the last bull testing set remains consistent across different conditions, it is an
market lasting over a decade. indication that the model is robust. Therefore, using the testing
set to evaluate robustness is an excellent practice. It helps to
4. Forecasting Accuracy Metrics and ensure that the forecasting model is not just accurate under ideal
Model Selection Criteria conditions (i.e., after optimization performed in the training set),
The evaluation of forecasting models becomes pivotal in but it can also provide reliable predictions in real-world scenarios,
determining their effectiveness, therefore forecasting accuracy where the data might not perfectly match the assumptions made
metrics and selection criteria are instrumental not only to assess during model’s training.
the effectiveness, but also the robustness of any forecasting
model. It is important to note that effectiveness and robustness 4.2 Accuracy evaluation using the testing set
are crucial when evaluating forecasting techniques, but they The accuracy of forecasting models is not a simple binary
address different aspects of a model's performance and behavior: concept, and it encompasses several measurements highlighting
different aspects of performance; thus, it is essential to have a
• Effectiveness refers to how well a forecasting model comprehensive understanding of the various accuracy metrics
accurately predicts future outcomes based on historical available and their implications. Forecasting models’ accuracy
data. In other words, it assesses the model's ability to is evaluated through scale-dependent and scale-independent
generate forecasts that are close to the actual values. A metrics. Scale-dependent accuracy metrics are used to assess
forecasting model is considered effective if it consistently the accuracy of forecasting models in relation to the actual
produces accurate predictions over time, minimizing the values of the target variable. These metrics provide insight into
discrepancy between its forecasts and the actual outcomes. how well the model's predictions match the observed data on an
Accuracy metrics such as Mean Absolute Error (MAE), Root absolute scale. On the other hand, scale-independent accuracy
Mean Squared Error (RMSE), Mean Absolute Percentage Error metrics are used to assess the accuracy of forecasting models
(MAPE), and Mean Absolute Scaled Error (MASE) are used to without being influenced by either the scale or magnitude of
measure the effectiveness of a forecasting model. The goal the data. These metrics provide a relative measure of accuracy
of improving effectiveness is to enhance the precision and that is not tied to the specific values of the target variable.
reliability of the model's predictions. This is particularly useful when comparing the performance of
• Robustness focuses on the model's ability to maintain its different models on datasets with different scales or units. The
performance across various conditions, including different scale-dependent and scale-independent accuracy metrics used
datasets, changing environments, and unexpected events. in this work are:
A robust forecasting model remains accurate and stable
even when facing variations, outliers, noise, or shifts in • Scale-dependent metrics: mean absolute error (MAE) and root
the underlying data distribution. A robust model does not mean squared error (RMSE);
break down or significantly degrade its performance when • Scale-independent metrics: mean absolute percentage error
subjected to different scenarios or when the underlying (MAPE) and mean absolute scaled error (MASE).
assumptions of the model are challenged. For example, a
forecasting model that performs well on historical data from These metrics provide distinct insights into forecasting
the training set, might not be robust if it is applied to data performance. Choosing the right metric depends on the nature
from the testing set with varying characteristics. A robust of the data, the goals of the forecast, and the tolerance for
model can handle such differences and provide reasonably different types of errors:

IFTA.ORG PAGE 19
 IFTA JOURNAL 2025 EDITION

The random walk mean absolute error is equal to:

• Mean absolute error (MAE) is a scale-dependent measure of
forecasting accuracy based on arithmetic mean of absolute
value of residuals or forecasting errors:

and the seasonal random walk mean absolute error is equal

to:

where the forecasting errors or residuals are expressed as:

where m is the corresponding forecasting seasonal lag.2

• Root mean squared error (RMSE) is a scale-dependent measure The MASE is a scale-independent accuracy metric comparing
of forecasting accuracy based on square root of arithmetic the forecast errors of a given model to the errors of a simple
mean of squared residuals or forecasting errors: baseline forecast or benchmark, typically a naive forecast
like the random walk. It is also valuable when dealing with
intermittent demand patterns or when the data contains
outliers. A value of 1 indicates that the model's performance is
equivalent to the benchmark, while values below 1 indicate that
the model is performing better than the benchmark, and values
above 1 indicate a worse performance. A limitation to consider
when interpreting MASE is that it does not indicate whether a
• Mean absolute percentage error (MAPE) is a scale-independent model is actually accurate or not, only how it performs relative
measure of forecasting accuracy based on arithmetic to a specific baseline. In summary, MASE is a useful metric for
mean of absolute value of residuals or forecasting errors as assessing a forecasting model's performance by comparing it to
percentage of actual data: a simple baseline forecast. It provides a standardized accuracy
measure that is not influenced by the scale of the data, making it
valuable for cross-model and cross-dataset comparisons.

4.3 Model Selection

Model selection consists of assessing the trade-off between
the goodness of fit and the complexity of a forecasting model
• Mean absolute scaled error (MASE) is a scale-independent within a training dataset. This evaluation is done by using
measure of forecasting accuracy based on arithmetic mean information loss criteria that penalize models with a larger
of absolute forecasting errors or residuals divided by training number of parameters. The primary goal is to avoid overfitting,
range random walk or seasonal random walk mean absolute a situation where a model captures excessive noise in the
error [5]: training data rather than the underlying patterns. Goodness
of fit, in general, measures how well a statistical or predictive
model aligns with the observed data. In simpler terms, it
assesses how closely the model's predictions match the actual
observations. A strong goodness of fit suggests that the model
effectively captures the underlying patterns and relationships
in the data. In the context of forecasting models, this aspect is
crucial because it helps measure the accuracy of the model's
predictions. If a forecasting model exhibits a good fit, it means
that its predictions closely align with the historical data,
indicating its potential to offer reliable forecasts in the future.
Conversely, a poor fit suggests that the model is unable to
adequately capture the data's patterns and may not perform
well when making predictions. The information loss criteria
commonly used in the evaluation process are:

• Akaike Information Criterion (AIC) [8] is used to assess

PAGE 20 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

the goodness of fit and complexity trade-off in a model. It in situations with limited data. The formula for AICc is
quantifies how well the model fits the data while considering expressed by:
its complexity and penalizes models with a larger number
of parameters. Lower AIC values indicate a better trade-
off between goodness of fit and model complexity. The AIC
formula is given by:
where n is the number of observations and k is the number of
model parameters.

• Schwarz Bayesian Information Criterion (BIC) [10] is a method

used to assess the trade-off between the goodness of fit and
where sse is the sum of squared errors, k is the number of model complexity. Like other information loss criteria, it
model parameters, p is the autoregressive order, q is the moving penalizes models with an increasing number of parameters,
average order, P is the seasonal autoregressive order, Q is the but BIC tends to be stricter in this regard compared to AIC.
seasonal moving average order, c is a binary term equal to 1 BIC is expressed as:
if the corresponding model has a constant coefficient for the
trend, and 1 is for the error term. In the Akaike Information
Criterion (AIC), the primary assumption regarding the
probability distribution of the forecast errors or residuals is
that they follow a normal distribution. Specifically, AIC assumes When utilizing BIC, there is a critical assumption made
that the residuals are normally distributed with a mean of zero concerning the distribution of forecasting errors or residuals.
and allows for varying levels of dispersion or heteroscedasticity. Specifically, BIC assumes that these residuals follow a normal
The assumption that the residuals are normally distributed distribution with zero mean and constant variance. In simpler
allows AIC to accurately evaluate the model's goodness of terms, it assumes that the errors are normally distributed and
fit and balance it against the complexity of the model (as homoscedastic, meaning they have a consistent variance across
represented by the number of parameters k). It is important to all observations. This assumption is essential because it forms
note that if this assumption is significantly violated, meaning the basis for how BIC penalizes models. BIC penalizes models
that the residuals do not follow a normal distribution, AIC that incorporate an excessive number of parameters relative
might not provide accurate model selection results. In such to the amount of available data. Models with more parameters
cases, alternative model selection criteria designed to handle are more flexible and can potentially fit the data more closely,
non-normal and heteroscedastic forecast errors may be more but they also come with a higher risk of overfitting. By
appropriate. imposing penalties on models with an excess of parameters,
BIC aims to model complexity and goodness of fit. However,
• Corrected Akaike Information Criterion (AICc) [9] is used to if the assumption of normality and constant variance for the
balance the goodness of fit and model complexity when residuals is significantly violated, BIC may not offer accurate
working with limited data samples. It is a modification of the model selection outcomes. In such instances, it may be more
Akaike Information Criterion (AIC) tailored for cases where appropriate to consider alternative model selection criteria
data is scarce, and overfitting is a concern. A key feature of designed to handle non-normal and heteroscedastic forecast
the AICc is its consideration of sample size relative to the errors.
number of model parameters. AIC may tend to favor more The Akaike Information Criterion (AIC) and the Bayesian
complex models with more parameters when the sample size Information Criterion (BIC) are the model selection criteria used
is small. This preference for complexity can be problematic, in this work, but they differ in several key aspects. Hence, it is
potentially leading to overfitting, where a model fits the beneficial to compare the two criteria:
noise in the data rather than the underlying patterns. The
AICc addresses this bias by introducing a penalty term based • Philosophy: AIC is based on information theory and attempts
on the sample size. In essence, it penalizes models with a to balance model fit and model complexity. It favors models
larger number of parameters more heavily when dealing that fit the data well but penalizes models with a large
with smaller samples. This correction helps establish a number of parameters. BIC is derived from Bayesian statistics
better equilibrium between model complexity and goodness and is more conservative. It not only penalizes models
of fit. In summary, the AICc is particularly useful when for complexity but also favors simpler models with fewer
comparing models with varying degrees of complexity parameters.
and working with relatively small samples. It offers a more • Penalty for model complexity: AIC applies a less conservative
appropriate assessment of the trade-off between goodness penalty for model complexity compared to BIC. It allows for
of fit and complexity, thus preventing overly complex models more flexibility in terms of model complexity, which can be an
from being favored due to smaller sample sizes. The AICc advantage when dealing with complex data or when a more
penalty term is proportional to the square of the number flexible model is needed. In contrast, BIC imposes a more
of parameters, serving as a safeguard against overfitting substantial penalty for model complexity. This results in a

IFTA.ORG PAGE 21
 IFTA JOURNAL 2025 EDITION

preference for simpler models, making it a more conservative One-step forecasting is referred to as “without re-estimation”
choice when selecting models. because it involves making forecasts for just the next single
• Sample size: AIC tends to perform better when the sample size time step without re-calculating or re-estimating the model’s
is relatively large compared to the number of parameters, parameters as new observations become available. In other
as it is less likely to favor overly complex models. BIC is words, the process of forecasting is repeated step by step using
more suitable when dealing with smaller sample sizes, as the same model’s parameters for each individual time step.
its stronger penalty for complexity helps protect against In contrast, in multistep forecasting we make predictions
overfitting. for multiple future time steps (over the entire testing range).
• Normality assumption: AIC assumes that the forecast Both one-step and multistep forecasting approaches have
errors (or residuals) follow a normal distribution but their own advantages and limitations. Multistep forecasting
allows for heteroscedasticity (varying variance). BIC also is simpler and computationally less intensive, but it might not
assumes normality for the residuals but with the additional capture evolving patterns in the data; hence, it can provide less
assumption of constant variance (homoscedasticity) across accurate forecasts over longer horizons. Moreover, multistep
all observations. forecasting consists of out-of-sample forecasting, while one-
• Model selection criteria: AIC often leads to the selection of step forecasting without re-estimation consists of in-sample
more complex models when there is a trade-off between forecasting but using the new testing range data.
model fit and complexity. BIC tends to favor simpler models,
even if they have slightly lower goodness of fit, due to its more 5.2 Arithmetic Mean Method
stringent penalty for complexity. The Arithmetic Mean method is one of the simplest and most
straightforward techniques used for forecasting time series
In summary, AIC and BIC differ in their approach to model data. It computes the average or mean of historical observations
selection, with AIC being more flexible and favoring model fit, (in the training set) and uses this value as the forecast for future
while BIC is more conservative and prefers simpler models. time periods (in the testing set). The formula for the Arithmetic
The choice between them depends on the specific goals of Mean method is quite simple, and the step-by-step process to
the analysis, the sample size, and the nature of the data. AIC apply for this model is:
assumes normality in the distribution of errors but allows
for heteroscedasticity, meaning the variance of the residuals 1. Collect the time series data representing past observations
can vary across observations. This flexibility in allowing for (in our case the historical data in the training set);
varying levels of dispersion in the residuals makes the AIC 2. Calculate the mean: Add up all the historical values and divide
a more robust choice when dealing with situations where the sum by the number of data points of the previous period
the assumption of constant variance is not satisfied, and the historical data (training range);
residuals exhibit different levels of spread or dispersion. When
choosing a forecasting model using criteria such as AIC or BIC,
it is a common practice to compare the AIC or BIC values of
various models. The model that exhibits the lowest AIC or BIC is
generally regarded as the most appropriate choice.
3. Forecast future values: Use the calculated mean as the
5. Simple Forecasting Methods forecast for all future time periods .
This Section analyzes straightforward forecasting
techniques, including the arithmetic mean, random walk, The arithmetic mean method is useful when the data is
seasonal random walk, and random walk with drift, providing relatively stable over time and does not exhibit significant
a foundational understanding of their principles, applications, trends, seasonality, or other complex patterns. It provides a
and limitations. Simple methods serve as benchmarks for baseline or naive forecast that can serve as a starting point
more sophisticated approaches and provide a starting point for more sophisticated forecasting techniques. However, it
for assessing the consistency of predictions and establishing is essential to be cautious while using the mean method for
a basis for comparison. The four simple forecasting methods forecasting, especially if the data contains outliers or exhibits
explored in this Section lay the groundwork for understanding significant variations. In such cases, the mean method may not
the essentials of forecasting. capture the underlying patterns accurately and may lead to
inaccurate predictions. Overall, the mean method is a simple
5.1 Multistep Forecasting vs One-step Forecasting and fast approach for forecasting when dealing with relatively
without Re-estimation stable and straightforward time series data. The model is
Multistep forecasting consists of forecasts done at the trained on the training range and then exercised for forecasting
beginning of the testing range for the full testing range without in the testing range, using multistep forecasting with a
using testing range data. In contrast, one-step forecasting multistep period h corresponding to the length of the testing
without re-estimation consists of forecasts done using the range and one-step forecasting without re-estimation
model’s parameters estimated in the training range and the In this case the two methods yield the same results, as
testing range data without re-estimating the parameters shown in Figure 6, where the and plots overlap.
obtained in the training range.

PAGE 22 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 6. Arithmetic mean in the testing range

5.4 Seasonal Random Walk Method
The Random Walk method can be improved by introducing
more sophisticated forecasting techniques that take into
account trends, seasonality, and other factors affecting the
time series data. The Seasonal Random Walk is an extension of
the basic Random Walk method that incorporates seasonality
into the forecast. It is used when the time series exhibits clear
seasonal patterns or cyclic behavior. In the Seasonal Random
Walk method, the forecast for the next time period is calculated
as the value from the corresponding seasonal period in the
previous season. The formula for the Seasonal Random Walk
forecast is:

where on a daily time frame, m = 21 days in a business calendar

represents a monthly seasonality. Monthly seasonality refers
5.3 Random Walk Method to the regular and predictable patterns or fluctuations that
The Random Walk method is a simple and basic time series occur monthly. In many real-world scenarios, certain events,
forecasting technique that assumes that future values will be behaviors, or factors tend to follow a monthly pattern, leading
equal to the last observed value. In other words, this approach is to recurring patterns in the data over each month. For example,
based on the idea that future variations are unpredictable, that in retail, there might be a higher demand for certain products
the series will follow a “random walk” [11] and will continue its around the end of the month, when people receive their
recent direction or behavior, making the last known value the paychecks. Like the basic Random Walk method, the Seasonal
best estimate for upcoming periods. The formula for the random Random Walk is relatively simple and can be useful for short-
walk forecast is given by: term forecasting when the data shows stable seasonal patterns.
However, it may not be appropriate for long-term forecasting
or when the seasonality in the data is subject to significant
changes or irregular patterns. The forecasts of the seasonal
where is the forecast for the next period, and is the random walk in the testing range are depicted in Figure 8, while
last observed value. The Random Walk method is often used Figure 9 shows a zoom-in to better assess the results obtained
for short-term forecasting when there is little information by the multistep forecasting and the one-step forecasting
about underlying trends or patterns in the data. It is especially without re-estimation methods.
applicable when the data is relatively stable, and there are
no significant trends or seasonality. However, the Random Figure 8. Seasonal random walk in the testing range
Walk method is not suitable for long-term forecasting or when
the data has clear patterns or seasonality. The Random Walk
method forecasting results are shown in Figure 7. The multistep
forecasting for the Random Walk considers the last value
of the training range and repeats it for the full testing range.

Figure 7. Random walk in the testing range

IFTA.ORG PAGE 23
 IFTA JOURNAL 2025 EDITION

Figure 9. Seasonal random walk in the testing range: Zoom-in Figure 10. Random walk with drift in the testing range

Figure 11. Random walk with drift in the testing range:

5.5 Random Walk with Drift Method Zoom-in
The Random Walk with Drift method consists of forecasts
equal to the previous period data plus the arithmetic mean of
previous periods historical data differences. The Random Walk
with Drift is an extension of the basic Random Walk forecasting
method, which includes an additional constant term or drift
component to account for a trend in the data. In the Random
Walk with Drift method, the forecast for the next period is
calculated as the last observed value plus the drift term. The
formula for the Random Walk with Drift method is given by:

In one-step forecasting without re-estimation, we are

where is the drift term representing the trend in the data, not re-estimating the corresponding drift and we include a
is the forecast for the next period and is the last constant drift in the model. The challenge with multistep
observed value. The drift term indicates the average change forecasting in Random Walk with Drift is that the uncertainty
or slope in the time series data over time and is computed in and randomness accumulate as we project further into the
the training range. If the drift term is positive, it suggests a future. This means that the confidence in our predictions tends
gradual upward trend in the data. Conversely, if the drift term is to decrease moving farther away from the present time. In
negative, it indicates a downward trend. If the drift term is zero, practice, multistep forecasting in the context of Random Walk
the Random Walk with Drift reduces to the basic Random Walk. with Drift can be useful for certain types of time series data,
The Random Walk with Drift is useful when the time series data where the underlying trend is expected to continue over a
exhibits a clear trend, and it provides a simple way to account certain period.
for this trend in forecasting. However, it is important to note
that Random Walk with Drift may not capture more complex 5.6 Forecasting Accuracy
trend patterns or sudden changes in the data. The results of the The forecasting accuracy metrics described in Section 4 have
multistep and one-step forecasting without re-estimation are been used to evaluate the performance of the simple methods
reported in Figure 10, while Figure 11 shows the corresponding presented in this Section. In particular, we considered the
zoom-in. metrics RSME (scale-dependent) and MAPE (scale-independent).
The numerical results are reported in Table 1.

PAGE 24 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Table 1. Simple methods forecasting accuracy results

Regarding multistep forecasting, Random Walk with Drift obtained the lowest RMSE and MAPE values, therefore the highest forecasting
accuracy. For one-step forecasting without re-estimation both for RSME and MAPE the Random Walk with Drift achieved the lowest values,
thus confirming the highest forecasting accuracy.

[7] Zakamulin, V. “The Real-Life Performance of Market Timing with

Footnotes Moving Average and Time- Series Momentum Rules,” Journal of
1
Certified SIAT Technical Analyst – SIAT (Società’ Italiana Analisi Asset Management, 15(4), pp. 261-278, 2014.
Tecnica) [8] Akaike, H. "A New Look at the Statistical Model Identification," IEEE
2
In this work, we use daily data on a business calendar, and we Transactions and Automatic Control, 19(6), pp. 716-723, 1974.
approximate the monthly seasonality to 21 days, therefore m = 21. [9] Sugiura, N. "Further Analysis of the Data by Akaike's Information
Criterion and the Finite Corrections," Communications in Statistics,
References 7(1), pp. 13-26, 1978.
[1] Makridakis, S. G., Wheelwright, S. C., and Hyndman, R. J. [10] Schwarz, G. E. "Estimating the Dimension of a Model," Annals of
“Forecasting: Methods and Applications.” 3rd Edition, John Wiley Statistics, 6(2), pp. 461-464, 1978.
and Sons, 1998. [11] Malkiel, B. G. “A Random Walk Down Wall Street.” W. W. Norton &
[2] Hyndman, R. J., Koehler, A. B., Ord, J. K., and Snyder, R. D. Company, Inc., NY, U.S., 1996.
“Forecasting with Exponential Smoothing: The State Space
Approach.” Springer Series in Statistics, 2008.
[3] Box, G. E. P., Jenkins, G. M., Reinsel, G. C., and Ljung, G. M. “Time
Series Analysis: Forecasting and Control.” 5th Edition, Wiley Series
in Probability and Statistics, 2016.
[4] https://finance.yahoo.com/
[5] Hyndman, R. J. and Koehler A. B. “Another Look at Measures of
Forecast Accuracy,” International Journal of Forecasting, 22(4), pp.
679-688, 2006
[6] Inoue, A. and Rossi, B. “Out-of-Sample Forecast Tests Robust to the
Choice of Window Size,” Journal of Business & Economic Statistics,
30(3), pp. 432-453, 2012.

IFTA.ORG PAGE 25
 IFTA JOURNAL 2025 EDITION

Dr. Patrick Winter

Trend-Adaptation of Moving Averages (TAMA) Berlin, Germany
research@patrick-winter.de

Figure 1: Typical bias of MAs during trends.

Abstract
Largely unnoticed so far, Moving Averages (MAs) exhibit a
trend bias: They underestimate (overestimate) the true value of
the underlying asset during upward (downward) trends. This
can not only result in misleading interpretations and forecasts,
it also propagates to downstream indicators and ultimately the
whole trading system. In this paper, we propose a robust method
– TAMA – to correct the bias, i.e., to make any MA trend-
adaptive. We exemplify this for common MAs and evaluate the
effects qualitatively and quantitatively. Our findings indicate
that trend adaptation greatly improves their performance
in typical use cases. Additionally, we outline some new chart
analysis tools that could be derived from TAMA: a Harmonic MA,
a forecasting procedure, and a bands indicator. We will investigate where this apparent contradiction
comes from in a moment, but let us first stay with the figure
1. Introduction and ask another question: Since it can clearly be seen that –
Moving Averages (MAs) are among the most fundamental and even how far – the MA is from the price, can it then not
tools of technical analysis. Their primary job is to filter out simply be shifted mathematically just as it can be mentally,
quasi-random noise from a price development – smoothing it formally correcting its distortion? This is in fact possible,
thereby –, which is supposed to allow a better insight into the but unfortunately only if the price development is viewed
current “true value” of the underlying asset. Analysts then can retrospectively (as we do it here), i.e., if one already knows when
interpret this value directly, forecast the continuation based on trends have occurred. The usual intention to use an MA is to
it, or use it as input for downstream indicators (e.g., the MACD calculate it during the ongoing price development (“online”), of
or Bollinger Bands) and trading systems. course, and at those points in time, the data necessary for such
For all these applications, it is obviously of crucial importance a correction are obviously not yet available. This may also be a
that the MA’s value actually represents the asset’s true value main reason why MAs, despite their great importance, still have
without distortion, i.e., unbiased. If it does not, interpretations not been trend-adapted.
and forecasts could be misleading, and, even worse, However, this does not mean that the approach has to
downstream indicators and trading systems could produce be abandoned altogether. Although the trend is not fully
wrong signals without anyone noticing, as the MA input is foreseeable at the time of MA calculation, methods exist
subsumed by them. to estimate it, even robustly despite the then very short
Fortunately, most commonly used MAs are unbiased. This trend window, and the information obtained thereby can be
is at least what many traders believe, and for good reasons integrated into the MA. In the following, we will develop a rather
indeed: When we look at the formula of the Simple MA (SMA), simple method to achieve this (Section 2) and show then, in
for instance, it is identical to the formula of the arithmetic Section 3, how it can be used to make any MA trend-adaptive1,
mean, which is theoretically known to be not only unbiased but i.e., to remove its bias during trends to a large extent. Our
even the best unbiased linear estimator under some conditions particular method also allows for some unique extensions,
(BLUE; e.g., Blom 1976). which are presented in Section 4. In Section 5, we evaluate by
In the practice of technical analysis, however, cases occur data whether and, if so, by how much MAs can be improved
daily like the one exemplarily shown in Figure 1 (for the this way regarding their typical use cases. Section 6 finally
EUR/USD exchange rate in mid-2022): The employed MAs – concludes this work with some outlooks for future research.
regardless of their concrete specifics – seem to systematically
underestimate price movements during upward trends and 2. The TAMA Method
systematically overestimate them during downward trends (red
circles) – and thus not to be unbiased at all! 2.1 A Closer Look at the Problem
Let us now return to the aforementioned seeming
contradiction: Why does a MA like the SMA, which has a formula
that is theoretically unbiased, still exhibit a bias during trends?

PAGE 26 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

The reason for this lies in the fact that the prices pt are not 2.2 Idea and Approach
realizations of the same – especially identically distributed But how can one know which paradigm is the correct one in
– random variable, as it is assumed in the calculation of the each case? Absolute certainty only comes in hindsight; while
arithmetic mean (i.e., they do not form a sample). Rather, each the price is evolving, estimations are necessary. Fortunately,
price pt should be interpreted as the sole realization of its own the situation without a trend can easily be modelled as a special
random variable Pt, and it is by no means guaranteed that this case of the situation with a trend, and then it can be evaluated
variable must have the same distribution – or even the same based on the past price how likely this special case is.
location (or expectation) – as the previous one (Pt-1). On the The most intuitive approach that comes to mind may be to
contrary: It is precisely the definition of a trend that this is not create a simple linear regression model of the price development
the case. with the time index as the only explanatory variable and to train
Figure 2 illustrates this problem with an idealized price it then with the data from the current MA-window (so for an
development smoothed by a 5-period SMA, examined at each n-period MA at time t, using the prices pt-n+1 to pt). The estimated
point in time t at the end of a day, without knowledge of the value for pt (even though this price is already known!) that
future. First consider t = 5. The price has stayed at a value of 52 results from the regression line obtained this way could then
for five consecutive days here; thus, SMA5 (5)=1/5 • 5 • 52=52. This represent the value of the trend-adaptive MA at t.
would also be the best possible forecast for t = 6; and indeed, In fact, this approach corresponds mathematically to the
that day ends again with a value of 52. Therefore, SMA5 (6)=52, trend-adaptive version of the SMA (the TASMA), as we will see
and the previous forecast would now also be the best for t = 7. later.2 However, it has some significant weaknesses (also when
But things turn out differently: The price now rises to 54. This compared to other TAMA variants); in particular, it is not robust,
information is one that holds significance for analysts. As a e.g., regarding the choice of n. Figure 3 illustrates this problem
counter example, consider t = 10: The price here has increased and its solution based on the price development from Figure 2.
equidistantly for five consecutive days. Under the paradigm
that all price values between t = 6 and t = 10 were drawn from
the same distribution, the best estimate of their expected Figure 3: Various regression lines with trend on the price
value – and thus the best smoothing for t = 10 as well as the best development from Figure 2.
forecast for t = 11 – would indeed be that of the SMA, namely
SMA5 (10)=1/5 • (52+54+56+58+60)=56. Under the paradigm that
a trend exists, i.e., E(Pt )=E(Pt-1 )+m for a constant m (here m = 2),
the corresponding values would, in contrast, lie exactly on the
trendline, at 60 for t = 10 and at 62 for t = 11. The deviation of the
SMA from this is therefore irrelevant – or rather misleading
– information for analysts under the latter paradigm, since it
merely results from the mathematical neglect of the trend.

Figure 2: Idealized price development with an SMA(5).

Let us first consider only the red line. This is the regression
line that would result from the process described above for n
= 5. Its calculation has included the entire downtrend present
at the current time t = 13, i.e., the points t = 13, t = 12, and t = 11,
but also the points t = 10 and t = 9, which – though only clearly
recognizable in this illustrative example – do not belong to it.
If a different period length had been chosen instead, another
line would have resulted, e.g. the green one for n = 3, which hits
the trend much better. So the choice is not irrelevant, and the
trend will not always have started just by chance at the edge of
the period window. Introducing a parameter i≤n that lets the
user specify a different start time would also not help much, as

IFTA.ORG PAGE 27
 IFTA JOURNAL 2025 EDITION

this parameter is not known, and certainly not the same for all 3. From MA to TAMA
windows.
The idea for "robustifying" the regression is therefore to 3.1 General Procedure
perform it not only for one point in time of the window, but Now that we have formalized our idea, we could call the
for all points in time t - i + 1 within it (cf. lines in Figure 3), i.e., TAGMA a new indicator, choose any weights wi and any residual
multiple times. Then, a trend that starts only after the left edge r(t) for it, and include it in our trading system to replace the
of the window, as it is the case in the illustration, is still found. classical MAs.
Additionally, while points that do not belong to it still have an However, our actual goal was different: Instead of replacing
influence on some estimates, only the "more leftward" ones the classical MAs by something new, we just wanted to
are affected by this anymore (in the example for i [4;5]). The trend-correct them. This means that what remains is to find
(overall) estimate for pt is then obtained as an average of the a relationship between any MA – call it “Generic MA” (GMA)
individual estimates i (cf. coloured dots). – and its trend-adaptive (TA) version, the TAGMA (hence the
name).
2.3 Formalization To do so, we start with the formula that (almost)4 every
We will now formalize this idea, proceeding “backwards” concrete MA is a special case of:
from the last step to the first. This last step is the calculation of
an average of the price estimates at time t:
(4)
(1)

with weights ui (which can, of course, also be 0) and a price-

To take into account that there are multiple ways to calculate constant residual s(t), both factors to characterize the concrete
“an average”, we have introduced here weights wi and a MA. This almost looks like the definition of the TAGMA in (1)
residual constant r(t) that may depend on t but not on the price already, but there is an important difference: The GMA is based
estimates. This flexibility will play a crucial role later, and it on different actual prices pt-i+1, while the TAGMA is based on
will also become clear then why we have denoted the result as different estimates i of the same price pt. 5
〖TAGMAn (t). To make both approaches comparable, an essential insight
The i-th price estimate should be the result of its own linear is required: If the TAGMA is the TA-version of the GMA, it must
regression, thus lie on the corresponding regression line. match the latter for all arbitrary price values and points in time
Denoting as its intercept and as its slope, then i can be when trend adaptation is “turned off”. We will later introduce
formalized as that capability as an extension to the TAMA method in more
detail; for now, it should suffice to say that TA enters (2) through
(2) the dynamic slope coefficient and thus can be deactivated by
setting = 0 instead of its value from (3b). Using (2) and (3a)
with this, we can demand the following equality independent of
since what is sought is the i-th point on the line in each case. To specific price values:
illustrate, it is helpful to look at Figure 3 again, for example for
i=3 (green line): This is estimated by the points p11, p12, and p13,
and if we number these from left to right, the estimation time (5)
t = 13 corresponds exactly to the i-th (and last) point.
(2) refers to simple linear regressions, and for such, general One can show (see Appendix) that this is satisfied if and only
formulas for the straightforward calculation of and are if the concretization parameters of the TAGMA are chosen as
known.3 Adapted to our case, the following relationships are follows:
derived (see Appendix):

(3a) (6a)

(3b)
(6b)

where denotes the arithmetic mean of the prices from pt-i+1 Thus, a fairly simple relationship has been found that allows
to pt. the calculation of the weights for the corresponding TAMA with
deactivated trend adaptation from the knowledge of the MA’s
formula. With the "reactivation" of TA by re-setting to its
value from (3b), its construction is complete.

PAGE 28 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

3.2 TA-versions of Common MAs However, if one decomposes the calculation formula as shown
We can now apply our newly found relationship to any MA, in (9), the issue becomes easily resolvable: All other prices for
and we will do so in the following for the three arguably most i ≥ n, including their coefficients, do not enter the weights wi but
common MAs. Our approaches will be illustrative for other MAs are instead represented by the residual term r(t); this term must
as well. be constant with respect to the prices in the window, but not
necessarily with regard to the other ones. Thus, the weights wi
Simple Moving Average (SMA) can be determined as usual from (6a):
The SMA is characterized by and s(t) = 0. Thus, we have
from (6a): (10)

(7)
3.3 The “Simplest” TAMA: The TAHMA
One might have expected that the simplest concretization of
the TAGMA in (1), which can be constructed by choosing equal
This means that all regressions except the “largest one” (the weights of wi =1 for all i and a residual of r(t)=0, making the
one taking into account all n points up to the beginning of the calculated average the arithmetic mean, were the TA-version
period window) are excluded from the calculation! Thus, the of the SMA, since the latter also uses equal weights and no
resulting TASMA is based only on this single regression, making residual. However, by (7) we have found that this is not the case.
it generally fragile and sensitive to the choice of n. This issue has Instead, it can be shown (cf. derivation of (6a) and (6b)) that the
been discussed in more detail in Section 2.1. As a consequence MA that corresponds to the “simplest” TAMA – the TAHMA –
of it, the TASMA is usually more of theoretical interest and not looks like follows:
recommended to be used in practice.
(11)
Weighted Moving Average (WMA)
The Weighted MA (WMA) is based on the idea of assigning the
weight to the price pt-i+1, which is i - 1 days away from
the current price p_t, so that older influences linearly decrease where
in importance compared to newer ones. Here, N = n + ( n - 1 ) + ...
+1= is a normalization factor. According to (6a), the weights denotes the i-th so-called harmonic number, that is, the
of the TAWMA are determined as follows: i-th partial sum of the harmonic series (H0 = 0 is defined for
notational reasons). We therefore call this MA the Harmonic MA
(HMA).7
(8) This relationship may appear surprising at first glance, but
upon closer examination, it turns out to be very natural: The
TAMA is by construction an average of regression estimates,
Similar to the SMA, the WMA represents an interesting those are also averages themselves, and the basis of the
special case in a sense, as the otherwise necessary distinction regressions are exactly i = 1, 2, …, n points. Now a consecutive
in (6a) is not needed due to the arithmetic progression of its average of averages corresponds precisely to the essence of the
weights. harmonic series.
This inherent connection to the method, its simplicity, and the
Exponential Moving Average (EMA) weaknesses of the TASMA make the TAHMA an especially good
The Exponential MA (EMA) extends the idea of the WMA; it candidate for practical use. In fact, we will see later that it also
also weights prices less that are further back in time, but by an performs well; however, the difference between various TAMAs
exponential progression: is generally not very pronounced.

4. Extensions
(9) Now that we have found how the common TAMAs look like, we
could proceed directly to their practical application. However,
the particular method we have derived for trend adaptation
allows for some special extensions that will proof useful there,
with a smoothing factor 0 < < 1, which is typically but not so that we will introduce them beforehand.
necessarily chosen as Consequently, the weights
become progressively smaller, but they never reach the value of
0 and thus never disappear in theory.6 At first glance, this might
seem problematic, as only a weighted sum of precisely n prices
within the current period window has been allowed so far.

IFTA.ORG PAGE 29
 IFTA JOURNAL 2025 EDITION

4.1 Trend-Adaptation Switch TA either have its default value of 0, for which (13) and (2) coincide,
The first extension has already been mentioned and used or be 1, so it effectively acts as a “forecasting switch”.
to obtain (5): a "switch" that allows deactivating the trend
adaptation if desired. This enables a direct evaluation of its 4.3 Generic Aggregation Function AF
effect. In (1), the n estimates for the price pt at time t were
The switch is formally a Boolean parameter, denoted here as aggregated to an overall estimate through averaging. Though
TA, which is “true” (i.e., 1) by default and must be set to "false" this is the most intuitive method, it is not the only possible
(i.e., 0) to deactivate it. TA can be incorporated by modifying one. The right side of (1) can thus be replaced by a more generic
(3b) as follows: aggregation function AF:

(14)
(12)

By default, AF should still correspond to a (linear) weighted

The effect is immediately clear: Setting the slope of the line β average – in which case (14) is (1) –, but there are applications
to 0 means nothing other than to rule out a trend – precisely for which other aggregation functions are even more useful.
as MAs implicitly do. This becomes evident in (2) in conjunction A prominent example for such an application are band
with (3a); since the terms associated with = 0 vanish, it indicators. Again, TAMAs could simply be used as a replacement
simply holds that for their conventional counterparts here and would likely
perform better than these, since the generated bands are then
laid around a mean value that more closely reflects the truth,
and the interpretation does not need to depend anymore on
whether there is a trend or not.
4.2 Forecasting Period P The TAMA method also entails a specific band indicator,
Many forecasting models for the future price development however, when appropriate aggregation functions are used.
are based on MAs (see Nau 2014 for an introduction). Of course, For instance, a price development could be described and
TAMAs can also be used as inputs for such a model, and since analysed using the 20%-quantile band, the 80%-quantile band,
they take into account not just averages but also their changes and possibly the median (i.e., the 50%-quantile band) of the
during trends, it would be unsurprising to obtain better estimated values . This can even be used to build a trading
predictions thereby alone. However, due to the particular system out of TAMA. We will demonstrate this idea later on.
calculation of TAMAs – after all, they are based on regression
models – it is additionally possible to use them for forecasting 5. Practical Evaluation
independently. Having specified the TAMA method and its extensions
A further extension is introduced for this: In (2), the price at completely, it is time to return to Figure 1, which has
time t, pt, was estimated based on the corresponding point on demonstrated the trend-problem of conventional MAs.
the i-th regression line, which was found at index i. This was Continuing the same example, we will now examine
done solely for the purpose of smoothing (averaging), as pt is qualitatively whether the problem can be solved by TAMA and
already known at (or at the end of) t. The price at a later point how this affects the typical use cases of MAs mentioned in the
in time t + P, on the other hand, is still unknown by then; of introduction: smoothing, forecasting, and downstream usage.
course, however, nothing prevents us from looking for it on the We will also perform a small quantitative study to investigate
regression lines, too, just at index i + P instead of i. their performance in the latter two areas.
Formally, this requires introducing another parameter, here
designated as P (for “prognosis” or "period"), which defaults to 5.1 Smoothing
the value 0. (2) is then to be modified as follows: Figure 4 reiterates Figure 1, but this time with a TAHMA
added to it that has the same period duration (of 10) as the SMA
(13) we have looked at earlier, once with and once without trend
adaptation.

(13) means that the i-th estimate for pt is calculated

based on the data available at time t - P, and then extrapolated P
units into the future from there. Accordingly, a TAMA with this
extension provides values not only up to t but up to t + P, as data
are available exactly up to t = ( t + P ) - P.
P has been modelled here as any integer for maximum
flexibility.8 In practice, however, it is generally not advisable to
attempt extrapolating the price by more than one unit of time;
additionally, the estimation error naturally increases with the
distance from the data available at t. P will therefore usually

PAGE 30 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 4: Figure 1 extended by a TAHMA(10) with and Figure 5: Exchange rate forecast using a TAHMA(10) with
without trend adaptation (TA). P = 1, with deviations from the actual price.

Let us first focus on the version without trend adaptation How accurate is this forecast over the whole window (blue
(orange line), which is equivalent to a HMA(10) from (11). While it line)? Examining it on its own at first, it can visually be deemed
is at the same “level” as the SMA, it already provides a smoother appropriate: It is usually not too far from the actual value (which
picture of the exchange rate: Not only is it practically closer is not known until the following day, of course); on some days,
to it at every point in time, it is also significantly less delayed. it is nearly exact (e.g., 09/05), on others (especially those with
This results from the HMA, like an EMA, putting more emphasis significant changes), the deviation is larger (e.g., 12/05). Over-
on more recent prices. However, it can clearly be seen that the and underestimations (red and green lines, resp.) generally
trend-problem (cf. red circles in Figure 1) is not solved yet. balance each other out, although the former naturally occur
That is only achieved when trend-adaptation is switched on, more frequently during downward trends, while the latter occur
resulting in the actual TAHMA(10) (green line): Both during the more frequently during upward trends due to the MA basis.9
upward and downward trends, it not only stays very close to the However, we are more interested here in the added value of
exchange rate, as it should, but also is no longer systematically trend adaptation. To evaluate this, let us first consider what it
below or above it — it is indeed unbiased even during trends. Of would mean to forecast a price using conventional MAs, i.e.,
course, this also means that with the same period, it smooths without accounting for trends: The best possible prediction
slightly less than conventional MAs; it is therefore a plausible would then be the trivial one, i.e., assuming the price at t + 1 to
recommendation to use somewhat longer period durations for have the value of the MA at t – in other words, simply shifting
TAMAs in general. the MA one time unit to the right! This is illustrated by the grey
We would like to demonstrate the superiority of TAMAs for line, for which trend adaptation was deactivated using the TA=0
smoothing also quantitatively. However, that were challenging switch: It corresponds exactly to the orange line from Figure 4,
given that entire studies can be dedicated to the question alone just shifted. The value added by the TAMA method can thus be
how “smoothness” should be captured mathematically and directly determined by comparing the blue to the grey line: The
weighed against the accuracy of price representation (e.g., former generally lies much closer to the actual value, especially
Raudys and Pabarskaite 2016). However, it obviously is always during trends, corresponding to a significantly better forecast.
preferable when the latter, given a certain level of the former, This is not all that surprising, as dedicated approaches often
does not depend on whether the price “by chance” is in a trend also rely on regressions.
or not, and since TAMAs do not exhibit this weakness anymore, To quantify the extent of this improvement in general, we
they should dominate MAs in any quantitative investigation on conducted a similar analysis as shown here for the 40 DAX
smoothing. members over a 1-year period (01/07/2021-30/06/2022) as
example. Forecasting accuracy was measured using the Root
5.2 Forecasting Mean Square Error (RMSE), which (in principle) looks at the
Figure 5 again shows a TAHMA(10) for the USD exchange rate average size of an estimation residual (cf. red and green lines
example (blue line, cf. green line in Figure 4), but this time with in Figure 5), independent of its direction. The predictions were
P = 1 instead of P = 0 – i.e., the value at t now only relies on based on an EMA with a period of 20 days vs. a TAEMA with the
data up to t-1. Consequently, it can now be plotted up to t + 1, same period. Table 1 shows the results (columns 2-4). These are
forecasting the exchange rate for the next day (circle). basically as expected: The estimation error was lower for each
DAX member when trend adaptation was activated, as this
makes additional information — specifically, information about
trends — usable. However, the magnitude of this effect may be
surprising: The RMSE for the TAEMA was on average lower by
nearly 1/3 (31.97%) than for the EMA!

IFTA.ORG PAGE 31
 IFTA JOURNAL 2025 EDITION

Despite these appealing characteristics, TAMAs likely of size 2 • 2 • around a 20-period SMA in the standard
should not be used as a sole forecasting model. While they can setting, where denotes the standard deviation of prices
accurately capture averages and their changes, they lack a in the current window (for details, see Bollinger 2001). Many
component that represents the variance around these values. investors believe that the price at the next time unit will then
This gap can be bridged, however; either through suitable lie within this channel with a probability of about 95%. This is,
combinations with "spread indicators" like the Stochastic of course, not true in general because none of the assumptions
Oscillator, or by forming and using bands, i.e., interval rather behind this rationale is actually given (incidentally, this is
than point estimators. The latter approach is a special case of related to the problem discussed in Section 2.1). Nevertheless,
what will be discussed next. there is a probability that the expected will happen; it just
does not necessarily amount to 95%. Its actual value can be
5.3 Downstream Usage well estimated through backtesting; in our study, it averages
Finally, we will evaluate whether trend adaptation improves around 82% (column 5 of Table 1). But this is only the case for the
not only MAs themselves but also their downstream usage, conventional SMA; if its trend-adaptive version, the TASMA, is
i.e., other indicators or whole trading systems that are based used instead, the probability rises to over 90% (column 6), and
on them. As a well-known example for such a downstream with a TAEMA, it would still be higher – all for the same channel
indicator, we consider Bollinger Bands; these can also be size. This again shows how trend adaptation can improve
regarded an example for the interval estimators mentioned existing indicators.
above. They are constructed by placing a symmetric channel

Table 1: Comparison of forecasts and bands with and without trend adaptation.

Asset (1.7.21- rmse_ta0 rmse_ta1 -% boll_ta0 boll_ta1 boll_sig mima mima_sig

30.6.22)
1/7/21-30/6/22
adidas 130.12 88.78 31.77% 81.36% 93.22% 2 76.17% 0.77
Airbus 58.39 44.30 24.14% 84.75% 87.71% 2 73.19% 0.92
Allianz 90.35 61.39 32.05% 81.78% 91.95% 2 75.74% 0.78
BASF 32.71 20.30 37.94% 81.36% 89.41% 2 77.45% 0.82
Bayer 27.75 16.34 41.11% 80.08% 91.95% 2 71.06% 0.72
Beiersdorf 34.29 26.03 24.10% 86.86% 86.44% 2 74.89% 0.93
BMW 45.34 31.16 31.27% 84.75% 91.95% 2 74.89% 0.83
Brenntag 30.42 22.14 27.23% 84.32% 89.41% 2 73.62% 0.90
Continental 65.30 42.88 34.33% 78.81% 89.83% 2 78.30% 0.79
Covestro 29.49 19.75 33.02% 78.81% 89.83% 2 75.32% 0.83
Daimler Truck 16.69 10.15 39.15% 80.00% 88.33% 2 79.83% 0.77
Deutsche Bank 9.66 6.12 36.64% 84.32% 87.71% 2 75.74% 0.80
Deutsche Börse 55.99 37.12 33.71% 80.08% 92.37% 2 75.32% 0.79
Deutsche Post 25.02 16.61 33.63% 80.93% 88.98% 2 74.89% 0.79
Deutsche 6.47 4.47 31.00% 82.63% 91.10% 2 75.32% 0.79
Telekom
E.ON 4.40 2.95 32.80% 80.08% 92.80% 2 74.89% 0.81
Fresenius 19.15 11.28 41.09% 79.24% 94.07% 2 75.74% 0.69
Fresenius 25.81 17.44 32.45% 82.20% 91.10% 2 74.89% 0.80
Medical Care
Hannover Rück 65.95 49.66 24.70% 80.51% 90.68% 2 72.34% 0.88
31.61 20.92 33.83% 83.90% 91.95% 2 79.15% 0.77
HelloFresh 64.32 41.13 36.05% 79.24% 90.68% 2 76.17% 0.77
Henkel vz. 31.30 21.77 30.43% 79.66% 87.71% 2 75.74% 0.80
Infineon 21.25 13.82 34.99% 80.93% 90.25% 2 75.74% 0.77
Linde 109.46 74.28 32.14% 81.36% 90.68% 2 72.34% 0.79
Mercedes-Benz 51.03 34.84 31.72% 87.29% 87.71% 2 77.87% 0.79
Merck 102.64 63.70 37.94% 77.97% 90.68% 2 74.04% 0.73
MTU Aero 104.40 76.85 26.38% 84.32% 88.98% 2 76.17% 0.83
Engines
Münchener 107.14 74.11 30.83% 80.93% 94.07% 2 73.19% 0.79
Rück

PAGE 32 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Asset (1.7.21- rmse_ta0 rmse_ta1 -% boll_ta0 boll_ta1 boll_sig mima mima_sig

30.6.22)
Porsche 55.81 42.22 24.36% 87.29% 88.98% 2 75.74% 0.85
PUMA 53.21 34.56 35.04% 80.51% 94.07% 2 77.02% 0.78
QIAGEN 18.29 13.62 25.56% 81.78% 89.41% 2 74.47% 0.88
RWE 15.68 12.48 20.42% 79.66% 91.10% 2 74.47% 0.90
SAP 43.96 29.07 33.87% 81.78% 92.37% 2 74.04% 0.76
Sartorius vz. 477.57 309.75 35.14% 82.63% 92.80% 2 76.17% 0.72
Siemens 72.59 51.51 29.04% 81.78% 87.71% 2 74.47% 0.85
Siemens 29.67 21.00 29.21% 82.20% 89.83% 2 76.60% 0.86
Healthineers
Symrise 50.83 33.44 34.20% 80.93% 90.68% 2 75.32% 0.77
Volkswagen 100.70 74.26 26.26% 85.17% 88.14% 2 74.89% 0.89
Vonovia 21.35 14.77 30.81% 86.02% 93.64% 2 77.87% 0.83
Zalando 52.13 32.14 38.35% 80.08% 92.37% 2 74.89% 0.75
Average 59.71 40.48 31.97% 81.96% 90.57% 2 75.40% 0.81

In addition, as already has been mentioned in Section 4.3, the such a crossing means that the price then falls below or exceeds
TAMA method also entails its own band indicator, and that will the smallest or largest possible extension of its current course;
be the object of our last analysis. For this purpose, let us once this should be a strong signal that something will change!
again look at the example of the EUR/USD exchange rate with a Figure 7 shows, to verify this, the content of Figure 6 but on
TAHMA(10), but this time with the aggregations functions a forecast basis, i.e., with P = 1. Indeed, there are several places
, respectively (Figure 6). Among where the envelope formed by the min-max-bands is crossed.
all quantile-based functions, these are of particular interest Most of these crossings are sequential in the sense that the
because they create the extreme bands, which envelop the price same band is pierced repeatedly one after another. This could
and do never intersect. This may not sound too exciting at first, be interpreted as a changing or (more likely) not yet stabilized
but it inspires two new evaluation possibilities: (estimated) slope. However, more significant is the view of the
respective first crossing points of a band after a breach point of
the other (shown in green and orange in the figure). If these were
Figure 6: min-max-Bands (AF) of TAHMA(10) around the interpreted as buy or sell signals in the example of the figure,
exchange rate.
one would have traded well, except for the very first case where
an apparent uptrend did not materialize: The large uptrend in
mid-May is recognized very promptly and exited exactly when a
plateau sets in; another buy signal then occurs only in mid-June
after the large downtrend is over.

Figure 7: Figure 6 on forecasting basis ( P = 1 ), with

respective first crossing points.

The first is the observation that the price is almost always

closer to one band than to the other, often – but not always (e.g.,
09/05, 30/05) – hitting the former. It can therefore be assumed
that its distance to the bands represents information that can
be used to determine the magnitude of the current trend.
The second idea may be even more interesting: If the
minimum and maximum of the estimates for the price
completely enclose it when pt is known, what is the meaning of
crossing this envelope when pt is instead forecasted? After all,

IFTA.ORG PAGE 33
 IFTA JOURNAL 2025 EDITION

indicator (e.g., Chande and Kroll 1994, p. 20ff.).

For a quantitative evaluation, the TAMA Bands were studied 3
It should be noted that we have no direct interest in these two
the same way as the Bollinger Bands. Their hit rate seems lower coefficients but only, indirectly, in the resulting estimated value
at first glance (column 8 of Table 1) – on average, 75% prices lie In contrast to the former, the latter is reliable even when the
within their channel –, but this channel also is significantly regression is based on very few data points; in the extreme case (i=1)
narrower: It requires only about 0.8 standard deviations even just one – undoubtedly, pt itself provides an adequate basis for
(column 9) instead of 2 (column 7). Therefore, the TAMA bands (since a straight line cannot be formed from a single point,
could be a good alternative for some applications. must be manually set to 0 in (3b) for this case).
4
The exception are “exotic” MAs that are non-linear in the prices,
6. Conclusion such as the Geometric Moving Average .
In this paper, we have developed a regression-based method However, many members of this group can be linearized.
to correct the bias MAs exhibit during trends. The method
5
There also are two notation-only differences: First, we do not factor
includes a “switch” extension by which trend adaptation can out 1/n as a normalization constant here because that would hardly
always simply be turned on or off, enabling traders to evaluate make sense in general. Second, MAs are calculated from right to
left while the regression lines in (1) are formed from left to right.
its effects easily.
However, this all is mathematically equivalent.
In our example study of 40 DAX members, we have found 6
In practice, there are two different approaches that trading programs
that turning it on greatly improves an MA’s performance in
can use to handle this. The approach described in the following
smoothing, forecasting, and downstream usage: Forecasting involves calculating the EMA up to the very first available data
errors were reduced by roughly 1/3 and Bollinger Bands became point. Alternatively, the calculation can be stopped after exactly
more consistent by roughly 10%. Future research can continue n steps. In that case, the coefficient of the n-th price p(t-n+1) must
to evaluate such effects, e.g. for other MA-based indicators such first be corrected by the factor . The procedure can then continue
as the MACD. analogously to the WMA.
The particular variants of TAMAs generally do not differ 7
One can sometimes find MAs under this designation that are harmonic
much regarding performance from our experience, at least no in the prices rather than in the coefficients (and thus “exotic”) (e.g.,
more than the underlying MAs. An exception to this, however, gorx1 2020); the HMA presented here should not be confused with
is the TASMA, since its calculation effectively uses only a single those.
regression, which is against the basic idea of TAMA of averaging 8
For some applications, even a negative P (i.e., a "retrospective") might
multiple estimations for robustness. In contrast, the TAHMA we make sense.
have introduced aligns most naturally with this idea, and thus it 9
This observation prompts the question of whether one could also
is the variant we recommend in the trend adaptation context. apply trend adaptation to the forecast value (instead of the actual
Finally, we have outlined some ideas on what else could be value), effectively meaning a "second-order" kind of TAMA, similar
derived from the TAMA method, namely a forecasting procedure to applying an MA to itself. This is presumably not feasible or
and a bands indicator. Both tools also showed promising results sensible since the crucial value pt is missing or already estimated
in the forecasting scenario, but this can be a subject for further
in our practical evaluation and are therefore worthy of further
research.
exploration in future research.

Implementation for MetaTrader 5 References

The TAMA method has been implemented for the popular Blom, G. (1976): When is the Arithmetic Mean Blue? The American
trading program MetaTrader 5 (for the MAs SMA, WMA, EMA, Statistician, Vol. 30, No. 1, pp. 40–42.
and HMA and the aggregation functions arithmetic mean, Bollinger, J. (2001): Bollinger on Bollinger Bands. McGraw-Hill
min, max, and quantile). It can be downloaded from the MQL5 Education.
marketplace (https://www.mql5.com/en/market). Chande, T.S. and Kroll, S. (1994): The New Technical Trader. John Wiley
& Sons.
Appendix gorx1 (2020): Harmonic Moving Average, https://www.tradingview.
The appendix contains the mathematical derivations of (3a) com/script/mpWRYb8F-Harmonic-Moving-Average.
and (3b) for one thing and of (6a) and (6b) for another thing. Kaufman, P. (1995): Smarter Trading – Improving Performance in
It has been omitted here for brevity but is available from the Changing Markets. McGraw-Hill.
author upon request. Nau, R. (2014): Forecasting with Moving Averages. Fuqua School of
Business, Duke University in Durham, NC, USA. https://people.
Footnotes duke.edu/~rnau/Notes_on_forecasting_with_moving_averages--
1
“Adaptive” refers to the MAs, as will become clear later, dynamically Robert_Nau.pdf.
adapting to trends. This has nothing to do with other “Adaptive MA” Raudys, A. and Pabarskaite, Z. (2016): Optimising The Smoothness And
indicators, such as the one by Kaufman (1995, pp. 129-153). Accuracy Of Moving Average For Stock Price Data. Technological
2
The TASMA described here (without the forthcoming extensions) is and Economic Development of Economy, Vol. 24, No. 3, pp.
essentially identical to the widely discussed Linear Regression Slope 984–1003.

PAGE 34 IFTA.ORG
Certified Financial Technician (CFTe) Program
IFTA Certified Financial Technician (CFTe) consists of the The CFTe II is also offered in English, French, German,
CFTe I and CFTe II examinations. Successful completion Italian, Spanish, Arabic, and Chinese, via Zoom, typically
of both examinations culminates in the award of in April and October of each year.
the CFTe, an internationally recognised professional
qualification in technical analysis. Curriculum
The CFTe II program is designed for self-study, however,
Examinations IFTA will also be happy to assist in finding qualified
The CFTe I exam is multiple-choice, covering a wide trainers. Local societies may offer preparatory courses
range of technical knowledge and understanding of the to assist potential candidates. Syllabuses, Study Guides
principals of technical analysis; it is offered in English, and registration are all available on the IFTA website at
French, German, Italian, Spanish, Arabic, and Chinese; http://www.ifta.org/certifications/registration/.
it’s available, year-round, at testing centers throughout
the world, from IFTA’s computer-based testing provider, To Register
Pearson VUE. Please visit our website at http://www.ifta.org/
certifications/registration/ for registration details.
The CFTe II exam incorporates a number of questions
that require essaybased, analysis responses. The Cost
candidate needs to demonstrate a depth of knowledge IFTA Member Colleagues Non-Members
and experience in applying various methods of technical CFTe I $550 US CFTe I $850 US
analysis. The candidate is provided with current charts CFTe II $850* US CFTe II $1,150* US
covering one specific market (often an equity) to be
analysed, as though for a Fund Manager. *Additional Fee (CFTe II only): $100 US Proctor Fee

Master of Financial Technical Analysis (MFTA) Program

IFTA’s Master of Financial Technical Analysis (MFTA) Timelines & Schedules
represents the highest professional achievement in the There are two MFTA sessions per year, with the
technical analysis community, worldwide. Achieving this following deadlines:
level of certification requires you to submit an original
body of research in the discipline of international SESSION 1
technical analysis, which should be of practical “Alternative Path” application deadline February 28
application. Application, outline and fees deadline May 2
Paper submission deadline October 15
Examinations
In order to complete the MFTA and receive your SESSION 2
Diploma, you must write a research paper of no less “Alternative Path” application deadline July 31
than three thousand, and no more than five thousand, Application, outline and fees deadline October 2
words. Charts, Figures and Tables may be presented in Paper submission deadline March 15 (of the
addition. following year)

Your paper must meet the following criteria: To Register

• It must be original Please visit our website at http://www.ifta.org/
• It must develop a reasoned and logical argument and certifications/master-of-financial-technical-analysis-
lead to a sound conclusion, supported by the tests, mfta-program/ for further details and to register.
studies and analysis contained in the paper
• The subject matter should be of practical application Cost
• It should add to the body of knowledge in the $950 US (IFTA Member Colleagues);
discipline of international technical analysis $1,200 US (Non-Members)
 IFTA JOURNAL 2025 EDITION

How the Deltachart Order Flow and Divergence Nashwan Mohammed Al-Thawr, MFTA
Cairo, Egypt
Delta Candles Works Together to Forecast the nashwan_abc@yahoo.com
01553594346
Price Movement on High Volatile Market

Abstract those executed through the act of hitting the bid price. In
With the technological development in all aspects of life, alternative terms, when Delta exceeds zero, there was a surplus
it was necessary to search for specific tools that search in of buying than selling, whereas when Delta falls below zero,
the depth of the market and give the trader signals about the there was a surplus of selling than buying (3) (“Bar Delta & Delta
movement of smart money that drives the market. For the Divergence”)
same concept, the most important thing about flow order and To illustrate how delta chart divergence functions consider
delta candles is that they show the aggregation when reaching the following examples:
important levels that are difficult for any other indicators
to detect, especially during times of sharp volatility due to 1. Bearish Divergence: Price hit a new high peak, but selling is
news. We will prove in this study that these tools and their more than buying.
combination with technical analysis are very powerful on 2. Bullish Divergence: Price struck a new low point but buying
futures market or the stock market as well we will prove this exceeded selling.
with interactive charts before and after as much as possible.
The Plots are:
Acknowledgement
I want to thank my family, especially my lovely wife, for their 1. A red arrow above the high when price makes higher high on
continuous support during the thesis submission process. As negative delta.
the first Yemeni seeking to establish a name for himself in the 2. A Green arrow below the low when price makes lower low on
financial markets, I am grateful for their support and faith in my positive delta.
endeavor. I am also grateful to my colleagues at the organization
for their great assistance throughout this amazing attempt. TS Trader, which is powered by Tradovate, has this open-
Their assistance has been critical in achieving this goal. source formula which is presented below:

INTRODUCTION Chart 1: (Delta Candle Divergence Formula)

1.1 How Delta Chart Operates

Delta chart trading is a technical analysis method used by
traders to evaluate asset buying and selling pressure over
a period of time (1) (“What Is Volume Delta Indicator?”) This
strategy is based on delta volume, which refers to the difference
between the buying and selling volumes at each price level.
Traders can more easily pinpoint areas of high liquidity and
places where price action is expected to occur when using delta
charts and Japanese candlestick patterns. Additionally, this
strategy helps traders to make informed trading decisions by
providing them with a clear picture of the supply and demand
forces behind the price movements using Japanese Candlestick
patterns.
This study aims to comprehensively examine the concept
of delta chart trading and assess its effectiveness within the
futures markets through the utilization of multiple live trade
examples.

1.2 How Delta Chart Divergence Operates

When making decisions in trading based on volume,
understanding the delta is essential (2) (Dubey)
The disparity exists in the contrasting quantities of market
orders executed through the act of raising the offer price and

PAGE 36 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Charts can be utilized as a tool for examining the historical is customary for expressing positive Delta values, which
fluctuations in currency prices, represented graphically over signal an inclination towards buying pressure. Conversely,
a specific period. These indicators are subsequently employed red hues are traditionally employed to depict negative Delta
to evaluate current prices and formulate forecasts regarding values, indicating a propensity toward selling pressure.
future price fluctuations (4) (“12-Month Japanese Candlestick 5. Analysis and Interpretation: The Delta Chart is employed
Patterns”) by traders to derive information pertaining to the disparity
Delta candles are widely recognized and utilized as a between buying and selling pressure at different price levels.
significant analytical instrument within financial markets, The individuals engage in the identification and analysis
primarily employed for the purpose of predicting and projecting of patterns, trends, and discrepancies in relation to the
future price fluctuations. The candlesticks exhibit distinct fluctuations in Delta and price (Chen)
components, including a body and shadows, which facilitate the
recognition of bullish and bearish market conditions. • Accumulation: The presence of a significant accumulation
By utilizing these visual cues, traders are able to accurately of positive or negative Delta in specific areas can signal
identify optimal entry and exit positions within financial robust buying or selling zones, respectively. These zones
markets. Delta candles are regarded as a valuable instrument may act as future support or resistance levels.
for individuals at all levels of trading proficiency, encompassing • Divergence: Divergence refers to disparities between
both novices and seasoned practitioners (4) (“12-Month Japanese Delta value changes and corresponding price movements,
Candlestick Patterns”) which can serve as valuable trading indicators. For
The body of a delta candle serves as an illustrative instance, instance, when prices rise but Delta decreases (showing
represents the thickest section of the candlestick, and signifies negative divergence), it may suggest a decline in buying
the difference between the opening and closing prices. A range pressure and potential trend reversal.
of patterns is utilized to indicate the formation of these candles • Momentum: Analyzing changes in Delta values in terms of
(4)
(“12-Month Japanese Candlestick Patterns”) speed and magnitude provides insights into the strength
A wide range of tools can be utilized to analyze price action of buying or selling pressure, indicating momentum.
in financial markets. These tools include trend lines, price
channels, moving averages, Fibonacci retracement levels, It is noted that Delta Charts primarily rely on order flow and
Bollinger Bands, as well as technical indicators such as the volume data, and their interpretation may exhibit variability
Relative Strength Index (RSI), Stochastics, the Average True contingent upon the trader's strategy and the particular market
Range (ATR), and the MACD indicator. under analysis. Furthermore, Delta Charts are frequently
There is no obvious differentiation between delta candles employed in conjunction with other technical analysis
and Japanese candles, as both are utilized for the purpose of instruments to facilitate informed trading judgments.
analyzing price action in a similar manner. Candlestick charts
can be analyzed using a range of analytical tools, such as MATERIALS AND METHODS
moving averages, support and resistance levels, and candlestick The researchers commence their work by formulating a
patterns. comprehensive research strategy that outlines the objectives
The delta chart typically works: of the study, the methods of data collecting, and the analytical
approaches to be employed. The initial phase involves doing
1. Data Collection: Delta Chart collects data on trading volume a comprehensive review of the existing body of literature in
and order flow at various price levels. This information can order to enhance understanding of the topic and identify any
be obtained through market exchanges or through order areas where knowledge is lacking. Consequently, this process
flow software. enables the formulation of research inquiries and hypotheses,
2. Calculation of Delta: The calculation of the Delta value which afterwards serve as the overarching structure for the
involves subtracting the selling volume from the buying subsequent phases of the investigation.
volume for each price level. A positive Delta denotes an Following this, the researchers proceed with the gathering
increase in buying volume, while a negative Delta indicates a of data, which includes both qualitative and quantitative data.
rise in selling volume. The Delta value serves as an indicator Qualitative data is obtained through conducting in-depth
of the overall buying or selling pressure observed at a interviews with participants, which serves as a medium for
specific price level (5) (“Delta and Cumulative Delta: How individuals to express their experiences and perspectives
Could They Help a Day Trader?”) regarding DELTACHART. On the other hand, quantitative
3. Visualization: The graphical representation of Delta values is data is obtained through the administration of surveys and
commonly observed on the y-axis of the chart, usually in the questionnaires, facilitating the acquisition of statistical data
shape of bars or lines, while their corresponding price levels and the quantification of distinct factors. The integration
are shown on the x-axis. Each discrete bar or line on the of qualitative and quantitative data is intended to provide
graph represents the Delta value that is linked to a specific a holistic comprehension of the impact and effectiveness of
pricing point. DELTACHART.
4. Color Coding: In order to enhance the process of visual Upon obtaining the available data, the researchers utilize a
interpretation, it is common practice to employ color-coded range of statistical techniques and software tools to carry out
Delta bars or lines. In general, the use of green or blue colors a thorough investigation. The objective is to discover patterns,

IFTA.ORG PAGE 37
 IFTA JOURNAL 2025 EDITION

trends, and correlations within the dataset. The purpose of this allow traders to analyze Depth of Market (DOM) data, hence
analytical undertaking is to identify and evaluate the merits facilitating the identification of patterns, trends, and the
and drawbacks of DELTACHART, as well as explore its possible prevailing market sentiment. For example, individuals have
uses and limitations. The research results are then analyzed and the ability to closely examine the changes in the order book
presented in a comprehensive research report, which includes throughout a period of time, evaluate the levels of market
specific suggestions for educators, policymakers, and other liquidity, determine the strength of support and resistance
stakeholders who are considering the adoption of DELTACHART levels, monitor the impact of significant orders, and identify
in their own settings. possible turning points in the market. Trading professionals can
In summary, the technique utilized in the investigation enhance their trading methods by incorporating CME Globex
of DELTACHART is distinguished by its rigorous and DOM data into DeltaChart in a seamless manner. These tactics
systematic precision. The process involves a thorough consider both the fluctuations in prices and the underlying
examination of relevant scholarly sources, followed by the complexities of the market. As a result, traders have the
gathering of data using both qualitative and quantitative ability to make informed judgments that might potentially
methods. The data undergoes thorough examination using result in more profitable trades and a more comprehensive
statistical methodologies, resulting in useful insights into the understanding of market dynamics.
effectiveness and performance of DELTACHART. In general, this CME Globex Depth data refers to the comprehensive market
methodological approach guarantees that the study is thorough, depth information provided by the electronic trading platform,
reliable, and relevant to the requirements and concerns of Globex, developed by the CME Group. The information includes
educational professionals and policymakers. the most competitive bid and ask prices, along with the related
order volumes at various price levels. The assessment and
2.1 Importing CME Globex Depth of Market Data application of CME Globex Depth data generally encompass the
CME Globex Depth of Market Data is a data offering provided subsequent stages:
by CME Group. This product consists of all necessary market
data messages for the recreation of the order book. Trade data 1. Determine Data Requirements: Initiate the process by
for all products traded on CME Globex is included, as well as five precisely identifying the requisite market depth data for
to ten orders deep in futures markets and three orders deep in the designated CME markets and financial instruments.
options markets that are provided by the Market Depth files(6) CME Group provides a wide range of futures and options
(“Market Depth - Electronic Platform Information Console - contracts that encompass a diversified selection of asset
Confluence”). classes.
In order to facilitate the importation of this data, it is 2. Choose a data vendor: Typically, gaining access to CME
imperative to possess a formal clearing firm relationship with Globex Depth data requires subscribing to a data provider
CME Group, as well as a CME Group-certified trading application that supplies the requisite market data services. Prominent
and the necessary connectivity to CME Globex (7) (“Trade on CME data vendors in the financial industry encompass Bloomberg,
Globex - CME Group”) CQG, Refinitiv (formerly known as Thomson Reuters), and
CME Group data can be accessed through a variety of Interactive Brokers. These companies provide users with
channels, including a network of over 300 distribution partners access to both real-time and historical market data, which
worldwide, as well as the proprietary market data platform includes detailed in-depth information.
and the highly regarded cloud distribution capabilities. The 3. Establish Data Feed Connection: Once a data source has
utilization of the CME Data Mine Market Depth files is also been selected, the subsequent action involves establishing
possible, as these files are exclusively accessible in FIX/FAST a connection to their data feed. The process often entails
format. establishing an account, obtaining the necessary access
In recent times, a number of service providers have emerged credentials, and configuring the trading or analytical
in the market, offering data services tailored specifically for platform to establish a connection with the data feed.
traders. Notable examples include E-Signal, Ninjatrader, ATAS, 4. Import and utilize the data: Once a connection has been
Clusterdelta, and Volfix. successfully established, the next step involves importing
In this study, we will be utilizing the Clusterdelta data that CME Globex Depth data into your trading or analytical
has been imported into the Metatrader 4 platform, along with platform. This particular procedure may exhibit variability
the Go Charting. contingent upon the platform and data provider. In general,
The integration of CME Globex Depth of Market (DOM) data data can be accessed by APIs (Application Programming
into DeltaChart has the potential to greatly enhance the trading Interfaces) or data feed protocols such as FIX (Financial
knowledge and understanding of both traders and others with Information Exchange) or custom APIs offered by the data
an interest in the financial markets. The CME Globex Depth of provider.
Market (DOM) provides immediate access to bid and ask prices, 5. Analyze and Interpret Market Depth: The study and
as well as different contract volumes. Traders can enhance their interpretation of market depth involve examining the
understanding of market dynamics and make informed trading imported data to gain a deeper understanding of the
decisions by effectively integrating this data into DeltaChart, a dynamics of supply and demand, patterns of order
comprehensive platform for charting and technical analysis. movement, and the levels of liquidity in the CME Globex
The extensive functionality and tools offered by DeltaChart market. The acquisition of this vital knowledge can be

PAGE 38 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

utilized to make informed trading decisions or develop to fluctuations in the underlying asset, so providing valuable
profitable trading strategies. insights into the potential for profitability and associated risks.
The process of analyzing the Delta Chart includes evaluating
2.2 How to read Delta Chart and Delta Chart both the slope and shape of the chart. A positive slope indicates
Divergence that the value of the option increases in tandem with the growth
Analyzing price movements and comparing them with the in price of the underlying asset, whereas a negative slope implies
delta chart Histogram is a crucial aspect in financial analysis. a loss in value as the price of the asset increases. Steep gradients
This involves identifying reverse candle patterns and examining on a graph are indicative of a greater degree of price sensitivity,
Volume Delta Divergence behavior, wherein a bullish candle is whereas shallower gradients suggest a lesser level of sensitivity.
observed alongside a red delta volume indicating selling, and The utilization of Delta Chart Divergence is advantageous
vice versa. in discovering discrepancies that arise between the Delta
In order to proficiently comprehend and assess the divergence Chart and the real-time price fluctuations of the underlying
of a Delta Chart, it is imperative to examine the data depicted asset. The analysis of these discrepancies enables traders to
on the chart with the aim of comprehending the intricacies of forecast potential fluctuations in the price of the option and
buying and selling pressure dynamics. the trajectory of the underlying asset. A positive divergence
It is imperative to underscore that the utilization of Delta arises when the Delta Chart indicates a bullish attitude, while
Chart analysis and divergence analysis should not be regarded the price of the asset exhibits a contrary movement, indicating
as independent methodologies. In order to make informed the potential for a reversal or correction. On the other hand,
trading decisions, it is advisable to supplement them with a negative divergence is observed when a bearish Delta Chart
additional technical analysis tools and indicators. Regular coincides with a bullish trend in asset prices, suggesting an
practice and accumulated expertise in analyzing Delta Charts imminent correction or change in market sentiment.
and identifying divergences will improve one's ability in Proficiently examining the divergence of Delta Charts
effectively employing these tools. can furnish traders with a distinct edge, facilitating the
identification and exploitation of lucrative prospects, while
adeptly mitigating hazards in their trading pursuits.
Chart 2: Entered a selling position with EURUSD (it's very
clear that bullish candle reached 1.0649 with aggressive
selling activity in the resistance area confirmed by Delta RESULTS
divergence behavior (when the candle is bullish while the
delta volume is red 3.1 Trading Delta Chart with Candle Patterns in
High Volatile Market
Engaging in transactions within a market characterized by
significant volatility is a considerable obstacle, requiring the
implementation of advanced tactics to effectively negotiate the
rapid changes in prices. One approach that can be employed
involves the employment of a trading delta chart, which is a tool
that provides significant insights into the relationship between
an option's price and changes in the value of the underlying
asset. This tool is commonly utilized by experienced traders,
enabling them to make informed judgments and capitalize on
chances that arise in tumultuous markets.
In order to comprehend the efficacy of utilizing trading delta
charts in market situations characterized by high volatility,
it is imperative to get a comprehensive understanding of the
notion of delta. The symbol delta is used to represent the pace
at which the price of an option varies in response to variations
in the value of the underlying asset. The relationship between
delta values and strike prices is visually represented through
the use of a delta chart. This chart displays the delta values
corresponding to different strike prices. A larger delta value
signifies a more robust link between the price of the option and
Proficiency in comprehending and analyzing Delta Charts the price of the underlying asset. In settings typified by swift
and Delta Chart Divergence holds paramount significance for fluctuations in prices, such as markets with high volatility, delta
investors and traders seeking to make well-informed judgments charts serve as a valuable tool for traders to detect options
inside the realm of financial markets. The Delta Chart is a that possess elevated delta values. The potential for increased
graphical depiction that illustrates the relationship between market volatility may result in greater financial gains or losses
the price of an option and the corresponding changes in the for these alternatives.
price of the underlying asset. The given statement effectively The astute analysis and effective application of trading
illustrates the manner in which the price of an option responds delta charts in exceedingly volatile markets are of utmost

IFTA.ORG PAGE 39
 IFTA JOURNAL 2025 EDITION

importance. Experienced traders possess the analytical acumen as engulfing patterns, hammers, or shooting stars, have
required to evaluate the patterns and trends depicted in the the potential to indicate trend reversals in markets
chart. The ability to identify crucial levels of support and characterized by extreme volatility. When combined
resistance and predict possible breakout or reversal points is with Delta chart order flow research, these patterns
facilitated by analyzing past delta values. Moreover, a thorough offer additional validation and increase the probability of
comprehension of options pricing models and efficient risk a successful reversal trade.
management strategies significantly augments their ability b) Breakout Patterns: Volatile markets frequently
to make prudent selections by utilizing the delta chart. experience breakouts from critical support or resistance
Individuals who possess this intellectual groundwork are more levels. Candlestick breakout patterns such as breakouts
aptly prepared to navigate the complexities of highly unstable from consolidation or breaches of trendlines can signify
markets and effectively optimize their trading techniques in the potential continuation of the existing trend. The
response. integration of Delta chart order flow analysis aids in the
In conclusion, the utilization of a trading delta chart identification of robust buying or selling pressure during
offers significant value in the context of market situations these breakout scenarios, bolstering trader conviction.
characterized by high volatility, enabling traders to make c) Determining Stop Loss Levels: The determination of
decisions that are better informed. Understanding the stop loss levels can be enhanced by the utilization of
relationship between the price of an option and the value of candlestick patterns in conjunction with Delta chart
the underlying asset, as illustrated by delta values on a chart, order flow analysis. Traders have the ability to create
is crucial for exploiting opportunities in volatile markets. As a stop losses at levels that are less likely to be triggered
result, individuals possessing advanced cognitive abilities and by short-term market noise by watching order flow and
a deep understanding of market dynamics are more adept at Delta behavior around crucial support or resistance
capitalizing on and maneuvering through the intricate nature levels that have been discovered through candlestick
of exceedingly unstable market circumstances through the patterns.
utilization of trading delta charts.
The integration of Delta chart order flow trading with By integrating these two analytical methodologies, traders
candlestick patterns in exceptionally volatile market conditions are endowed with enhanced capabilities to traverse the
can provide traders with important insights into market intricacies of exceedingly volatile markets, provide judicious
dynamics and potential trading opportunities. Here is an trading judgments, and efficiently exploit chances.
explanation of the potential outcomes that can be achieved by
combining these two approaches: 3.1.1 Trading Non-Farm Payroll, CPI Index, and
other major news
1. Order Flow Analysis Chart: Participating in trading activities focused on non-farm
a) Detecting Buying and Selling Pressure: The examination payroll, CPI index, and other noteworthy news releases
of order flow plays a crucial role in the real-time requires a heightened level of intellectual capacity and
detection of buying and selling pressure by closely thorough comprehension. As a graduate-level student with a
examining the movement of orders. The Delta chart specialization in finance, I have cultivated a deep understanding
provides a visual depiction of the disparity between of the complexities involved with economic indicators and have
buying and selling volume, serving as a graphical comprehended their potential implications on the financial
representation of the order flow imbalance. markets. Staying knowledgeable and skillfully evaluating
b) Validating Candlestick Patterns: The validation of market emotion are crucial foundations for producing accurate
candlestick patterns can be achieved by the utilization predictions and skillfully navigating the field of trading. With
of delta chart order flow analysis. For example, when a a solid foundation of information and analytical abilities,
bullish candlestick pattern arises and the Delta chart engaging in the trading of significant news releases can become
reveals a positive Delta or strong buying pressure at that a powerful tool for attaining financial prosperity.
particular point, it reinforces the bullish indication. The This action observed in chart (3) is evident at an early stage.
process of validation serves to enhance the trader's level Non-farm employment appeared with a total of 311,000 jobs
of confidence in the observed candlestick pattern. and hit the expectations of 224,000. As a result, EURUSD
c) Spotting Institutional Activity: The analysis of Delta experienced aggressive selling activity, reaching the resistance
chart order flow is essential in identifying instances of level of 1.06947.
institutional action within the market. The impact of This was shown by the divergence pattern on the Delta Chart,
orders placed by major investors on price and volume which was seen in a market that was extremely volatile as the
is substantial, and their identification can be achieved price rapidly increased upwards and stopped at the level of
through the examination of order flow. Cognitive ability resistance with a shooting star candle formation.
demonstrates notable advantages in unpredictable This would imply that traders have already begun buying
market circumstances, wherein the participation of USD in preparation for an upcoming increase in Non-Farm
institutions might lead to significant changes in prices. Employment, which has the potential to lead to an additional
2. Candlestick Patterns In Volatile Markets: interest rate increase by the FED. The Delta Chart's divergence
a) Reversal Patterns: Candlestick reversal patterns, such behavior further supports this interpretation. EUR/USD price

PAGE 40 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

movement between 10:00 AM and 11:00 AM is shown in this Furthermore, the unemployment rate experienced a decline
chart. The currency's price dynamics are shown by two red and to 3.5%, instead of the anticipated rate of 3.7%. Consequently,
green lines. Trading volume is shown by two sets of red and the price experienced a rapid rise to 33302.78. It is worth noting
green bars at the chart's bottom. The stock ticker is "EUR/USD," that this upward movement was accompanied by the presence
and the current price is "1.09" in the upper left corner. The chart of selling positions. The confirmation of this observation was
shows "10:00" and "11:00" in the upper right corner. confirmed by the presence of the Delta Divergence pattern,
followed by the appearance of a Doji candlestick.
Chart 3: EUR/USD at the time touching the resistance The positive delta divergence in chart 6 indicates that
before reversing significant market participants, or 'whales,' began to accumulate
more buying contracts. Nobody expected the Dow Jones to set a
new daily high. Despite this, the news caused a drop in prices.
Remarkably, the Dow Jones moved by over 700 pips in the
same day, demonstrating the market's high volatility.

Chart 5: YM Futures Contract Touching Support Area

EUR/USD price movement between 10:00 AM and 11:00 AM

is shown in this chart. The currency's price dynamics are shown
by two red and green lines. Trading volume is shown by two sets
of red and green bars at the chart's bottom. The stock ticker
is "EUR/USD," and the current price is "1.09" in the upper left
corner. The chart shows "10:00" and "11:00" in the upper right
corner.
The red line represents the current EUR/USD price, while the Chart 6: YM Futures reverse and changes its direction
after the positive delta divergence appeared in Support
green line shows the price moving average over time. Moving
Areas
averages measure short- and medium-term price movements.
Prices rising above the moving average indicate an uptrend,
while falling below it indicates a downturn.

Chart 4: EURUSD hit the target and drooped to 1.0635

Chart (7) shows a significant crash in the Crypto market

on November 4, 2022, which was further exacerbated by the
shocking announcement of FTX's bankruptcy and fraudulent
news. Bitcoin prices were expected to fall significantly,
according to traders.

According to chart 5, in the 15-minute chart of the Dow

Jones, after January 6, 2023, the Non-Farm Payroll exceeded
expectations, as 223K employment positions were reported,
surpassing the market's anticipated figure of 200K.

IFTA.ORG PAGE 41
 IFTA JOURNAL 2025 EDITION

However, contrary to forecasts, aggressive buying activity During the CREE earnings announcement, the same
increased unexpectedly. The Delta Volume Divergence verified observation can be made regarding both Stocks and Option
this surprising buying, demonstrating that traders bought Contracts. By analyzing the market depth and Delta Chart, we
more Bitcoins from 15000/16000, which was contrary to can determine the legitimacy of the reversal candle, which in
traders' expectations. Notably, the positive volume was high, an this instance was a bottom formed Doji.
occurrence that had not been seen since 2020. This provided investors with sufficient conviction to buy
CREE stock options. The volume analysis, in particular the
Chart 7: BTC Futures Chart Orders Flow, revealed that pending buy and sell contracts
were concentrated within the Doji candle at the support level.
This indicates that the volume of trading for the succeeding
green candle was bullish, despite the overall volume of trading
being red or bearish. Notably, while options contracts initially
reflected negative sentiment during the announcement decline,
the presence of positive buying orders on options contracts
indicated a comfortable buying sentiment emerging from the
support level.

Chart 9: Cree

The Bitcoin Futures market suffered a significant event on

May 10, 2023, following the announcement of the Consumer
Price Index (CPI) with complex results relating month-on-month
(m/m) and year-on-year (Y/Y) data.
Chart (8) shows that the price of Bitcoin fell from 28246.15
to 26775.88 when we suddenly noticed an aggressive buying
activity with Delta Volume Divergence indicating that traders
bought more Bitcoins confirmed by Hammer pattern formation The chart no. 10 below demonstrates a clear reversal pattern
indicating that the market moved up against the trader's whereby the price action demonstrates a movement contrary
expectation from 26970 to 27963 within a few hours. to that of the support area. The claim of this reversal is
substantiated by the existence of a Doji candlestick pattern and
Chart 8: BTC Futures on 10th May 2023 a positive order flow Delta volume.

Chart 10: CREE dropped after earning released, it shows

the reverse from the support area supported by Doji
candle + Positive order flow Delta volume with delta
divergence

PAGE 42 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Chart 13: AAPL 9 – 20 June 2023

According to the filing made with the Hong Kong Stock
Exchange (HKEX), Morgan Stanley's ownership stake in Xpeng,
a company involved in the production of electric vehicles,
had a significant surge in volatility. Specifically, their long
position in Hong Kong shares of Xpeng escalated to 6.13%. The
aforementioned rise was substantiated by a typical trading
volume observed in the option contract associated with a strike
price of $14 on the 7th of July.
At the outset, traders observed that the substantial
trading volume associated with a bearish candle might
potentially suggest a persistence of the downward trend in
prices. Nevertheless, the Delta Chart Divergence analysis
indicated that the apparent increase in trading volumes was
predominantly attributable to robust buying orders for the
call option contract, which was initially priced at $0.42 and
experienced a significant jump to $1.32.

Chart 11: XPEV 7th July.23 – Call $14 Strike

After experiencing a notable decrease in prices after to a
gap-up opening in the American market, concerns emerged
regarding the possibility of a stock price collapse. The earlier
apprehension increased due to the observation of a red
engulfing candlestick pattern.
Nevertheless, the delta candle offered significant insights
by uncovering that these downward movements were indeed
accompanied by buying activity from important market
participants. Consequently, there was a subsequent increase in
price from 430 to 436 on the next day.
This observation not only facilitated the understanding of
investor behaviors but also emphasized the impact of their
buying actions in reducing the initial downward momentum.
Chart 12: XPEV 7th July.23 – Call $14 Strike hit $1.30
Chart 14: NVDA 16 Jun.23

The significance of Delta Divergence becomes evident

when conducting an analysis of stocks, particularly in the
case of AAPL. During a period of price drops, the Delta Candle
signifies the strategic choice made by buyers to engage in the
buying of stocks from the support areas. The buying behavior
is additionally affirmed by the existence of a bullish Doji In times of uncertainty, there exist critical moments which
candlestick pattern. have importance for individuals engaged in trading activities.
The comprehension of the candle's depth and the examination
of the order flow are essential in determining whether these
market movements signify a desire to maintain the existing
trend or suggest a possible reversal. Within the context of the

IFTA.ORG PAGE 43
 IFTA JOURNAL 2025 EDITION

SPY box, it is of utmost importance to ascertain the verification 3.2 Trading in Natural Volatile Market Conditions
of deceptive highs and lows, as these occurrences serve as The GBP/USD chart below exhibits an evident upward
compelling evidence during periods of recovery. direction in price, originating from the support area. The
The prevailing pattern is mainly supported by the existence increase in price was accompanied by the formation of a
of reversal candlestick patterns, thereby enhancing the Morning Star candle pattern, which was further supported by a
significance of identifying these patterns in market analysis. notable increase in aggressive buying activity. Furthermore, an
occurrence of Delta Divergence was identified, characterized by
Chart 15: SPY 15 May – 19 June.23 the presence of bullish candles and a positive delta histogram.
These incidents occurred under normal market conditions.

DISCUSSION
4.1 Advantages and Disadvantages of Delta Chart
trading

4.1.1 Advantages of Delta Chart Trading

The utilization of delta chart trading provides numerous
advantages to traders in comparison to alternative technical
analysis tools.
First of all, the utilization of this tool enables individuals
to observe and analyze the supply and demand forces that
affect the market at any given time. The use of this approach
facilitates the identification of regions characterized by higher
liquidity levels, as well as the identification of potential levels of
support and resistance.

Chart 16: GBP/USD 21st April 2023

PAGE 44 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

The second advantage to consider is that the utilization of assist traders in identifying divergences, reversals, breakouts,
delta chart trading enables traders to enhance their decision- and trends based on volume activity (9) (“TradingView – Track
making process by providing them with a clearer picture of All Markets”)
market trends and potential reversal points. Trading Japanese candles on Support and Resistance levels
Furthermore, one notable advantage of using the delta chart uses candlestick patterns with support and resistance levels to
trading methodology is its effectiveness across different market make sound trading decisions. Support and resistance levels are
conditions. In order to effectively analyze price action and areas where buyers and sellers have set up defenses and where
execute profitable trades, traders have the option to employ prices may shift direction. Candlestick patterns can assist
delta charts, regardless of whether the market is exhibiting a traders in confirming or anticipating these movements, allowing
trending or consolidating behavior. them to enter or exit transactions accordingly.
Finally, delta chart trading can be applied to different asset
classes, including stocks, futures, options, and currencies. CONCLUSION
The utilization of delta chart trading as a technical analysis
4.1.2 Disadvantages of Delta Chart Trading instrument holds significant value for traders, as it enables
While delta chart trading offers numerous advantages, it is them to assess the levels of buying and selling pressure of an
important for traders to be aware of the drawbacks associated asset. This approach facilitates the identification of regions
with Delta Chart Trading. characterized by substantial liquidity and the possibility
Firstly, one of the primary drawbacks is that it is challenging of encountering potential support and resistance levels. By
to accurately interpret the delta values. In order to effectively providing valuable perspectives on the dynamics of supply and
employ this strategy, traders must possess an in-depth demand that influence fluctuations in prices, it enables traders
understanding of the market dynamics and the various factors to make informed and rational choices.
that influence the delta values. However, the successful application of delta chart trading
The second disadvantage to consider is that delta values can requires a thorough comprehension of the fundamental market
fluctuate rapidly, making it challenging to keep up with market dynamics and the various factors that influence delta values.
shifts. In addition, it is recommended to integrate delta charts with
Lastly, one additional drawback associated with delta chart supplementary tools such as candle patterns, volume analysis,
trading is its limited ability to offer a comprehensive depiction fundamental analysis, and market sentiment indicators in
of market dynamics. While it effectively displays the buying order to obtain a comprehensive understanding of the market.
and selling pressures associated with different price levels, it Incorporating delta chart trading has the potential to enhance
does not consider the overall market sentiment, news events, traders' decision-making processes and improve profitability
or fundamental analysis. In order to obtain a comprehensive within financial markets.
understanding of market conditions, traders are required to
utilize additional tools in conjunction with delta charts. 5.1 Improving the Trend direction with imbalance
areas (Future Improvements)
4.1.3 Which is better Delta Chart Candles analysis or Trading The Imbalance indicator from Cluster Delta is specifically
Japanese candles on Support and Resistance levels designed to emphasize trade imbalances that occur during the
Both the Delta Chart Candles analysis and the Trading trading process. Imbalances commonly arise when there exists
Japanese candlesticks on Support and Resistance levels a substantial disparity in the ratio between buyers and sellers,
techniques utilize candlestick patterns to identify trading such as a ratio exceeding 3 to 1 or 4 to 1 in either direction or
opportunities by examining volume and price movements. when one side is entirely absent. This indicator effectively
Both the Delta Chart Candles analysis and the Trading demonstrates various imbalances, such as the ratio of buyers
Japanese candles on Support and Resistance levels techniques to sellers within a specific price range and market imbalances
utilize candlestick patterns to identify trading opportunities where the Ask price consistently surpasses the Bid price.
by examining volume and price movements. However, these The presence of imbalances is readily apparent on cluster
techniques for trading possess distinct merits and drawbacks charts, and this particular indicator enhances their visual
reliant upon the market conditions, the time frame, and the representation on price charts.
trader's preferences (8) (“CVD - Cumulative Volume Delta Candles The ability to interpret imbalances in buying orders and
— Indicator by TradingView”) selling offers is essential for validating the analysis of order
The analysis of Delta Chart Candles involves the utilization flow and the delta chart. The provided statement effectively
of intra-bar information in order to acquire volume delta identifies the specific points of entry that lead to an imbalance
information that is more accurate compared to approaches that between the supply and demand within a given candlestick.
solely rely on the timeframe of the chart. Furthermore, the recurrence of these imbalances at close
The volume delta refers to the disparity between the volumes price levels serves as an indication of successive transactions
of upward and downward movements within a bar, serving carried out by substantial market participants, thereby
as an approximation of the level of buying or selling activity supporting the strength of the trend in either buying or
exerted on an instrument. This methodology has the potential to selling. The identification of imbalances and the assessment

IFTA.ORG PAGE 45
 IFTA JOURNAL 2025 EDITION

of the persistence of buying or selling trends can be efficiently As the price of gold fell below the level of 2014, the presence of
achieved by employing Footprint and Order Flow analysis selling pressure became apparent, even though the delta chart
techniques. order lacked clarity.
The market makers' decision to lower the price was confirmed
Chart 17: GC 16th June 2023 by the consistent execution of consecutive selling deals, as well
as the imbalances observed following minimal buying activity.
All subsequent activities to return the price level from 2014 to
2012 were unsuccessful, as resistance zones promptly formed.
This further confirms the existence of substantial selling orders
within those regions.
The observation of price imbalances can yield valuable
insights into investor behavior, informing subsequent buying
or selling decisions. This finding instills a sense of assurance
regarding the strength of the observed trend.
The provided chart below depicts the consistent upward
movement of gold, despite occasional instances of corrective
movements. Nevertheless, these corrections can be interpreted
as a strategic move by market makers to strengthen their
buying orders and consequently increase the price.

Chart 19: GC hit target on 16th June 2023

The presence of both positive and negative signals in
the trend, as validated by the analysis of Order Flow and
Delta charts, revealed clear price imbalances in gold upon
encountering resistance levels. The observed disparities
unveiled the aggressive behavior of important market
participants, commonly referred to as "whales," and their
clustering in particular regions and locations where deals
were conducted. The price promptly reverted to these regions,
implying their importance as levels that large investors would
probably uphold when making the decision to sell.
This was obvious in the provided illustration of the gold
chart. After encountering resistance at the 1967 level, the price
experienced a subsequent decrease. Significantly, upon reaching
the identical level on the minute chart, the price encountered
a pronounced downward rebound. The observed behavior,
in conjunction with the observed price imbalance, serves as
evidence of the dominant selling pressure in contrast to the
relatively weaker buying orders. This can be attributed to the Enhancing the precision of market direction determination
encounter of a critical price resistance area that affects the through imbalance areas could benefit from several strategic
direction of gold. considerations:
1. Finesse Imbalance Area Analysis: The focus is on
Chart 18: GC Dropped on 16th June 2023 continuously refining and improving the existing approach
to ensure more accurate and comprehensive results. This
study aims to explore alternate methodologies for accurately
detecting and delineating these regions by considering the
inclusion of particular volume thresholds or corroborating
imbalances with other technical indicators.
2. Incorporate Multiple Timeframes: Utilize a multi-timeframe
methodology while conducting an analysis of areas of
imbalance. Evaluating the alignment of these regions
across different timeframes might offer a more thorough
perspective on the overall market trend and prospective
trading prospects.
3. Cross-Validation with Other Techniques: Cross-validation
in conjunction with other analytical techniques can be
employed to validate and corroborate findings related
to regions of imbalance and market direction. The

PAGE 46 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

incorporation of imbalance area analysis into market ENDNOTES

assessments, when combined with supplementary technical (1) This source provides insights into the Volume Delta indicator and its
analysis tools such as trendlines, chart patterns, or application in trading.
oscillators, has the potential to strengthen evaluations of (2) This is a GitHub repository maintained by Sanjay Dubey, containing
market direction and enhance trading decisions. custom trading indicators for the Tradovate platform.
4. Leverage Volume Profile Analysis: Utilize the technique (3) This source explains the concepts of Bar Delta and Delta Divergence
of volume profile analysis to gain a more comprehensive in the context of order flow trading.
understanding of volume distribution within areas of (4) This source discusses 12-month Japanese candlestick patterns in the
imbalance. The identification of significant volume clusters context of forex trading.
can provide further confirmation of support or resistance (5) This source explores the benefits of using Delta and Cumulative
levels and aid traders in assessing the intensity of market Delta indicators for day trading.
imbalances. (6) This source provides information related to market depth on the
5. Back testing and Data Examination: Conduct comprehensive CME Group's electronic platform.
back testing and data analysis to verify the effectiveness (7) This source explains the process of trading on the CME Globex
of imbalance areas in forecasting market trends. By doing platform, operated by CME Group.
an analysis of historical data and conducting a comparative (8) This source introduces the Cumulative Volume Delta (CVD) indicator
examination of trade outcomes in relation to regions of available on the TradingView platform.
imbalance, traders can enhance their strategies and identify (9) TradingView is a versatile platform for tracking various financial
any recurrent patterns or biases that may impact market markets and related data.
trends.
6. Continuous Learning and Adaptation: Continuous learning
and adaptation are essential for individuals to remain up to
date with the latest advances and innovations in the field
of market analysis. Participate in workshops, webinars, or
educational courses to enhance one's knowledge in the field
of imbalance area analysis and its potential for determining
market trends.

REFERENCES
“12-Month Japanese Candlestick Patterns.” Admirals, 20 Mar. 2023,
admiralmarkets.com/ar/education/articles/forex-basics/namathej-
alshomoo. Accessed 1 Sept. 2023.
“Bar Delta & Delta Divergence.” Emoji Trading, 7 July 2019, www.
emojitrading.com/docs/order-flow-basics/key-order-flow-traded-
volume-concepts/delta-delta-divergence/.
“CVD - Cumulative Volume Delta Candles — Indicator by TradingView.”
TradingView, 4 June 2022, www.tradingview.com/script/NlM312nK-
CVD-Cumulative-Volume-Delta-Candles/. Accessed 7 July 2023.
“Delta and Cumulative Delta: How Could They Help a Day Trader?” Atas.
net, 9 Sept. 2019, atas.net/atas-possibilities/indicators/what-is-
delta/. Accessed 1 Sept. 2023.
Dubey, Sanjay. “Tradovate-Custom-Indicators.” GitHub, 2 Nov. 2022,
github.com/sdmiami/tradovate-custom-indicators.
“Market Depth - Electronic Platform Information Console - Confluence.”
Www.cmegroup.com, www.cmegroup.com/confluence/display/
EPICSANDBOX/Market+Depth.
“Trade on CME Globex - CME Group.” Www.cmegroup.com, www.
cmegroup.com/globex/trade-on-cme-globex.html. Accessed 7 July
2023.
“TradingView – Track All Markets.” TradingView, in.tradingview.com/.
“What Is Volume Delta Indicator?” TabTrader, 23 Nov. 2023, tabtrader.
com/academy/articles/what-is-volume-delta-indicator.

IFTA.ORG PAGE 47
 IFTA JOURNAL 2025 EDITION

Using Renko Charts for Noise Reduction and

Directional Insights in the US Equity Market: Loïc Bellina, CFTe, MFTA
Hauteluce, France
+33635205868
loic.bellina@gmail.com
A Multi-Dimensional Approach Integrating Investment
Factors, Sector Analysis, and Volatility Filters

Abstract in price changes but not time. By offering a time-agnostic

This thesis introduces a multi-dimensional trading visualization, Renko charts excel in highlighting market trends
methodology that intertwines Renko chart patterns, volatility by reducing noise and presenting more discernible and less
expectations, and financial ratios. We present a unique subjective chart patterns.
approach to segmenting implied volatility into distinct contexts Despite their utility, the application of Renko charts remains
and assessing three strategic components: segments of market largely uncharted territory in existing literature and market
conditions, investment factor strategies, and sector trend practices. This paper seeks to fill this gap by investigating
strategies. This methodology is articulated through in-depth how Renko charts, when combined with other analytic tools,
discussions on bullish and bearish positioning rules, selection of can enhance directional insights in the U.S. equity market. To
Renko patterns, exit strategies, and considerations of exposure. achieve this objective, we strive to construct a holistic approach
A key innovation is the Renko Directional Z-score (RDZ), that intertwines market filtering mechanisms using the VIX
a quantitative measure that extends beyond traditional term structure and Treasury yields, a comprehensive sector
visual analysis by objectively quantifying trend strength and trend and relative strength analysis, and factor investing
momentum. The RDZ has multiple applications, notably in strategies.
identifying overbought and oversold conditions. Building on The paper unfolds as follows. We begin with a brief literature
this, we introduce the Leading RDZ, a cutting-edge market review covering the essential elements of our study, followed by
breadth indicator engineered to respond to future volatility a detailed exposition of our backtesting methodology and data
expectations by incorporating the VIX term structure. modeling. Subsequently, we present the results of our empirical
Our approach offers a dynamic and adaptable framework research, focusing on the performance of long/short Renko
for trading operations, empowering traditional Renko chart patterns, optimizing Renko performance with market condition
analysis with a dimension of forecasting market volatility, a filtering, and identifying optimal investment strategies
layer of financial ratios, and a statistical perspective. While across various volatility contexts. We conclude by unveiling
this empirical research is exclusively focused on the U.S. equity a methodology anchored in our findings, introducing two
market, it acknowledges the inherent limitations of historical pioneering indicators centered on Renko charts and volatility,
backtesting and the influence of external factors beyond our and acknowledging limitations while outlining potential paths
predictive capacity. The study not only provides a practical for future research.
and versatile trading compass for navigating diverse market Through this exciting exploration, our aim is to provide
landscapes but also lays the groundwork for future research. a multi-dimensional compass for guiding traders' decision-
This opens avenues for the evaluation of this methodology in making processes and to contribute meaningfully to the
other financial markets and the exciting potential incorporation existing body of literature on technical analysis, particularly
of cutting-edge data analytics, neural networks, and focusing on the integration of Renko chart patterns within
revolutionary AI. diverse volatility contexts.

1. INTRODUCTION 2. BRIEF LITERATURE REVIEW

The global financial crisis of 2008, the more recent COVID-
19-induced market turmoil, and the ongoing conflict in Eastern 2.1. Construction and Relevance of Renko Charts
Europe emphasize the formidable challenge of navigating the The basis of Renko charts is price movement, not time
volatile U.S. equity market. Geopolitical events, inflation shifts, intervals. Each brick on a Renko chart represents a specific
and interest rate dynamics introduce intricacy, complexity, and price increment, or the “box size”. This could be a fixed price
uncertainty, leaving investors grappling with the identification increment, or one dynamically calculated using the Average True
of trends and directional opportunities. Range (ATR), a method that adjusts the brick size in line with
Traditional technical analysis tools like moving averages recent price movements (Wilder, 1978). Renko charts simplify
and oscillators sometimes prove insufficient in filtering out trend analysis by eliminating minor price fluctuations and the
market noise, leading to misleading signals and contradictory time factor, creating an unambiguous representation of price
predictions. Their reliance on fixed time intervals can fail to trends. These charts clearly visualize support and resistance
capture the market's true dynamism. This is where Renko levels and can be used alongside other technical indicators for
charts step in, a type of price charting method that factors a comprehensive market view. However, Renko charts may not

PAGE 48 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

fit time-centric strategies, work less effectively in range-bound conducted by Cao, Wei, and Zhong (2015) and Alizadeh, Brandt,
markets, and are inefficient when tracking prices gaps (Dawson and Diebold (2002).
& Steenbarger, 2021). Their box-size determination, subjectively
set, requires careful adjustment to accurately reflect market 2.3. US Treasury Yield Curve
dynamics and prevent overlooking significant trends or creating Over time, the US Treasury yield curve - especially its
excess noise (Smith, 2020). The effectiveness of Renko charts inversion, where short-term rates surpass long-term ones - has
has been documented by research papers conducted by Bernal acted as a predictor for upcoming economic recessions (Estrella
and Venegas-Martinez (2018), Yang and Su (2019), and Zhang and and Mishkin, 1998). Adrian and Wu (2019) employed machine
Ma (2020). learning techniques to confirm the predictive power of yield
curve shape changes, including the role of the yield curve slope
Figure 1. Price chart and ATR and its curvature. Changes in the yield curve can also impact
risk sentiment and asset prices.

2.4. Common Investment Factors

Factors such as size, value, quality, momentum, and risk
volatility are commonly utilized in the investment sector
to create extra returns (Fama and French, 1993; Asness et
al., 2013). However, their effectiveness may fluctuate across
market cycles, igniting spirited discussions and debates. For
instance, the value factor might underperform during periods
of economic expansion but could outperform during economic
downturns (Baker et al., 2011).

2.5. Top-Down Sector Analysis

The approach of top-down sector analysis - evaluating
macroeconomic data and sector performance to pinpoint trends
and opportunities - has proven effective in generating superior
returns (Sharpe, 1992; Lin and Sun, 2019). Combining sector
Figure 2. Renko chart with ascending triangle breakout analysis with other investment factors, such as momentum and
value, can further enhance investment strategies. However,
we must consider the risks associated with inaccurate
macroeconomic forecasts and unforeseen events. Potential
improvements might involve diversifying across sectors
and integrating bottom-up stock picking with the top-down
approach (Arnott & Bernstein, 2002).

2.2. VIX Index and Its Term Structure

Often referred to as the “fear index”, the VIX Index gauges
market volatility and is derived from S&P 500 index option
prices (Whaley, 1993). The term structure of the VIX index,
which illustrates the relationship between the prices of VIX
futures contracts with multiple maturities, can be a powerful
tool for forecasting upcoming market volatility and risk
sentiment. However, its predictive capacity might decrease
during periods of market uncertainty or following major
economic events like the 2008 Financial Crisis (Engle & Rangel,
2008). This has been demonstrated in studies, such as those

IFTA.ORG PAGE 49
 IFTA JOURNAL 2025 EDITION

3. BACKTESTING METHODOLOGY AND DATA MODELING

3.1. Identifying Twenty Renko Chart Patterns

Figure 3. Selecting Bullish and Bearish Renko patterns

PAGE 50 IFTA.ORG
IFTA JOURNAL 2025 EDITION

IFTA.ORG PAGE 51
IFTA JOURNAL 2025 EDITION

PAGE 52 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

3.2. Market Filtering Conditions: Defining Ten • Size:

Segments • Segment F1: small caps with capitalization of up to $2
In the following specifications, EMA(ticker,5) represents the billion
5-day exponential moving average. The terms 50-SMA and 200- • Segment F2: big caps with capitalization of more than $10
SMA designate the 50-day and 200-day simple moving averages, billion
respectively. • Value:
Our focus is on companies that seem undervalued, as
3.2.1. Four Segments for VIX Filtering Analysis suggested by their price-to-book and current ratios.
• Segment F3: price_book_ratio>0 AND price_book_ratio<2
• Fear and Market Volatility Expectations: AND price_cash_ratio>0 AND price_cash_ratio<20 AND
• Segment V1: VIX<19 current_ratio>1 AND current_ratio<40
• Segment V2: VIX>24 • Quality:
• VIX Term Structure Contango or Backwardation: We target firms with consistent capital returns over recent
• Segment V3: EMA(VIX3M,5)/EMA(VIX,5)>1.1 (contango years and relatively lower debt profiles compared to their
context) industry counterparts.
• Segment V4: EMA(VIX3M,5)/EMA(VIX,5)<1 (backwardation • Segment F4: ROE_5-year_average>15 AND ROI_5-year_
context) average>15 AND Debt/Equity_to_Industry<100%
• Momentum:
3.2.2. Six Segments for US Treasury Yield Filtering Analysis The common guideline is that the trend continues until
TNX is a widely recognized ETF used to measure the yield of substantial evidence indicates a shift. We characterize
benchmark 10-year Treasury notes, while TYX is an analogous an intermediate bullish trend as: [stock_above_50-SMA]
ETF for 30-year Treasury bonds. AND [13-week_relative_strength_above_50%]. A 13-week
relative strength exceeding 50% implies that the stock has
• U.S. Economy Health and Interest Rate Trends: outperformed 50% of its exchange-listed counterparts in the
• Segment Y1: EMA(TNX,5) below EMA(TNX,200) past 13 weeks, or about 3 months. An intermediate bearish
• Segment Y2: EMA(TNX,5) above EMA(TNX,200) trend can be similarly defined as: [stock_below_50-SMA] AND
• Segment Y3: EMA(TNX,5) below EMA(TNX,50) [13-week_relative_strength_below_50%].
• Segment Y4: EMA(TNX,5) above EMA(TNX,50) • Segment F5: bullish intermediate trend on the stock
• Yield Curve Slope, an Indicator of Economic Expansion or • Segment F6: bearish intermediate trend on the stock
Recession:
• Segment Y5: EMA(TYX,5)/EMA(TNX,5)>1.5 (very steep yield 3.3.2. Four Segments for Sector Trend and Relative Strength
curve) Analysis
• Segment Y6: EMA(TYX,5)/EMA(TNX,5)<1 (relatively flat • Sector Primary Trend (Medium to Long Term Trend):
yield curve) In analyzing any given stock, one should refer to its
corresponding sector-based ETF. A primary bullish trend
3.3. Specific Filtering Strategies: Defining Ten is defined as: [ETF_above_200-SMA] AND [ETF_26-week_
Segments relative_strength_above_50%] (26 weeks ≈ 6 months).
Similarly, a primary bearish trend is characterized as: [ETF_
3.3.1. Six Segments for Factor Investing Analysis below_200-SMA] AND [ETF_26-week_relative_strength_
Four key investment factors are considered: size, value, below_50%].
quality, and momentum. • Segment S1: bullish primary trend on the corresponding
sector-based ETF

IFTA.ORG PAGE 53
 IFTA JOURNAL 2025 EDITION

Figure 4. Establishing screeners in MarketInOut

• Segment S2: bearish primary trend on the corresponding
sector-based ETF
• Sector Intermediate Trend (Short Term Trend):
In analyzing any given stock, one should refer to its
corresponding sector-based ETF. A bullish intermediate trend is
defined as: [ETF_above_50-SMA] AND [ETF_13-week_relative_
strength_above_50%] (approximately 3 months). Similarly, an
intermediate bearish trend is characterized as: [ETF_below_50-
SMA] AND [ETF_13-week_relative_strength_below_50%].
• Segment S3: bullish intermediate trend on its
corresponding sector-based ETF
• Segment S4: bearish intermediate trend on its
corresponding sector-based ETF

3.4. Backtesting Methodology, Data Modeling and

Performance Metrics

3.4.1. Data Source and Stock Universe

Data is sourced from the MarketInOut Stock screener
(marketinout.com), chosen for its comprehensive coverage of
market data. This platform's Formula Screener tool allows for
creating expressions to identify instruments matching specific
investment criteria. The Strategy Backtest tool assists in
evaluating strategies against historical data and exporting the
resulting data into CSV files for further analysis.
The criteria for defining the stock universe Ω are:
• Exchange: NYSE, NASDAQ
• Capitalization: Large-cap, Mid-cap, Small-cap
• Security Type: Stock
• Price Is Greater Than 10

3.4.2. Constructing and Executing the Backtesting Procedure

• Renko Tactical Strategies:
Our methodology spans from January 1, 2000, to April 30,
2023. For each of the 20 predefined Renko patterns within the
stock universe Ω, we perform the following steps:
• Run 20 backtests with a long position, a 5-ATR trailing stop
and a 5-ATR take profit
• Run 20 backtests with a long position and a 5-ATR trailing
stop
• Run 20 backtests with a short position, a 5-ATR trailing
stop and a 5-ATR take profit
• Run 20 backtests with a short position and a 5-ATR trailing
stop
In the Strategy Backtest tool of MarketInOut, the portfolio
size cap is set at 500, a limit that was not exceeded in our
testing. The order execution model is based on closing prices.
• Market and Specific Filtering Strategies:
We develop stock screeners for each of the twenty predefined
filtering segments: {V1,V2,V3,V4} + {Y1,Y2,Y3,Y4,Y5,Y6} +
{F1,F2,F3,F4,F5,F6} + {S1,S2,S3,S4}.

Exit stock screeners are also established, incorporating a

“Not” function into their expressions. These screeners form
the foundation for market filtering strategies, with the original
screener being used to initiate positions and the exit screeners
to close them.

PAGE 54 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

The results are exported into CSV files, presenting a data structure as follows:

Figure 5. CSV structure of screener results

IFTA.ORG PAGE 55
 IFTA JOURNAL 2025 EDITION

The collected data then undergoes an ETL (Extraction, Transformation, and Loading) process, for which we use the Microsoft Power
Query Editor to manage data from the CSV sources. The objective is to create a unique entry for each stock on each distinct day that
meets our predefined criteria. As a result, if a filter applied to the table returns a row corresponding to a specific date and stock, it
implies that the stock was included in the corresponding screener on that date. Subsequently, we construct a relational data model and
develop metrics to measure performance and risk. This process enables the investigation of factors that have consistently influenced
outcomes over the 23-year study.

3.4.3. Creating the Power BI Data Model

Figure 6. Microsoft Power BI data model

3.4.4. Implementing Performance and Risk Metrics

We seek to identify statistically significant segments by optimizing a Renko risk-adjusted performance metric. These metrics are
developed using DAX language within the Power BI data model. While our focus here is on the implementation for Renko trades, the
underlying logic is also applied to SPY and the investment benchmarks incorporated in this model.

Figure 7. Renko performance and risk DAX measures

PAGE 56 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

4. RESULTS
4.1. Performance of Long/Short Renko Patterns

4.1.1. An Overview of the Data

The most prevalent strategies are “R04_new_black”, “R01_white_trend”, “R03_new_white”, and "R02_black_trend”. However,
the less frequently appearing strategies, such as “R06_double_top”, “R05_double_bottom”, “R20_bull_trap”, and “R19_bear_trap”
strategies, have maintained a steady presence over the years, suggesting their potential utility in spotting trend reversals. In contrast,
the least observed strategies correlate with complex patterns, such as “R18_long_tail_up_reversal”, “R17_long_tail_down_reversal”,
“R13_bullish_catapult”, “R08_triple_top”, and “R07_triple_bottom”. As these patterns require significant market shifts, their
infrequent occurrence is understandable.

IFTA.ORG PAGE 57
 IFTA JOURNAL 2025 EDITION

Figure 8. Renko trades count over time

The bar charts display performance differentials of various
Renko strategies using two distinct Average True Range (ATR)
exit strategies: “5_ATR+TP” (Take Profit) and “5_ATR”. A higher
differential suggests superior performance for long positions.
The Average Long/Short Renko Performance Differential
consistently indicates superior results for “5_ATR+TP”, implying
that setting a profit target typically yields better results than
relying solely on a trailing stop. Over the last 23 years, Bullish
Renko patterns have consistently outperformed bearish ones,
mirroring the long-term bullish trend of the market.

Figure 9. Renko trade proportions

PAGE 58 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 10. Long/short Renko performance differential

4.1.2. Classifying the Renko Patterns

We propose classifying Renko strategies into groups based on the differential between their long and short performances.
For Bullish Renko Patterns:

• Group A+: Top performers include “R17_long_tail_down_reversal” and “R09_inverted_head_shoulders”, displaying differentials
exceeding 350%.
• Group B+: Mid performers include “R11_ascending_triangle_breakout”, “R19_bear_trap”, “R07_triple_bottom”, and “R15_
symmetrical_triangle_bullish_breakout”.
• Group C+: Lower performers, still respectable, include “R05_double_bottom”, “R01_white_trend”, “R03_new_white”, and “R13_
bullish_catapult”.

IFTA.ORG PAGE 59
 IFTA JOURNAL 2025 EDITION

Figure 11. Classifying Bullish Renko patterns

These findings emphasize the critical need to consider the
specific market context when applying and systematically
evaluating Renko strategies for the best possible outcomes.

4.2. Optimizing Renko Performance across

Volatility Contexts: Top Market Segments
We propose using a machine learning tool, the Power BI Key
Influencers visualization, to conduct a volatility-segmented
analysis. Our aim is to identify and rank influential market
conditions for optimizing the Renko Risk-Adjusted Performance
measure of long positions across various volatility contexts.
This study considers six volatility contextual slicers:

• A: V1_low_VIX=True
• B: V1_low_VIX=False AND V2_high_VIX=False
• C: V2_high_VIX=True
• D: V3_contango=True
• E: V3_contango=False AND V4_backwardation=False
• F: V4_backwardation=True

4.2.1. Context A: V1_low_VIX=True

Over the past 23 years, bearish patterns have shown Figure 13. Market influencers and top segment in volatility
inconsistent performance, indicating that going against segment A
the prevailing bullish market trend often leads to
underperformance. Therefore, recognizing more favorable
market conditions before selecting Renko patterns is key.
For Bearish Renko Patterns under the “V4_backwardation”
context:

• Group B-: Decent performers include “R02_black_trend”,

“R10_head_shoulders”, and “R12_descending_triangle_
breakout”.
• Group C-: Underperformers, with inconsistent results,
include “R06_double_top”, “R16_symmetrical_triangle_
bearish_breakout”, “R04_new_black”, “R20_bull_trap”,
“R08_triple_top”, “R18_long_tail_up_reversal” and “R14_
bearish_catapult”.

Figure 12. Classifying Bearish Renko patterns

PAGE 60 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

4.2.2. Context B: V1_low_VIX=False AND V2_high_VIX=False 4.2.4. Context D: V3_contango=True

Figure 14. Market influencers and top segment in volatility Figure 15. Market influencers and top segment in volatility
segment B segment D

4.2.3. Context C: V2_high VIX=True 4.2.5. Context E: V3_contango=False AND V4_

No influencer found. backwardation=False
No influencer found.

4.2.6. Context F: V4_backwardation=True

No influencer found.

IFTA.ORG PAGE 61
 IFTA JOURNAL 2025 EDITION

4.3. Identifying Optimal Investment Strategies across Volatility Contexts: Top Strategic Clusters
Our analysis conducts a risk-adjusted evaluation of various investment strategies, incorporating both investment factors and
market trend strategies. The insights gained are:

• Risk-Adjusted Performance:
This measure calculates the downside risk associated with the returns yielded by each strategy.
• Risk-Adjusted Performance vs SPY:
This metric evaluates each risk-adjusted strategy against a risk-adjusted long position in the SPY ETF.
• Strategic Clusters:
This methodology groups similar strategies, enabling a more nuanced exploration of the relationships among diverse strategies.

These insights equip investors with essential data to choose strategies aligning with their risk tolerance and return expectations. By
assessing these factors, investors can craft tailored portfolios targeting their financial objectives while balancing potential risks and
rewards.

4.3.1. Context A: V1_low_VIX=True

Figure 16. Investment strategies in volatility segment A

PAGE 62 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

4.3.2. Context B: V1_low_VIX=False AND V2_high_VIX=False

Figure 17. Investment strategies in volatility segment B

4.3.3. Context C: V2_high_VIX=True

Figure 18. Investment strategies in volatility segment C

IFTA.ORG PAGE 63
 IFTA JOURNAL 2025 EDITION

4.3.4. Context D: V3_contango=True

Figure 19. Investment strategies in volatility segment D

4.3.5. Context E: V3_contango=False AND V4_backwardation=False

Figure 20. Investment strategies in volatility segment E

PAGE 64 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

4.3.6. Context F: V4_backwardation=True

Figure 21. Investment strategies in volatility segment F

5. DISCUSSION 5.1.1. Summary of Key Insights from the Empirical Study

5.1. Proposing a Multi-Dimensional Methodology

Figure 22. Key insights: three strategic pillars per volatility
From analyzing Renko patterns, market conditions and segment
investment factors across various volatility contexts, we
propose a practical trading methodology. While not fully data-
driven, this roadmap integrates insights from our previous
study and can be combined with existing fundamental stock
ranking processes.
The proposed approach segments volatility into several
contexts (A…F) and examines three strategic aspects within
each context: top market segment, top strategic cluster for
investment factor strategies and top strategic cluster for sector
trend strategies.

IFTA.ORG PAGE 65
 IFTA JOURNAL 2025 EDITION

Figure 23. Key insights: Renko patterns classification

5.1.4. Exit Strategy
We recommend the following exit guidelines:

• Set a trailing stop and a take-profit level, each at a distance

equivalent to five times the Average True Range (ATR).
• Alternatively, consider a more dynamic exit strategy. Initiate
an exit when a confirmed color change in the Renko box is
detected, such as when a second black box appears during a
bullish sequence of white boxes. Maintain a 5-ATR-trailing
stop for risk management.

In summary, our approach combines Renko chart patterns,

market volatility forecasts, and detailed historical data to offer
a robust, multi-dimensional, and systematic framework for
decision-making in trading operations.

5.1.2. Bullish Positioning Rules 5.2. Introducing the Renko Directional Z-score for
For each volatility context and for a specific stock, we suggest Quantitative Stock Selection
the following rules:
5.2.1. Defining the Renko Directional Z-score (RDZ)
• If two strategic aspects are validated, search for Bullish We introduce a new indicator, the Renko Directional Z-score
Renko patterns in A+, B+, and C+ groups. We recommend (RDZ), to add a fresh perspective to the visual analysis typically
increasing trade exposure by 30%. associated with Renko charts. This tool quantifies the strength
• (For instance, the current volatility context is “SEGMENT and significance of a trend by comparing the current momentum
D: V3_contango” and, for a stock named ABC, two strategic to historical trend using Z-score. For the Bullish RDZ, it
aspects are validated: “F4_qual” and “S1_bull prim_trend”). measures upward trend strength, whereas for the Bearish
• If only one strategic aspect is validated, search for Bullish RDZ, it evaluates downward trend strength. This statistical
Renko patterns in A+ and B+ groups, maintaining regular computation effectively highlights overbought/oversold
trade exposure. conditions and points out potential momentum shifts.
• If no strategic aspect is validated, search for Bullish Renko The RDZ, for both bullish and bearish situations, is calculated
patterns in the A+ group, reducing trade exposure by 30%. as follows:
• When structuring trades, consider either direct stock
investments or 1 standard deviation Out-The-Money (OTM)
debit spreads with a duration of 60 to 120 days to capitalize
on the payoff ratio’s convexity.
Where:
5.1.3. Bearish Positioning Rules ConsecutiveRenkoBars counts the sequence of bullish or
For each volatility context and for a specific stock, we suggest bearish Renko bars for the Bullish or Bearish RDZ respectively.
the following rules: μ stands for the average number of consecutive Renko bars.
σ is the standard deviation of consecutive Renko bars.
• In the C/V2 high VIX context, it may be preferable to adopt
delta-neutral strategies that involve selling option premiums
against stocks with a high implied volatility percentile, rather Figure 24. Normal distribution and Z-scores
than pursuing directional strategies.
• In the F/ V4 backwardation context, if no strategic aspect
is validated, search for Bearish Renko patterns in B- and C-
groups, maintaining regular trade exposure.
• In all other contexts, if no strategic aspect is validated, search
for Bearish Renko patterns in the B- group, reducing trade
exposure by 30%.
• When structuring trades, consider Bearish At-The-Money
(ATM) credit spreads with a duration of 30 to 60 days to
capitalize on time decay, given that bearish technical patterns
have historically shown lower reliability.

PAGE 66 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Overbought Bullish RDZ signals might necessitate position size

reduction or risk hedging.
• Entry and Exit Timing:
The RDZ primarily functions as a timing tool. Bullish RDZ
reversal signals might trigger entry, while Overbought Bullish
RDZ or Bearish divergences on Bullish RDZ could indicate exit
points.
• Multi-Asset Portfolio:
The RDZ can be useful in a multi-asset context with
preference possibly given to asset classes showing Bullish RDZ
within an asset allocation strategy.
• Machine Learning Models:
The RDZ, as a statistical indicator, could contribute to
enhance the predictive accuracy of machine learning models
used for stock selection or ETF allocation.

5.3. Introducing the Leading RDZ: a VIX-Adjusted

5.2.2. Analyzing the RDZ’s Technical Signals Market Sentiment Indicator
Several traditional technical signals can be extracted and
analyzed for enhanced market insights: 5.3.1. Defining the Leading RDZ
With the Leading RDZ, we push Renko analysis up to a new
• Bullish RDZ Reversal Signal: level by engineering a sophisticated and innovative market
This indicates potential uptrends when the Bullish RDZ shifts breadth indicator. It measures the proportion of S&P 500
from negative to positive. stocks displaying overbought Bullish and oversold Bearish
• Bullish RDZ Momentum Signal: RDZ conditions, providing a means to detect market extremes,
This indicates bullish momentum when the Bullish RDZ is confirm prevailing trends and detect divergences signaling
positive and increasing. potential market shifts.
• Overbought Bullish RDZ: The true originality lies in the uniqueness of these thresholds
This suggests overbought stocks when the Bullish RDZ of overextension, which are adjusted according to the VIX term
significantly exceeds its mean by more than two standard structure, represented by the ratio EMA(VIX3M,5) / EMA(VIX,5).
deviations (i.e., when Bullish RDZ > 2) By embedding future volatility forecasts, we have forged a tool
• Bearish Divergence on Bullish RDZ: that is more dynamic, echoing the inherent fluctuation present
This suggests potential bearish reversals when stock prices in market conditions.
peak but the Bullish RDZ does not. Furthermore, we introduce a scaling factor “k” that adjusts in
response to the VIX Z-score. This scaling allows the thresholds
Naturally, comparable signals can be extracted from the for overbought and oversold conditions to marginally increase
Bearish RDZ for a comprehensive analysis. with higher VIX levels and decrease with lower VIX levels,
offering even more granular control and insight into market
5.2.3. Integrating the RDZ into Existing Stock-Picking behavior.
Processes
Consider the RDZ as another tool in your toolbox, one that • Overextended Bullish Component Formula (OBUC):
can be seamlessly integrated into existing quantitative stock-
picking processes:

• Fundamental Analysis Integration:

The RDZ’s signals can enhance fundamental analysis. For Here, “k” is a scaling factor, centered around 1 and responsive
instance, preference could be given to undervalued stocks to future volatility
showing Bullish RDZ reversal signals.
• Quantitative Factor Models:
The RDZ could serve as an additional factor. For instance,
stocks showing a high Bullish RDZ might receive higher
momentum scores. Where μ is the VIX population mean and σ the standard
• Portfolio Construction: deviation.
Stocks displaying strong Bullish RDZ may receive larger
allocations, while those with robust Bearish RDZ could be
underweighted.
• Risk Management:
The RDZ can offer an additional layer of risk management.

IFTA.ORG PAGE 67
 IFTA JOURNAL 2025 EDITION

• Overextended Bearish Component Formula (OBEC): Moreover, external variables such as geopolitical shifts,
policy changes, or Black Swan events represent a factual
limitation. These factors lie beyond the predictive capacity of
any systematic analytical model. Despite providing a roadmap
based on Renko patterns, volatility contexts, and fundamental
ratios, this methodology is not immune to these unpredictable
Here, “k” is the same scaling factor, adjusted with the VIX factors.
Z-score.
5.4.2. Future Research Directions
5.3.2. Visualizing and Interpreting the Leading RDZ This study exclusively targets the U.S. equity market, possibly
The OBUC and OBEC are plotted on a scatter plot, offering a limiting its applicability. Yet, it paves the way for evaluating
visual interpretation of market breadth. Each quadrant on the our methodology's adaptability across varied markets. Training
scatter plot conveys a unique market sentiment: neural networks on historical data could reveal even richer
nuances with Renko chart patterns.
• Quadrant I (High OBUC, Low OBEC): Additionally, leveraging AI in detailed market simulations
This indicates a strong bullish sentiment with most stocks could precisely uncover subtle anomalies like irregular cycles
in overbought conditions, suggesting potential upward market and prevalent trader biases. By combining Renko techniques
trend. with such advancements, we could gain deeper, more actionable
• Quadrant II (High OBUC, High OBEC): market insights. The emergent world of decentralized finance
This represents a volatile or uncertain market, with many (DeFi) and crypto sectors could provide a fertile ground for
stocks either overbought or oversold, suggesting disparity refined Renko analysis. As financial systems rapidly evolve,
across sectors or investment styles, necessitating drill down exploring innovative smart contract strategies, especially
investigation via sector-specific or group-specific Leading RDZ integrating the RDZ, could also become a captivating avenue.
calculations. In summary, the combination of cutting-edge data analytics
• Quadrant III (Low OBUC, High OBEC): and revolutionary AI with traditional technical analysis, could
This indicates a strong bearish sentiment with most stocks very well shape the future landscape for market technicians.
in oversold conditions, suggesting possible downward market This synergistic approach has the potential to increase the
trend. versatility and relevance of technical analysis in a world
• Quadrant IV (Low OBUC, Low OBEC): increasingly influenced by data and AI.
This represents a calm or stagnant market, with few stocks
either overbought or oversold, suggesting less directional 6. CONCLUSION
opportunities and relative market stability. In this study, we embarked on an exploration of the intricate
relationship between market implied volatility, investment
This scatter plot analysis can be applied not just to the S&P strategies, and the less frequently examined ATR-based Renko
500, but also to each sector within the index, and even to chart patterns. Our objective was to construct a systematic,
various investment style groups of stocks, thereby offering a flexible, and practical trading methodology that could serve as
granular view of market sentiment. a reliable beacon for investors and traders in navigating diverse
market conditions. We categorized volatility into multiple
In summary, the Leading RDZ, a pioneering market breadth segments, and within each, we examined three strategic pillars:
indicator, harnesses the principles of Renko charts to examine market conditions, investment factors, and sector trends.
overbought and oversold conditions in the S&P 500 or specific Our tactical roadmap combines the strengths of technical
stock clusters. Seamlessly integrating future volatility and fundamental analysis, while blending the clarity of Renko
expectations, this tool provides a compass for market analysts charts with a nuanced understanding of volatility contexts.
and traders. It empowers them to differentiate between various The multidimensional and adaptive nature of this approach,
market sentiments - bullish, bearish, volatile, or calm - and with its ability to simultaneously consider a spectrum of
interpret the significance of their transitions. factors and adjust to market volatility, provides traders with
a comprehensive decision-making framework. As with any
5.4. Limitations of the Study and Future Research empirical research, our insights are drawn from historical data,
Directions requiring ongoing validation with live market data to ensure
relevance and accuracy.
5.4.1. Limitations Leveraging the principles of Renko charts, we introduced two
Like any empirical research, this study has inherent groundbreaking technical indicators: the Renko Directional
limitations, primarily its reliance on historical data for Z-score (RDZ) and the Leading RDZ. The RDZ quantifies the
backtesting. While past performance can offer valuable strength and significance of trends displayed by Renko charts
insights, it doesn't guarantee future results. Therefore, any using a Z-score approach, promising to enhance various trading
interpretations drawn from past data should be approached aspects from quantitative factor models to risk management
with caution and regularly reassessed against up-to-date and portfolio construction. The Leading RDZ, a dynamic market
market data to stay in sync with market dynamics. breadth indicator, monitors market extremes within S&P 500

PAGE 68 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Cao, J., Wei, J., & Zhong, Z. (2015). Option-implied variance and future
stocks while incorporating future volatility expectations, stock returns. The Journal of Business, 78(5), 1947-1972.
thereby offering an original lens through which to interpret Dawson, E. R., & Steenbarger, B. N. (2021). The psychology of trading:
market sentiment. Tools and techniques for minding the markets. John Wiley & Sons.
Looking forward, we perceive abundant opportunities to Engle, R. F., & Rangel, J. G. (2008). The Spline GARCH Model for Low-
expand this practical research. Applying our methodology Frequency Volatility and Its Global Macroeconomic Causes. Review
to different markets and integrating it with emerging of Financial Studies, 21(3), 1187-1222.
technologies like artificial intelligence and machine learning, Estrella, A., & Mishkin, F. S. (1998). Predicting U.S. Recessions: Financial
holds substantial promise. Future research could also focus Variables as Leading Indicators. Review of Economics and Statistics,
on refining the ATR and RDZ parameters to resonate with the 80(1), 45-61.
rhythm of market cycles. Fama, E. F., & French, K. R. (1993). Common risk factors in the returns
In essence, this applied research sought to recalibrate the on stocks and bonds. Journal of financial economics, 33(1), 3-56.
scope of technical analysis through the creative use of Renko Lin, C., & Sun, Y. (2019). The investment performance of sector rotation
charts. This exciting reinterpretation of traditional techniques strategy on the U.S. stock market. Journal of Empirical Finance, 50,
has the potential to uncover additional strata of market 1-15.
analysis, leading to more precise and insightful discoveries. We Smith, T. (2020). How to create a Renko chart manually. Journal of
hope that this contribution will incite a paradigm shift among Stock & Forex Trading, 9(2).
market technicians, paving the way for future advancements Whaley, R. E. (1993). Derivatives on Market Volatility: Hedging Tools
that blend traditional technical analysis seamlessly with Long Overdue. Journal of Derivatives, 1(1), 71-84.
contemporary quantitative data analytics. Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend
Research.
7. REFERENCES Yang, Y., & Su, X. (2019). An empirical examination of the model-free
Adrian, T., & Wu, H. (2019). The Term Structure of Growth-at-Risk. implied volatility in options markets. Journal of Futures Markets,
International Journal of Central Banking, 15(S1), 163-202. 39(1), 17-44.
Alizadeh, S., Brandt, M. W., & Diebold, F. X. (2002). Range-based Zhang, Y., & Ma, F. (2020). The effect of global oil price shocks on China's
estimation of stochastic volatility models. The Journal of Finance, metal markets. Energy Policy, 137, 111122.
57(3), 1047-1091.
Arnott, R., & Bernstein, P. (2002). What Risk Premium Is “Normal”?
Financial Analysts Journal, 58(2), 64-85.
Asness, C. S., Frazzini, A., & Pedersen, L. H. (2013). Quality minus junk.
Available at SSRN 2312432.
Baker, M., Bradley, B., & Wurgler, J. (2011). Benchmarks as Limits to
Arbitrage: Understanding the Low-Volatility Anomaly. Financial
Analysts Journal, 67(1), 40-54.
Bechu, T., Bertrand, E., & Nebenzahl, J. (2014). L'analyse technique
théorie et méthodes (7ème édition). Economica.
Bernal, O., & Venegas-Martínez, F. (2018). A functional time series
approach to the spectral density estimation of Renko returns
in foreign exchange. North American Journal of Economics and
Finance, 45, 55-65.

IFTA.ORG PAGE 69
 IFTA JOURNAL 2025 EDITION

The Similarities in Various Markets and

Timeframes, Through Quantitative Comparison Nobuaki Kakimoto, CFTe, MFTA
Nippon Technical Analysts Association
Tokyo, Japan
Methods, and Its Application to Trading gi358dd9@bu.iij4u.or.jp

Systems

ABSTRACT different markets and timeframes. Different markets have

While the extent of mutual resemblances in the charts of different currency units and digits, and different timeframes
various markets and timeframes has been widely discussed have different magnitudes of price fluctuations. The methods
by employing mathematics and physics, this thesis presents ought to be simple so that investors can resolve the question of
several simple ways to measure such similarities and discusses whether the methodology and parameters should be changed
their application to trading systems. The observations were for each market and timeframe. Although there seems to be no
made on the past 30-year worth of data of monthly, weekly and settled opinion on this point, two types of ideas exist.
daily on S&P500, Nasdaq Composite, Nikkei225, TOPIX, USD-JPY
and on a “simulated market”, generated by random numbers, in a. Different markets and timeframes have different
which the price fluctuations follow a normal distribution. characteristics, so methods and parameters should be
First, it was learned that the deviation rate from moving changed accordingly.
averages and price fluctuation range irrespectively showed b. If the same method does not work in different markets and
long-term convergence to a constant value or a range, when timeframes, it cannot be considered a robust method.
taking ATR into account and processing them. This allowed to
observe strong similarities and to discover outliers without These are two opposites. An example of the former is that
adjusting parameters per market. Additionally, while the many of the automated trading systems sold are market and
convergence was also observed in the simulated market, whose timeframe specific.*¹ Regarding the latter, Pardo explains it in
appearance resembled the actual one, the converging value terms of "scalability" and recommends multi-market, multi-
apparently differed. period verification using the same system (Pardo 2008, pp.
Second, in respect to the durability of moving averages, the 75-76, pp. 261-262).
relative magnitude of price movements, probability of price This discussion ends in to the question of how much similarity
continuity, and the behavior of major technical indicators, it was there is among them. Some perspectives on how investors can
learned that some of the values were almost identical across practically measure the similarity are proposed on this thesis.
all markets and timeframes, including simulated markets,
suggesting a strong similarity, whereas others showed regular MATERIALS AND METHODS
and tendentious differences. These observations are statistically
processed by distribution analysis based on standard deviation, 2.1 DATA USED
arithmetic mean and frequency distribution. Daily, weekly, and monthly data on S&P500, Nasdaq
For the purpose of applying the research results to trading Composite, Nikkei225, TOPIX, USD-JPY, and WTI Crude Oil for
systems, adopting the probability that is nearly common to each the past 30 years were used. In order to validate the smaller
market to improve their performances and utilizing outliers timeframes, from 1-minute to 60-minutes timeframes of the
against other markets are discussed. In addition, methods for year 2022 for the markets corresponding to the above were also
adjusting the parameters of technical indicators reflecting the used.*²
values that tend to vary by market are also discussed.
2.2 CREATION OF SIMULATED MARKET
ACKNOWLEDGEMENTS In order to verify the similarity of markets, a simulated
I would like to thank the senior members of The Nippon market was created to compare and contrast with actual
Technical Analysts Association :： markets. The simulated market is a so-called random walk
President Y.Higashino, Vice President K.Nakamura, Director market where price fluctuations follow a normal distribution.
T.Miyajima, Chairman of the Board of Trustees K.Kojyo, Member It was created based on an Excel file for generating simulated
of the Board of Trustees N.Yonekura, Secretariat member markets which had been posted on the members-only website
N.Niijima and all the secretariat members, for their useful of The Nippon Technical Analysts Association (NTAA), with
advices and help. modifications (Figure 1).*³

INTRODUCTION
For investors who apply technical analyses to different
markets and timeframes, this thesis intends to devise
quantitative methods to see if the same approach works at

PAGE 70 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 1 Examples of graphing the price actions with

350,000 random prices 2.3 USING TYPICAL PRICE IN CONJUNCTION WITH
CLOSE PRICE
In this thesis, the typical price (hereinafter referred to as
"TP") were used in addition to the close price (hereinafter
referred to as "Close"), as the use of TP often stabilized a
technical indicator or highlighted the original purpose of the
indicator.*⁴

2.4 APPLICATIONS AND PROGRAMMING

LANGUAGES USED
Microsoft Excel was used. Various functions and VBA
programming were used for data aggregation and verification.

2.5 TECHNICAL INDICATORS AND ANALYSIS

METHODS
In each of the following analyses, the simulated market and
the actual market were analyzed using the same method. Also,
the differences between the two, as well as the differences
between the markets, were analyzed.

2.5.1 Generalized Volatility Ratio (GVR) (invented by the

author)
GVR is the absolute value of {(TP of the bar - TP of the
previous bar)/ATR(14)}. Using ATR as a divisor, it aims to
obtain a generalized value that allows direct numerical
comparison between markets in terms of the magnitude of price
fluctuations. The unit is %.

2.5.2 Generalized Deviation ratio from Moving Average(GDMA)

The values corresponding to price fluctuations created by this (invented by the author)
method follow a normal distribution, as shown in Figure 2. The absolute value of {(TP - EMA(20))/ATR(14)}, where ATR is
used as a divisor for the same purpose as GVR.
Figure 2 Standard normal distribution data generated by
NORMINV function (2^20 cases) 2.5.3 Durability of Moving Averages (DoMA) (named by the
author)*⁵
Number of times the MA turned / total number of bars in the
measurement period. Moving averages are calculated using TP
with EMA (5, 20, 50).

2.5.4 Analyzing the Distribution of Representative Technical

Indicators and Price Fluctuations Based on Their Standard
Deviations
The distribution based on the standard deviation at TP and
Close(Bollinger bands), and that of RSI values were compared to
the normal distribution. The calculation periods used were the
common 25 and 14. The distribution of price fluctuations, which
was said to be close to the normal distribution, was also verified.

2.5.5 Analysis of the Number of Consecutive Rises And Falls

The percentage of occurrences for each number of
consecutive rises and falls were measured*⁶. It was also
analyzed whether the data had any meanings or not.

IFTA.ORG PAGE 71
 IFTA JOURNAL 2025 EDITION

RESULTS Table 2 GVR measured in the actual markets at TP

(continued)
3.1 GVR
In the simulated market*⁷, GVR converged to 50% (units
abbreviated below) as measured at TP. However, when the
number of data was as small as monthly or weekly, there was
a variation of less than 3 around 50. The data ranged from 0 to
300, with almost 90% of the data within 100, and the values
measured at TP were generally about 13 lower than those
measured at Close (Table 1).

Table 1 GVR measured in the simulated market

3.2 GDMA
In the simulated market measured at TP, GDMA converged to
130, with 94% within 300 (Table 3). GDMA at Close were almost
the same, but they were slightly larger and the percentage that
settles within 300% was smaller.

Table 3 GDMA measured in the simulated market

In the actual market, stock indices were slightly larger in GVR

than FX and commodity markets, with the former averaging in
the upper 40s and the latter in the lower 40s in daily or above
timeframes*⁸. In less than 60-minutes timeframes, there were
little difference between markets, with values in the lower
40s. The values of 1-minute timeframe were slightly larger
than those of the other minute timeframes. In any case, the
values were smaller than the simulated market with a central
value of 50. About 95% of the values were within 100, a higher
percentage than in the simulated market (Table 2). Although
there were very rare instances of values above 200, the values
were almost always between 0 and 200.

Table 2 GVR measured in the actual markets at TP

PAGE 72 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

In the actual market, stock indices were larger in value than FX and commodity markets, with the former averaging in the mid-130s
and the latter just under 110 in daily or above timeframes. The differences between the markets were small in the minute-timeframes,
but slightly larger in the stock indices, and also larger in the 60-minutes timeframes than in the 15- minutes or under. The probability
of getting within 300 was clearly higher than in the simulated market, averaging 96.2% (Table 4).

Table 4 GDMA measured in the actual markets at TP

Table 4 GDMA measured in the actual markets at TP (continued)

3.3 DoMA
Measurements were taken at 5, 20, and 50 EMAs. When measured at TP, DoMA persisted longer than at Close in all markets and
timeframes without exception. Therefore, only the values measured at TP, which were more persistent, were presented here. First, the
values measured in the simulated markets are shown (Table 5).

Table 5 DoMA measured in the simulated market

In the actual market (Table 6), the overall trend was as

follows: 1) the 1-minute timeframes were the closest to the
simulated market, and the values increased slightly as the
timeframe expanded, 2) with the exception of DoMA(5), the
values were larger than those of the simulated market. The
DoMA(5) were almost the same as the simulated market except
for the U.S. stock indices (weekly and monthly). 3) the U.S.
stock indices (especially the 20 and 50 DoMA above 60 minutes)
showed larger values than the other markets.

IFTA.ORG PAGE 73
 IFTA JOURNAL 2025 EDITION

Table 6 DoMA measured in the actual markets at TP

Table 6 DoMA measured in the actual markets at TP (continued)

3.4 ANALYZING THE DISTRIBUTION OF REPRESENTATIVE TECHNICAL INDICATORS AND PRICE

FLUCTUATIONS BASED ON THEIR STANDARD DEVIATIONS

3.4.1 Bollinger bands

In the simulated market, the distribution measured at TP was slightly thinner in the middle of the distribution and thicker toward
the edges than at Close (Table 7). Note that this is far from a normal distribution and is unrelated to it.

Table 7 Price distribution based on Bollinger

bands measured in the simulated market In the actual market, there were no significant differences
among markets except for some monthly exception
values(Table 8)*⁹. In the timeframe, focusing on values above
2 , the daily values were about the same or slightly smaller
than the simulated market. The weekly values were slightly
larger, and the monthly values were much larger (Nikkei225
and USD-JPY were exceptions). In the 1-minute timeframe,
the values were almost the same as in the simulation, slightly
larger in the 5-minutes timeframe, and maximum in the 15 or
60-minutes timeframe.

PAGE 74 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Table 8 Price distribution based on Bollinger bands measured in the actual markets at TP

3.4.2 RSI
The upper and lower thresholds were set at 90/10, 80/20, and 70/30 to measure how many of the prices settled within the
thresholds. In the simulated market, based on Close, slightly less than 4.6% (2 ) exceeded the 80/20 threshold; on TP, as with
Bollinger bands, the outside of the distribution was thicker (Table 9).

Table 9 Price distribution based on RSI measured

in the simulated market Both TP and Close measurements are included here (Table
10) in order to use them for the parameter adjustment
described later in 4.5.3 and because the differences in values
are somewhat large. In the actual market, there were few
differences among markets, except that the tips of the
distribution were noticeably thicker at weekly and monthly
timeframes of the U.S. stock indices. In the timeframes, the
daily and the above timeframes of the U.S. stock indices were
also striking. Looking around the whole markets, the values
in the 80/20 were much higher than the 7.2% and 3.7% of the
simulated market for the 60-minute and above timeframes
across all markets. The smaller timeframes were not much
different from the simulated market, and the 1-minute
timeframe was almost identical to the simulated market. The
trend of “larger values for larger timeframes” was also similar
to that of the Bollinger bands.

IFTA.ORG PAGE 75
 IFTA JOURNAL 2025 EDITION

Table 10 Price distribution based on RSI measured in the actual markets

3.4.3 Price Fluctuations

In the simulated market, an almost normal distribution was shown. The values between 1 and 2 were slightly thicker than
those of the normal distribution, while those above 2 were slightly thinner. The distribution at Close was slightly closer to a normal
distribution (Table 11).

Table 11 Price fluctuations measured in the

simulated market In the actual market, the distribution was also close to a
normal distribution, but slightly thicker above 2 and below 1
(Table 12). Expressed as a distribution diagram, the central
peak was a little higher and both ends were a little thicker (fat-
tail structure). This phenomenon is a known fact (Tabuchi 2005,
pp 43-44) and here it is clearly evident. There were no notable
differences among markets, but in timeframes, the trend
became more pronounced as the timeframes expanded in the
minute-timeframe. A closer look reveals that the maximum for
above 2 was the 15-minutes, while the maximum for below
1 was the 60-minutes, which was common to all subjects.
There was no significant difference between daily and weekly,
and only the monthly was more pronounced. Generally, values
measured at Close were closer to the normal distribution and
the fat-tail structure was more pronounced at TP.

PAGE 76 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Table 12 Price fluctuations measured in the actual markets

IFTA.ORG PAGE 77
 IFTA JOURNAL 2025 EDITION

3.5 ANALYSIS OF THE NUMBER OF CONSECUTIVE

RISES AND FALLS
Table 13 left shows the results of counting the number of
times in 350,000 simulated markets (Close). The results show
that the probability of rises and falls is almost 1/2, which can be
called a random walk. When the same data is counted at TP, the
result is shown in Table 13 right, showing that the percentage of
"0" time decreased significantly, and the probability of “above 1”
times increased at all counts.

Table 13 Number of consecutive rises and falls and its

percentage measured in the simulated market

Table 14 are the examples of actual markets. Only a few

examples are presented, because the results were virtually
indistinguishable in any market or timeframe, except for the
monthly, for which data were scarce When measured at Close,
the rate of occurrence was almost halved for every 1 count
in any market or timeframe, which was very similar to the
simulated market. However, when observed in detail, there were
some areas where a slight fluctuation in probability could be
detected. The left and middle are with a little visible fluctuation
and the right is with almost no fluctuation. In the DISCUSSION
section, the results are tabulated in a different way to explore
how much the fluctuations affect the trade.

PAGE 78 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Table 14 Number of consecutive rises and falls and its percentage measured in the actual markets

DISCUSSION
4.1 MARKETS TO BE ANALYZED
To begin, the price movements of each market over a 30-year period are shown in a small graph (Figure 3) *¹⁰. The characteristics of
the price movements ensure diversity. This suits the purpose of this thesis, which is to analyze the similarity of the markets. Figure 4
is a graph for each market for the year 2022, using 60-minutes timeframe.

Figure 3 Price movements of each market over 30 years (line graph of daily Close)

IFTA.ORG PAGE 79
IFTA JOURNAL 2025 EDITION

PAGE 80 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 4 Price movements of each market during 2022 (line graph of 60-minutes’Close)

This section presents how similarities and differences can be found in the above completely different markets and in various
timeframes, using the simulated market as a benchmark.

4.2 GVR
ATR, devised by J.W. Wilder Jr. is a technical indicator measuring market volatility. It shows the "average of the true price range”.
For the average, 14 was used following the inventor. Since ATR uses the previous Close, High and Low of the current bar, GVR greater
than 100 indicates that the price fluctuation between the previous bar and the current bar is greater than ATR(14). ATR tends to
have a larger value than GVR because ATR also takes into account the range in the current bar. In the measurement of the simulated
market, the mean value of GVR was 50 at TP and about 90% of all data settled within 100. The significance of using ATR for the
divisor is, first, that it allows markets with different prices and currencies to be directly compared using the same yardstick, making

IFTA.ORG PAGE 81
 IFTA JOURNAL 2025 EDITION

it a powerful tool for measuring similarity. ATR is calculated of value distribution, GDMA that exceeds 300 are approximately
based on the price itself, so it is not directly comparable for 5% (equivalent to almost (4.6%) for a normal distribution)
different markets and timeframes. The same is true for the in most cases in all markets and timeframes, so it is a stable
price fluctuation between the previous bar and the current one. indicator because it provides an approximate probability in
This can be generalized as a magnification by dividing the price advance for judging trends and sudden price excesses.
fluctuation by ATR for direct comparison. The second reason is GDMA in the stock indices, especially in the two U.S. stock
to mitigate the characteristics of ATR, which directly reflects indices, are large in the 60-minute or longer timeframe, equal
the effects of rises and falls in the current bar, as Ehlers defines to or even larger than the 130% (standard value in the simulated
"Noise in trading is the average of the daily trading ranges" market). The larger value for longer timeframes suggests that
(Ehlers 2001, p 98). the trends are more long-term valid. This is deeply related to the
The expectation before calculating was that GVR would not analysis in the next section (DoMA), which will be described in
only vary considerably from market to market and timeframe to 4.4.
timeframe, but also exceed 50, relative to the average value of
the simulated market. Thus, the constancy in the 40s in almost 4.4 DoMA
all markets and timeframes was unexpected (Tables 1 and 2 in Durability of Moving Average measures the degree to which
the previous chapter). price fluctuations are sustained when they take a certain
Next, a detailed discussion about GVR is added. direction. It calculates the average number of bars required to
First, the difference between the measured values at TP and change direction of EMA.
at Close means that the range of change at Close is larger than First, since DoMA of 5EMA for any market shows no difference
at TP. In the simulated market, the mean GVR was 63 at Close from that of the simulated market, it seems that the short-
compared to 50 at TP, which was more than 20% larger. In the term exponential moving average is the same as measuring a
actual market, the difference was almost the same, the values random-walking prices and not measuring what can be called a
changed but the mutual relationship remained the same*¹¹. trend*¹⁵.
In addition, the value of 50 is easy to understand. Thus, the On the other hand, for the 20 and 50 EMAs, DoMA are larger
discussion here is based on the values measured at TP. than for the simulated market as the timeframe expands in all
Second, consider the fact that GVR in the actual market markets: slightly for the 20, but clearly for the 50. This suggests
is smaller than that of simulated market. This indicates that that even artificial market with random walk can look trend-
the actual market has milder price fluctuations (compared to like, but in real "human-involved" markets, the persistence of
the previous bar) than the random-walking market. The GVR the trend increases due to human bias. Furthermore, as with
compares the current price fluctuation to the recent average the GDMA, particularly large values are observed in the U.S.
price fluctuation. It is usually in the 40s and the probability of stock indices at 20 and 50 EMA of the daily or above timeframes.
a value above 100 is less than 10% in any market or timeframe, Those of Japan are also somewhat larger than those of the
both smaller than those of the simulated market. non-stock index markets. Relating this to 4.3, it can be said
Third, there is a fact that not only are the values of above- that trends tend to persist longer in stock indices than in other
daily timeframes larger than those of minute-timeframes, but markets and deviations from moving averages also tend to be
also exceptionally large values of around 50 (comparable to the larger*¹⁶.
simulated market) are observed in the above-daily values of 　　In summary, we can conclude that 1) the larger the
the U.S. and Japanese stock indices. One reason for this is very timeframe and the longer the period over which price direction
likely to derive from the large gaps due to the influence of other is measured, the more meaningful direction can be captured;
markets outside of the session hours. The situation differs from 2) this is especially true for stock indices; and 3) a direction has
that of FX, etc., which is continuously traded for nearly 24 hours. more staying power in any market than in simulated market.
Therefore, the values for the case where nighttime session of Regarding the market similarity, all markets are similar at
stock indices is also included are reported in Endnotes*¹². the minute timeframe level, while at the daily level and above,
The above discussion indicates that the relative magnitudes differences are measured between stock indices and other
of price fluctuations represented by GVR are generally similar markets.
among the various markets and timeframes*¹³. In addition,
there are differences between stock indices and other markets 4.5 ANALYZING THE DISTRIBUTION OF
at the daily and longer timeframes, while other differences are REPRESENTATIVE TECHNICAL INDICATORS AND
small. It is also shown that small differences with certain trends PRICE FLUCTUATIONS BASED ON THEIR STANDARD
can be observed among different timeframes*¹⁴. DEVIATIONS

4.3 GDMA 4.5.1 Bollinger bands and Price Fluctuations

This indicator is calculated by dividing the deviation from a Bollinger bands are invented by John A. Bollinger in
moving average by ATR. It expresses, as a percentage, how many 1980s. At the beginning of this section, the existence of a
magnifications the deviation is from the average price range in misunderstanding about the relationship between Bollinger
the market. As with GVR, ATR is used as a divisor to allow direct bands and the distribution of price fluctuations is described. It
comparison between markets. When graphed, GDMA shows is the range of the price fluctuation that makes the distribution
similar behavior to the moving average deviation ratio. In terms similar to a normal distribution, not the price itself. It is clear

PAGE 82 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

from the results in 3.4.1 and 3.4.3, and has already been pointed be seen even in the 1-minute timeframe, with the above-2
out in 2004 (Kimura 2004 The Nippon Technical Analysis increasing as the timeframe expands (the largest in the
Compendium, pp 169-170). However, the misunderstanding 15-minute timeframe). While the discontinuity between
that "about 95% of prices are expected to settle within the 60-minute and daily timeframe is aforementioned *¹², it
line of the Bollinger bands" is still widely believed*¹⁷. It is hoped should be noted that the 1-minute and daily timeframes show
that this thesis will also help to make people aware of this nearly indistinguishable values and are the closest to a normal
misconception. distribution of all timeframes. In addition, it can be said that the
larger the timeframe, whether it is a minute or a day, the larger
4.5.2 Bollinger bands Analysis the distribution for the tail, and thus the more extreme the price
The appearance of above 2 of the Bollinger bands is still fluctuation likely occurs.
a point of interest. On the minute-timeframes, the rate of Note that while Bollinger bands and RSI have larger above-
appearance increases slightly as the timeframe size expands. 2 percentages when measured at TP than at Close, this is not
However, the maximum difference between the 1-minute and necessarily the case for price fluctuations, especially for stock
60-minutes timeframes is less than 2%, so it is difficult to say indices. Nor did the trends differ between U.S. stock indices and
whether this is a useful difference. The percentage above 2 others, as observed elsewhere.
σ also tends to increase as the timeframe expands for daily The above analyses of the Bollinger bands, RSI, and price
and longer timeframes, and the values of monthly-timeframes fluctuations all show a tendency to be thicker for the tail of
are noticeably different among the markets and countries. The the distribution than the simulated market, and the trend
percentage above 2 increases significantly for the U.S. stock also becomes clearer as the timeframe expands. Furthermore,
indices, crude oil and gold, while it does not increase much or differences are observed between stock indices and the rest
rather decreases for the Japanese stock indices and USD-JPY. of the markets in daily or above timeframes, except for price
This is one aspect of the market's characteristics. It is also fluctuations. In price fluctuations, there is a strong similarity
considered useful in trading to focus on markets where values among all markets, including stock indices. From the above, a
which are far away from the standard value of 13% measured in strong indication about the similarities and differences in the
the simulated market. markets is obtained in this section 4.5, as in previous sections.

4.5.3 RSI 4.6 ANALYSIS OF THE NUMBER OF CONSECUTIVE

RSI was invented by J. Welles Wilder, Jr. in 1978. To observe RISES AND FALLS
the behavior of RSI in any markets, the percentage exceeding In the simulated market, the probability of a rise or fall is
thresholds set at 90/10, 80/20, and 70/30 was examined. The almost strictly 1/2, however, roughly speaking, even in the
thresholds will be easier to adjust if it is known in advance actual market, the probability is closer to 1/2. It was somewhat
what percentage of cases exceed them. The results of this surprising that the probability is almost 1/2 even for stock
measurement can be used for the adjustment. For example, indices with a strong long-term upward trend. However, there
depending on the market or timeframe, It might be a good idea are some cases where the probability deviates slightly from
to determine the threshold at the point where the percentage 1/2, and furthermore, not only the probability but the price
exceeding it will be 2 (about 5%). Since RSI is unimodally range should be taken into account. Therefore, the results of
distributed (Matoba 2004 The Nippon Technical Analysis trend-follow and contrarian trading according to the number
Compendium, pp 254-255), it makes sense to determine the of consecutive rises and falls are presented on table 15 and 16.
threshold by a percentage based on the standard deviation. In The trading are taken price ranges into account and 1.00 is the
this idea, if using the daily Close of the Nikkei 225, 80/20 will be neutral value*¹⁸. The implication of values is the same as profit
acceptable, but for the weekly TP of the S&P 500, 80/20 will be factor, which does not include spreads or commissions. The
too frequent, and a value closer to 90/10 will be better. true ratio of win and lose is calculated by multiplying the price
fluctuations by the number of rises or falls. If it exceeds 1.00, it
4.5.4 Price fluctuations is a win. Considering the frequency of occurrence, no more than
The distribution of price fluctuations measured in 5 consecutive times are included.
the simulated market deviates slightly from the normal In the simulated market, both trend-follow and contrarian
distribution. Since the continuous sequence of each value trading were close to 1.00, indicating that a random walk could
itself in the simulated market follows a normal distribution be represented (Table 15).
exactly, the act of cutting the continuous data by 10 (Close) and
comparing the Close with the previous Close is presumed to be
the cause of this discrepancy. Although the mechanism could
not be elucidated in this thesis, it is interesting to note that
the percentage of above-2 values decreased in the simulated
market relative to the normal distribution, whereas above-2
increased in the actual market. The fat-tail structure can

IFTA.ORG PAGE 83
 IFTA JOURNAL 2025 EDITION

Table 15 Number of consecutive rises and falls and winning percentage in the simulated market

The results for the actual markets are shown in Table 16. Some extreme values were seen in the monthly timeframes. A common
phenomenon with one exception were found in each market where the number of consecutive times was 2 or 3 for the weekly and
daily timeframes. For example, if the same direction appeared for two consecutive weeks, the third week was preferable to follow the
direction (with the exception of S&P 500), and the fourth week was preferable to reverse the direction. This seems to be an anomaly.
Or the "gambler's fallacy" of behavioral finance theory might be at work here. When the same direction appears several times, they
"underestimate the trend" and go against it (Tabuchi, 2005, p. 62). On the other hand, there are no extreme values seen in the minute-
timeframes. As the timeframe gets smaller, they are approaching 1.00, but there might be somewhat distinctive information. The
stock indices have a slight advantage of trend-follow up to three times in the 1-minute timeframe and a slight advantage of contrarian
in the 5-minute timeframe. However, this might be an error.
It is found that all markets are close to 1/2 in terms of mere probability of a rise and fall, but when the price fluctuation is taken into
account, deviations from 1.00 occur, and some of the trends are common to the markets.

Table 16 Number of consecutive rises and falls and winning percentage in the actual markets

PAGE 84 IFTA.ORG
IFTA JOURNAL 2025 EDITION

IFTA.ORG PAGE 85
 IFTA JOURNAL 2025 EDITION

Although Close has used to analyze consecutive rises and falls in this section considering the real trading, other prices, such as High,
Low and TP, are also analyzed at Table 17. Calculated using S&P 500's daily timeframe, the left two tables are almost indistinguishable,
although the S&P 500's 30-year long-term uptrend is clear. From this it can be said that when a high or low is renewed, the probability
that it will be renewed on the next bar is greater than 1/2. The probability of no more than one renewal is about 44%, and the following
each renewal is more than half of the probability of the previous one. This result is even more pronounced when measured in TP.

Table 17 Number of consecutive rises and falls and its probability in S&P500 (Daily High, Low, TP)

Based on the above discussion, the analysis of the number of consecutive rises and falls in Close shows that all markets are similar
including the simulated market, with no significant difference in timeframes. This is the most striking example of similarity, the
theme of this paper. However, when aggregating the data considering the price fluctuation, it can be said that a few characteristics
common to each market, and different from simulated market, can be extracted.

4.7 APPLICATION TO TRADINGYSTEMS

4.7.1 Evaluation Methods

Finally, three samples of trading systems based on the previous results and discussions are presented and evaluated. Along with
profit/loss, number of trading and profit factor, the smoothness of the profit/loss curve and the size of drawdowns are visually verified
*¹⁹ *²⁰.

PAGE 86 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

4.7.2 Application 1: Arbitrage Trading Using GVR 4.7.3 Application 2: Trend-follow and Contrarian Trading using
Utilizing the characteristics of GVR's ability to detect GDMA and RSI
extreme price fluctuations, two entries are made at the timing Using the daily Close of each market, a trend-follow entry
of extreme value on one market and normal value on another in is made where the GDMA is very small and moves along the
two highly correlated markets. Using daily timeframes of S&P direction the price is heading, closed when the GDMA is a
500 and Nasdaq, if one has a GVR above 100 and another has little past its standard value. A contrarian entry is made
below the half, sell the one above 100 and buy the another. After when the GDMA is very large and close when it shrinks to a
the next day, the rule is set to close both trades if the overall predetermined value. The GDMA for entry and exit decisions
profit is gained, and cut the loss if any profit was not gained are adjusted to three types, as shown in Table 19, based on the
by the end of the third day*²¹ (Table 18, Figure 5). Only Close is probability of occurrence of above-300 in Table 4. In addition,
used. The PF and win rate are good and the objective seems to be RSI(14) at Close is used as a filter. In the case of a trend-follow,
met, but the low frequency, long stagnation periods and a large RSI should also be along the direction of price, and in the case of
drawdown must be improved. a contrarian, it should be above 85 for selling and below 15 for
buying. For RSI, each market is in the same condition. Table 20
and Figure 6 show that the average trading days and the number
Table 18 Arbitrage trading using GVR
of trading are close in each market, thus the adjustment of
parameters is generally reasonable.
The overall win rate is probably the strong point of this
system. Although this system does not work in some markets,
there are no very bad ones.*²² However, the long periods
of stagnation, even with good results, were an issue. The
contrarian trading is too selective in terms of entry points,
resulting in very infrequent trading in all markets.

Table 19 GDMA parameter settings (displayed in 1/100)

Figure 5 Arbitrage trading of S&P500 and Nasdaq using

GVR

IFTA.ORG PAGE 87
 IFTA JOURNAL 2025 EDITION

Table 20, Figure 6 Trend-follow and contrarian trading using GDMA and RSI

Table 20, Figure 6 Trend-follow and contrarian trading using GDMA and RSI (continued)

4.7.4 Application 3: Trading Multiple Stock Indices Using Anomalies Appearing in the Number of Consecutive Rises and falls
This is a simple logic to make a contrarian position after three consecutive ups or downs, and close it at the Close of the next day,
using the anomaly found in Section 4.6, which is common to each market. This application is designed as a basket trading style. Four
markets are selected, one each from stock indices, commodities, and FX representing the U.S. and Japan. The lots are multiplied based
on the prices and currencies of each market. Japanese yen was converted to U.S. dollars at the rate of the closing day, and results are
presented in U.S. dollars (Table 21, Figure 7). Including underperforming markets, the advantage of basket is utilized and the profit and
loss trends are passably smooth. The results are in line with the anomaly read from the analysis.

PAGE 88 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Table 21, Figure 7 Trading using anomalies appearing in the number of consecutive rises and falls

4.8 RESEARCH ISSUES TO BE SOLVED

For many indicators in this thesis, 1-minute timeframe and the simulated market were almost indistinguishable. However, the
actual market was moving away from the simulated market with a certain trend as the timeframe expanded. Also, with moving
averages, the 5-EMA was indistinguishable from the simulated market no matter what it was applied to, but not for the 20 or 50
periods. There must be an essential fact lurking in this mechanism that distinguishes the random walk from the actual market. On
this very point, further analyses are needed.

CONCLUSION
5.1 SIMILARITY OF THE MARKETS
For different markets and timeframes, this thesis proposes and analyzes methods to measure across different markets and
timeframes from four perspectives: volatility, moving average durability, deviation and distribution of representative technical
indicators and price fluctuations, and continuous rises and falls. The results show that, with respect to the subject of this thesis,
similarities are strongly suggested if the markets and timeframes are different, and also there are commonalities in the ways in
which the actual market differs from the simulated market. Therefore, in a larger sense, this thesis concludes that the similarity is
quite strong. In the details, in terms of market type, certain differences are observed between stock indices and other markets, and
in terms of timeframes, values slightly increase almost regularly as timeframe expands. Therefore, it can be concluded that there are
certain patterns in the differences of markets and timeframes.

5.2 DIFFERENCES BETWEEN ACTUAL MARKET AND RANDOM WALKING SIMULATED MARKET
A simulated market, where price fluctuations follow a normal distribution, is very similar to actual markets when displayed
on a chart. The differences are examined on each of the perspectives of this thesis and it is found that quantitatively measurable
differences exist in all of them. While there are few differences only in the probability of rises and falls, a difference appears when the
price fluctuation is taken into account. Therefore, this thesis concludes that actual market and simulated market are distinguishable
quantitatively from various perspectives.

5.3 APPLICATION TO TRADING SYSTEMS

It is considered how to apply the indicators used in this thesis to a trading system and this thesis gives three examples. Although all
of them have simple logic, the results obtained seem to be able to reach a practical level by some improving.*²³

IFTA.ORG PAGE 89
 IFTA JOURNAL 2025 EDITION

NORMINV(RAND(),0,1) so that it would be a standard normal

ENDNOTES distribution. No integer conversion was applied.
*¹ A study of the top 30 sales of one month at a large system-trading 4) For simplicity, the initial value is set to 0. Therefore, the value may
software sales site showed that 22 of them specified a single target naturally be negative, but here the value movement was important,
market and/or timeframe in which to operate. (researched by the and the apparent positive or negative value did not affect the
author on August 25, 2023 at https://www.gogojungle.co.jp/). verification.
*² In principle, daily, weekly and monthly data were obtained from Note that even in the process up to 2), the price fluctuations were
Yahoo! Finance (US), which is publicly available. However, for the quite close to a normal distribution.
data listed below, the following publicly available data were used *⁴ As an example, when measured the daily timeframe from January
due to the period of data availability or not being present on Yahoo! 1962 to June 2023 (a period that OHLC are recorded) using DoMA,
Finance (US). the average required for the 20EMA to turn was 9.7 days when
USD/JPY, Crude Oil(WTI), Gold : Investing.com (US) calculated using TP, whereas 8.3 days at Close. The same was always
TOPIX : Yahoo! Finance (JPN) the case with EMAs of other calculation periods and other markets.
For timeframes of 60-minutes or less, 5-minutes, 15-minutes, and Also, the same trend was observed for other technical indicators.
60-minutes timeframes were created by the author's computer Therefore, in many places in this thesis, TP is given priority in
program using 1-minute data for FX and CFD trading published by analyses.
Axiory Global Ltd. (USD/JPY, Gold) or GMO CLICK Securities, Inc. *⁵ Focusing on the turn of moving averages is common, and there is
(others). Although some markets have data prior to 30 years ago likely precedent for similar measurements.
(1993), data for the recent 30 years (from July 1,1993 to June,30 *⁶ The number of consecutive bars is counted as follows. If the
2023) were used due to variations of start dates. previous bar falls compared to the another previous Close, and the
The number of data is shown in the table below. It is determined to current bar rises compared to the previous Close, the current bar is
be statistically sufficient, with the exception of monthly timeframe. set to 0. If the next bar also rises, the bar is set to 1. If the next one
Note that this number of data includes the overhead required for also rises, the count goes up to 2. If it falls, the count goes back to
the calculation. Otherwise, for example, the calculation period 0. As long as the price keeps falling and rising for each bar forward,
would increase by 45 for 50 EMA compared to 5 EMA, and data that the counts will all be 0.
exceeds the predetermined period (July 1993 to June 2023) in the *⁷ The simulated market was measured by randomly selecting an
past direction would be included in the calculation. interval of the required number of bars from the one million bars.
Although the data used are publicly available information, the data *⁸ The number of bars per year varies slightly from market to market,
have been compiled and processed for use in this paper, therefore so the average here is not a strict term.
the responsibility for any errors is attributable to the author. *⁹ The values above 2σ measured at TP were slightly larger than those
measured at Close, which was also the same as the simulated
market.
*¹⁰ Only Gold has data starting in 1999.
*¹¹ In view of the difference of GVR between at TP and at Close, it can
be said that the change of TP is more moderate than that of Close,
whereas the change of Close in the simulated market follows a
normal distribution. The cause is presumably due to an equalization
effect.
*¹² The table below compares the S&P500, Nasdaq100 and Nikkei225
by creating 1440-minutes bars from one-minute bars of 2022.
The 1440-minutes data here is of CFD, which is continuous for
almost 24 hours because the nighttime session is connected to the
intraday trading. On the other hand, the daily-timeframe of stock
indices that GVR shows large values of around 50 contains the
prices of only the stock markets' opening hours. In the 1440-minute
*³ ©NTAA, 2002, now unlisted. timeframe, normal levels of GVR values are shown. This suggests
The specifications of this EXCEL file are as follows. that the gap created during off-session hours would be larger for a
1) First, one arbitrary number is given, and then a random number normal daily bar, since the gap is entered directly when calculating
generated by the RAND function is added to it, converted to an the difference between the previous day's TP or Close and that of the
integer, to create a sequence of 10 numbers. The first and the last of current bar. On a weekly-timeframe, the impact of the gap is once a
these are the Open and Close, and the maximum and minimum are week, but on a daily-timeframe, the impact is more significant since
considered High and Low, creating a bar. it is five times a week. This might be one reason for the phenomenon
2) Add a random number generated by the RAND function to the that there appears to be little continuity between the 60-minutes
previous Close to create the next Open, and repeat the operation and daily timeframe on the table. A similar phenomenon can be
above to create random prices one after another. observed in other indicators.
In this way, the system creates a simulated market with random
price fluctuations. Some improvements made by the author are as
follows;
3) To ensure that price fluctuations explicitly follow a normal
distribution, the NORMINV function was applied to the values
obtained by RAND function. Parameters were set to
*¹³ Some markets which have data of older periods before 1993 show

PAGE 90 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

that GVR have historically declined (white log-approximation curve

in the figure below), but it cannot be said that this is certainly right
due to the paucity of samples. However, it is confirmed that there
has been almost no change in the average of GVR in any of the
markets over the last 30 years covered in this thesis.

*²¹ The daily timeframe was used. The maximum trade period was
set at 3 days, since a large reversal of the arbitrage trading is
*¹⁴ This observation demonstrates that expressions such as, for dangerous. Since the two markets differ greatly in value, the
example, “short-term trading is risky because the five-minute bar Nasdaq/S&P500 multiple was measured daily and the number of
is rougher than the daily bar”, or “the stock market has milder price lots in that ratio was used. However, when closing the position, the
fluctuations than the commodity market”, are only subjective and same lot was used as when new. A multiple of 2.3 means 2.3 S&P500
essentially unconvincing. for every 1.0 Nasdaq. Since the multiplier is to the first decimal
*¹⁵ Here, a trend is a directional price movement accompanied by an place, the actual number would be 23 and 10. The win/loss amount
effect of artificial bias. A directional movement can also be seen in a also takes into account the spread.
simulated market. *²² Because of the huge losses involved in market abruptness, setting
*¹⁶ As shown in Figure 3, the U.S. stock indices show a long-term a loss cut to a generally acceptable value improved performance,
up trend with less noise than the others throughout the 30-year but it was not reflected in the Table 20 because it would be an extra
observation period, but in this figure, periods such as 20EMA of factor in the comparison between markets.
daily or weekly timeframe are too small to observe. However, the *²³ As for the application to trading systems, it has not yet been tested,
results in 4.4 show that even moving averages on weekly or daily but could include the following.
have better durability in stock indices than in the other markets.
･While the DoMA at EMA(50) is in duration (but shorter than the
*¹⁷ In many cases, the explanations of Bollinger Bands on the websites average duration of that market and while the trend is young),
of securities companies and other personal sites in Japan are enter when the price goes backwards and RSI shows a value
written this way. For example, "It is used to predict future prices corresponding to above 2σ.
based on the expectation that prices will move between the +2σ
(standard deviation) and -2σ lines with high probability. Note that, ･Instead of going against the Bollinger Band as soon as it crosses 2σ
statistically, the probability of settling between +2σ and -2σ is line, entering against it when the percentage of above-2σ exceeds
95.45%." the standard value in the simulated market (approx. 13% at TP).
(https://www.smbcnikko.co.jp/terms/japan/ho/J0054.html
REFERENCES
English translation is by the author.)
Aronson, David R. Evidence-Based Technical Analysis : Applying the
However, prices will not settle between those lines 95% of the time.
Scientific Method and Statistical Inference to Trading Signals.
*¹⁸　　This calculation was done at Close because it mimics an actual Translated by Emiko Yamashita, Pan Rolling, 2014.
trading. An example is showed below.
Ehlers, John F. Rocket Science for Traders : Digital Signal Processing
If the number of consecutive times = 2 in ascending; Applications. Translated by Emiko Yamashita, Pan Rolling, 2002.
at the Close of the bar that has a consecutive count of 2, enter the The Nippon Technical Analysts Association, editor. The Nippon
market with a buy (trend-follow) or sell (contrarian), and close the Technical Analysis Compendium. Nikkei Inc., 2004.
position at Close of the next bar. The price fluctuation is recorded
Pardo, Robert. The Evaluation and Optimization of Trading Strategies.
as positive for a win and negative for a loss, and the average range
Translated by Emiko Yamashita, Second Edition, Pan Rolling, 2009.
of positive and negative values for the entire aggregate period is
calculated. The number of wins and losses are counted, then the Tabuchi, Naoya. Random Walk & Behavioral Finance. NIPPON JITSUGYO
win/loss ratio is calculated, and then the average win price range PUBLISHING, 2005.
× win ratio (A) and the average loss price range × loss ratio (B) are
calculated, and finally the absolute value of (A/B) is obtained.
*¹⁹ For a detailed evaluation of the trading system, it should be
aggregated by splitting the entire period. Furthermore, with the
methods as described in Aronson's "Evidence-Based Technical
Analysis" (Aronson 2009), it should be finally undergone rigorous
verifications. Since trading system construction is not the
main subject of this thesis, this section is limited to the part of
the verification process that corresponds to the "preliminary
verification" proposed by Pardo (Pardo 2008, pp 245-270).
*²⁰ Listing below are the spread and the number of lots used in section
4.7.

IFTA.ORG PAGE 91
 IFTA JOURNAL 2025 EDITION

Chicken and Egg: Robert Hanna

Capital Advisors 360
Should You Use the VIX to Time the SPX, or Use Medfield, MA, USA
robh@capitaladvisors360.com
+1 781-956-6952
the SPX to Time the VIX?

Abstract a long-SPX strategy when looking to avoid lengthy periods of

This article presents an exploration into the reciprocal drawdown.
predictive relationships between the S&P 500 Index (SPX) With this in mind, studies are shown that are constructed in
and the CBOE Volatility Index (VIX), with a particular focus on a similar manner to those that examined SPX movements. The
deriving actionable insights for market timing strategies. This paper again utilizes short-term RSI readings and short-term
comprehensive study challenges the prevailing market wisdom high-low readings of both SPX and VIX. This time it examines
by positing that SPX action offers a more reliable basis for their value as indicators for VIX futures movements. Again it is
forecasting VIX movement than VIX action does in forecasting found that SPX readings tend to act as a superior indicator. Data
SPX movement. here is also broken down to examine behavioral differences
The introduction sets the stage by questioning the traditional when SPX is in a long-term uptrend (above its 200ma) versus a
reliance on the VIX as a market timing tool, suggesting instead long-term downtrend (below its 200ma).
that SPX movements may offer valuable insights into future VIX While the data tables show strong evidence that SPX readings
trends. Numerous works are cited sampling publicly available could be used effectively to find favorable opportunities for
research. Most past research has utilized VIX as an indicator for shorting VIX futures, the final section takes it a bit further. To
SPX. The study probes deeper to understand the extent to which show the consistency that the edge has exhibited over time, a
one index can predict the movements of the other. simple model is built using some of the concepts and indicators
The paper then explains the history and construction of the previously examined. The model makes it clear that the research
VIX. It shows how the VIX has moved during all major market is capable of serving as a core component in winning strategies
selloffs since 1990, which is as far back as our VIX data is that could be developed for investors.
available. Final analysis declares that SPX indicators, particularly
The next section is a discussion of implied and historical overbought/oversold measures, possess a markedly higher
volatility (HV). It begins with a dialogue of the “Rule of 16”, and predictive value for VIX movements than previously recognized.
what different VIX readings actually represent. This section The paper challenges conventional wisdom and presents a
lays the groundwork for better understanding the interaction compelling case for the predictive superiority of SPX over VIX
between SPX and VIX, and why the movements of SPX influence in the context of trading both SPX and VIX-based securities.
option traders and their willingness to pay more for portfolio This sets the stage for further research while providing traders
protection. As it turns out, options traders are more likely to be and investment managers tools to realize untapped potential in
reactionary than anticipatory. leveraging SPX action for more informed and effective trading
At this point, the paper examines VIX versus SPX readings decisions in the volatility space.
as short-term market indicators. Two types of studies are done
here. One uses short-term RSI readings of 2, 3, and 4 days. The Introduction
other looks at short-term high and low readings, from 5 to 25 Do bigger chickens produce bigger eggs? Do bigger eggs lead
days in length. Readings are calculated on SPX and also on VIX. to bigger chickens? We might not know whether the chicken or
Data is then examined which shows the SPX readings tend to be the egg came first, but we can probably learn the answer to the
more useful in anticipating SPX movement than VIX readings above. When looking at the market, we know that the S&P 500
are. The data is sliced several ways, and the answer consistently (SPX) came before the CBOE Volatility Index (VIX). But which
suggests SPX readings offer more robust edges for trading. does a better job of predicting the other? And more importantly,
There is then a discussion of VIX-based securities and why where do the biggest trading edges lie?
traders and investment managers may want to consider them There has been ample research published over the years
in a portfolio. One of the primary reasons demonstrated is that that examined CBOE Volatility Index (VIX) action, and how VIX
they have shown an incredible downside persistency. While readings might be used to predict S&P 500 movement. In their
counter moves can be sharp during market crises, VIX-based 2004 book, “How Markets Really Work”, Larry Connors and
securities (futures and exchange-traded products) have Conor Sen showed how short-term overbought and oversold VIX
typically gone substantially shorter periods between making readings were a useful indicator for short-term SPX movement.
new all-time lows than SPX has in making new all-time highs. Connors has since published additional research along these
The Great Financial Crisis of 2007-2008 along with the 2022 lines.
bear market are discussed in detail. Analysis implies that a In 2018, Fassas and Hourvouliades published “VIX Futures As
well-constructed short-VIX strategy should have an edge over A Market Timing Indicator” which examined VIX futures term

PAGE 92 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Figure 1
structure for anticipating forward S&P 500 returns.
In his 2020 NAAIM Founders Award paper, “Actively Using
Passive Sectors to Generate Alpha Using the VIX” Michael Gayed
used the VIX in timing allocation adjustments between sectors.
Studies have also been published that examine short volatility
strategies, such as “VIX Index Strategies, Shorting volatility
as a portfolio enhancing strategy”, by Dondoni, Montagna, and
Maggi (2018). There they identified the long-term tendency of
VIX futures to decline and used the VIX futures term structure
to aid in timing short volatility positions.
A primary use of VIX among researchers and market
participants has been as an indicator for timing the SPX and
other markets. Years of trading experience and research has
taught me that more value may be found in doing the converse.
Therefore, I will examine approaches that use SPX action as
an indicator for trading VIX-based securities, primarily VIX
futures. Before getting to that, I will discuss the VIX itself, what
its readings represent, and historical tendencies of the VIX. I’ll
then examine whether VIX readings are any more predictive VIX spikes occurred during all of the tumultuous S&P 500
of SPX movement than simply using SPX readings. I’ll examine drawdowns, including the 1998 Long-Term Capital blowup, the
VIX futures (VX) movement, the use of SPX indicators as a VX 2000-2003 tech bear, the 2008 Great Financial Crisis, the 2020
trading filter, and finally, a simple model that incorporates some COVID Crash, and others. But as you can see, the VIX tended
of the research. to revert to more normal levels quite quickly, even when the
S&P 500 took an extended amount of time to recover its losses.
VIX Overview Quicker recovery time is a key concept. And it not only applies
The VIX was created by the Chicago Board of Options to the VIX, but as I will show later, it also applies to tradeable
Exchange (CBOE). It measures 30-day S&P 500 volatility VIX-based products.
expectations based on SPX option prices. To get the 30-day I will also share another observation here. The 2022 high
average, it considers option prices in the 23 to 37-day range. level is shown by the red line. Along with the December 2018
The VIX will generally deliver “high” readings in volatile and bear, this is about as low a reading as we have seen during a
uncertain conditions. “Low” readings are often found during bear market. That surprised some people, but it shouldn’t have.
quiet market environments. From 1990 – 2023 the range of There was nothing particularly scary about the 2022 bear
closing VIX prices was between 9.14 and 82.69. market. It was primarily caused by a rise in interest rates. Rising
The VIX is not tradeable, but there are tradeable instruments interest rates are not nearly as scary as a worldwide financial
based on the VIX. VIX futures were the first instrument to be crisis or a 100-year pandemic. Further, realized volatility never
based on the VIX index. They began trading in 2004, though spiked in 2022 like we saw in other periods. So with actual
volume was quite muted in the early years. In 2006 VIX options price movement not extreme and the "big scare" being rising
were introduced. Then in 2009, we saw the first VIX-based interest rates, why would people expect the VIX to reach the
ETFs arrive. Interest in using VIX-based instruments has kind of levels that were seen in some other bear markets? They
grown substantially over the years. For some they are used shouldn’t.
as a hedging instrument, and for others they are traded as a
separate strategy. Implied & Historical Volatility
The reason VIX products became a popular hedge is that the I think part of the problem with people misunderstanding
VIX will often trade inversely to the SPX. The basic reason for where the VIX “should” be is that they do not understand
this is that falling markets will generate fear. The desire for implied volatility, and they especially don’t understand what the
protection increases and traders are generally more willing to VIX readings represent. Implied volatility is an estimate of daily
pay a larger premium for options to protect their portfolios. variations in price. The VIX measures options that are priced on
When SPX is on the rise, investors feel good. There is less the S&P 500. It is the implied volatility of those SPX options that
demand for portfolio protection. People are not willing to pay as average 30 days to maturity. But it is not a "daily" number. It is
much for options in a rising market as in a falling market. So the “annualized”. This leads us to the “Rule of 16”.
reduced options premium is reflected in the lower VIX reading. To convert an annual implied volatility number into a daily
Below is a long-term chart of daily VIX closing prices. Though number you need to take the square root of the number of days.
SPX is not shown on the chart, yellow highlighted sections are There are approximately 252 trading days in a year. The square
times where the SPX experienced a 20% drawdown or more. root of 252 is 15.875 (about 16). So a VIX reading of 16 represents
daily implied volatility of about 1% over the next 30 days. (More
specifically, it suggests that 68.2% of days will see price changes
of less than 1%, and 31.8% of the time the SPX could change by

IFTA.ORG PAGE 93
 IFTA JOURNAL 2025 EDITION

more than 1%.) A VIX of 32 implies 2% moves are likely, 48 would So VIX is a measure of implied volatility for S&P 500 options.
anticipate numerous 3% moves, and for the VIX to be at 80, It is not related to actual, historical volatility, but the two are
implied volatility on SPX options would be pricing in 5% daily highly correlated.
moves. I mentioned the VIX often trades inversely to the SPX. Let’s
If you have observed the S&P 500 for very long, you will look at a chart so you can see what I mean by this.
realize that price changes of 5%, 4%, 3%, or even 2% in one day
are quite rare. Going through an extended period of time where Figure 3
daily moves are averaging even 2% is highly unusual. So it is
understandable that a VIX of even 30 or 40 is not common.
When price action in the S&P 500 becomes more volatile, the
VIX therefore naturally spikes. And when S&P 500 volatility
begins to quiet down, it is unlikely that the VIX will remain
elevated.
When considering what volatility expectations might be over
the next month, the easiest way to estimate this would be to
look at what volatility has been over the last month. One month
forward is rounded to 30 days for the VIX. One month backward
can be rounded to 21 (trading) days for historical volatility (HV).
The chart below shows VIX readings in red and 21-day HV of SPX
in blue, going back to the inception of the VIX in 1990. This chart is from 2022. You can see here how moves up in
SPX generally coincide with moves down in the VIX, and vice
Figure 2
versa. Because VIX direction is often inverted versus SPX, the
tendency over the years has been to use VIX as an indicator for
SPX. Let’s take a quantitative look to evaluate whether VIX is
likely a useful indicator for SPX movement. I will look at it from
both a long-term and a short-term perspective.

VIX vs SPX Readings as a Long-Term

Market Filter
I do not recall reading any research in the past that used
VIX as an effective long-term indicator. Of course SPX long-
term trend measures and long-term moving averages can be
used to potentially sidestep portions of bear markets, reduce
drawdown, and in many cases increase long-term returns.
So I took 3 simple S&P 500 long-term trend filters to see how
A few notes: utilizing them might compare to a Buy & Hold approach with the
S&P 500. The 3 trend filters I chose were:
• Perhaps the most apparent observation I could make about
the chart is that the lines tend to move together. They rise and 1) Whether the SPX closed higher than it closed 1 year ago (252
fall at about the same time, never getting too far off track. trading days)
• The red line (VIX) is typically above the blue (HV). So 2) Whether the SPX closed above its 200-day moving average.
estimations of implied volatility are generally a little higher 3) Whether the 50-day moving average closed above the 200-day
than what has been seen from a volatility standpoint in the moving average (also known as the Golden Cross formation).
recent past. This makes sense. The next 30 days' action is
unknown, and options sellers will typically be able to demand The test assumed that when the trend filter suggested an
a premium for taking on unknown risks. uptrend that SPX was entered. When SPX was not in an uptrend,
• There are a couple of instances marked on the chart that HV then the portfolio would go to cash, and earn interest at the
reached similar levels to the 2022 highs, but VIX spiked quite overnight Fed Funds rate.
a bit higher. They were in 2010 and 2015. Important to note Lastly, I took the same indicators and applied them to the VIX.
about both these instances is that they were quick market Since the VIX typically mirrors the SPX, I reversed them. So the
shocks. HV transitioned from a low level to up over 30 very portfolio would go long if:
quickly both times. In other words, there was a brief but
powerful market shock that VIX quickly adjusted to, and this 1) The VIX closed below where it did 1 year ago.
caused it to elevate more substantially above HV. 2) The VIX closed below its 200-day moving average.
• The 2022 high levels in HV were achieved through a more 3) The 50-day moving average of the VIX closed below the 200-
gradual rise. Options traders had more time to adjust to day moving average.
increased volatility, and so VIX remained more in line with HV
as compared to the 2010 and 2015 instances.

PAGE 94 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Results using these six filters, along with “Buy & Hold” returns can be found in the table below.

Figure 4

We see here that using SPX trend filters was generally successful. Two of the three SPX filters outperformed Buy & Hold total
returns. They all reduced drawdowns. And the Compound Annual Return / Max Drawdown (CAR/MDD) stats were improved using all 3
filters.
But the long-term VIX filters did not seem to help anything. They all showed lower net profits, and the CAR/MDD was worse than
Buy & Hold in every case. It appears long-term measures of the VIX are not helpful as a filter for SPX.

VIX vs SPX Readings as a Short-Term Market Indicator

Most publicly available VIX-related research has focused on using VIX measures to anticipate SPX action. However, my research has
not found VIX action to be any more predictive of SPX movement than simply using SPX action.
I’ll use two different measures of overbought/oversold to demonstrate. The first indicator to consider is Wilder’s RSI. Short-term
measures of RSI are used here for a couple of reasons: 1) I have found that shorter-term RSIs tend to indicate short-term action better
than longer-term RSIs. 2) Short-term RSIs tend to offer more evenly distributed delineation. For instance, a 14-period RSI will see
most readings fall between 30 and 70. But a 2-period RSI will have plenty of instances below 30 and above 70. This makes for neater
delineation.
The first series of tests utilize RSI(2). All of these tests are run from 2007 – 2023. This was done to include the Great Financial Crisis,
and also to measure a period where both VIX futures and options were available. 2007 is the first full year where that was the case.
This first results table shows next-day returns in SPX when its RSI(2) closed in 5 different quintiles. It also incorporates a long-term
filter (SPX 200-day moving average) for the bottom 10 rows.

Figure 5

IFTA.ORG PAGE 95
 IFTA JOURNAL 2025 EDITION

Results here show some interesting delineation, and are suggestive of an edge that would be worthwhile to consider. Low RSIs
(especially below 20) provide the strongest numbers.
Such oversold conditions led to net positive next-day returns. Meanwhile, overbought conditions (RSI > 80) showed negative
returns. This was especially true during a long-term downtrends (bottom row). During long-term uptrends (above the 200ma), RSI>80
results were actually slightly positive. Also notable: 1) Some people may not be familiar with “Profit Factor”. It is Gross Gains / Gross
Losses. So a profit factor above 1 means that the strategy is profitable. A reading below 1 means the strategy has lost money over time.
Profit factor can be a helpful measure of reward/risk when evaluating strategies. 2) Results when SPX is in a downtrend tend to be
more extreme – both bullish and bearish. It may seem odd that average gains are much higher during “oversold in downtrends” than
they are during “oversold in uptrends”, but volatility tends to be higher during downtrends. When volatility is higher it makes moves
more sizable. Hence the more extreme numbers during downtrends. This will be evident in results throughout the paper. 3) The theme
of oversold providing an upside edge and overbought suggesting a neutral or bearish market condition is something you will continue
to see throughout the research as well.
So now let’s use the same RSI(2) indicator and apply it to the VIX to see if that might provide an edge for trading SPX.

Figure 6

Some of the same themes play out here, but not as impressively. An overbought VIX (RSI > 80, 0 Long-Term Filter, 6th row down)
suggests the SPX may be primed to bounce. But an oversold VIX (RSI <= 20, 2nd row down) shows numbers that are very close to the
numbers shown when there is no filter at all (top row). So while an overbought VIX could provide an edge, an oversold VIX does not
appear to generate a quantifiable edge.
Let’s next look at RSI(3) for SPX and VIX, in order to assess whether the themes play out here as well.

Figure 7

PAGE 96 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Once again, an RSI of SPX under 20 suggests a short-term bullish edge, and an RSI over 80 suggests a neutral (during uptrends) or
bearish (during downtrends) edge.
Next let’s apply RSI(3)of VIX as a delineator.

Figure 8

Here again, an overbought VIX suggests an upside edge for SPX, but an oversold VIX does not appear greatly predictive. An oversold
VIX when SPX is in a downtrend does show slightly bearish results, but not nearly to the extent that an overbought RSI(3) of SPX does.
Finally…RSI(4). First, SPX:

Figure 9

Again low RSI readings suggest an upside edge and high RSI readings suggest a next-day downside edge. Results here are a bit more
extreme. Part of the reason for this is that it takes larger moves to get extreme readings with a 4-period RSI than a 2-period RSI. That
is why you also see the lower number of instances where RSI was below 20 or above 80.
And VIX…

IFTA.ORG PAGE 97
 IFTA JOURNAL 2025 EDITION

Figure 10

No surprise here. An overbought VIX suggests an upside edge, but not as strongly as an oversold SPX. Interestingly, the number of
instances where VIX registered an RSI(4) reading greater than 80 was the same number of instances that SPX registered an RSI(4) less
than 20. It was 285 instances. RSI of SPX showed a higher % Winners (60% vs 55%), larger average move (0.31% vs 0.23%), and a better
profit factor (1.63 vs 1.51). Using RSI to measure overbought/oversold VIX, could identify opportune times to take short-term trades to
the long side. But simply using the same measures on SPX appears to provide a more reliable edge.
Of course RSI is just one measure of overbought/oversold. Next let’s look at an even simpler one. Here we are going to examine
whether SPX (and later VIX) closed at a multi-day high, a multi-day low, or somewhere in between. I’ll also delineate “in between” into
two buckets using a moving average. So the 4 quadrants are 1) at an X-day low, 2) below the X-day moving average (but above an X-day
low), 3) above the X-day moving average (but below an X-day high), and 4) at an X-day high.
The table below uses 10 for X.

Figure 11

After seeing the RSI results, these results are no surprise. Oversold suggests an upside edge, and overbought a mild downside edge.
Breaking it out by the 200ma also tells a similar story.

Figure 12

PAGE 98 IFTA.ORG
 IFTA JOURNAL 2025 EDITION

Closing at a low suggests an upside edge in either case. But a low close below the 200ma shows greater potential gains, while a low
close above the 200ma shows more reliable gains. Again, this is similar to what we saw using RSI.
Next is a similar breakdown, but using VIX readings as the indicator, rather than SPX readings.

Figure 13

Here we see a high VIX suggests an upside edge for SPX, while a low VIX does not appear to suggest any edge. Comparing the 10-day
low SPX close to the 10-day high VIX close you’ll note that the number of instances is similar, but the low-SPX setup shows a higher win
rate and a higher Avg % Profit/Loss. Here is the 200ma breakout like we just did with SPX:

Figure 14

Again below the 200ma shows more extreme readings. I will note that a “lowest” VIX reading when SPX is below the 200ma does
show a negative forward return for SPX, but not nearly as negative as a “highest” SPX reading showed (-0.03% vs -0.28% average loss).
Lest there is concern that the 10-day period was cherry-picked, below is performance for “Lowest” and “Highest” closes for 5, 10, 15,
20, and 25 days. First for SPX, then for VIX.

Figure 15 & 16

IFTA.ORG PAGE 99
 IFTA JOURNAL 2025 EDITION

The lessons are the same no matter the timeframe. 1) Oversold SPX and overbought VIX both suggest an upside edge. The edge
provided by SPX appears to be stronger. 2) Overbought SPX suggests a downside edge, while oversold VIX is more neutral.
You may find it curious that the downside edge for overbought SPX does not appear to increase when looking at more extreme
readings (a 25-day high vs a 5-day high). This is because the 20 and 25-day high readings are so much less common below the 200ma
where the edge primarily exists. You can see this in the table below that contains the 200ma breakdown.

Figure 17

Note the downside edge continues to increase as you make higher-level highs below the 200ma (bottom 5 rows). But the instances
shrink enough that the total impact is dulled when lumped together with instances above the 200ma.
Whether we use short-term RSI readings or whether we simply look at short to intermediate-term highs and lows to gauge
overbought/oversold, the message appears to be the same.

• Oversold SPX readings suggest a next-day upside edge.

• The edge is more reliable above the 200ma, but more powerful below it.
• An overbought VIX provides a similar edge, but the overbought VIX stats are not as impressive.
• Overbought SPX readings suggest a downside edge, particularly when SPX is in a long-term downtrend (below its 200ma).
• Oversold VIX readings don’t show much of a downside edge. If it appears at all, it is much less pronounced than the downward edge
an overbought SPX suggests.

In general, the VIX could be used to design short-term models to trade the SPX. But it is not clear that using VIX readings instead of
SPX readings would enhance model returns. In fact, the more robust edges seem to simply be based on SPX movement itself.

Considering VIX-based Securities for Trading

As discussed earlier, VIX readings appear highly correlated to, and perhaps dependent on SPX action. High VIX readings are unlikely
without high realized volatility in SPX. And low VIX readings are extremely unlikely when SPX is highly volatile. So rather than use VIX
movement to identify edges for trading SPX, would there be more value in the converse? Could SPX movement be used to trade VIX-
based securities?
First, let’s recall that the VIX itself is not tradeable. But there are VIX futures, options, and exchange-traded products available for
trading. So let’s examine how some of these products have traded over time.
Below is a split-adjusted look at UVXY, which is the VIX-based ETF that has had the longest continuous history.

PAGE 100 IFTA.ORG

 IFTA JOURNAL 2025 EDITION

Figure 18

With the value dropping from over $24 billion in 2011 down to $8.44 at the end of 2023 there is a clear long-term downside edge. Of
course, some very sharp countertrend rallies happened during big market selloffs. Still, a move from $24 million down to $8 seems like
one that could be taken advantage of. And what is especially encouraging is that there was never a long, extended period between new
lows.
Notable about UVXY is that the leverage changed from 2x to 1.5x on Feb 28, 2018, after "Volmageddon” in early February wreaked
havoc on several VIX-based ETFs. UVXY went from a low of 4260 to a high of 15,090 (354% gain) during that period. During March
of 2020 UVXY rose as much as 1300%. I am not going to get into position sizing and risk management in this paper, but it is clearly
something that investment managers need to take into account when trading VIX-based products. Still, the long-term edge appears
sizable.
Trading the ETFs and ETNs also includes counter-party risk, borrow costs, available shares for shorting, margin requirements, and
ETF structure issues. These all create complications that I am not inclined to delve into in this paper, but have discussed elsewhere.
Most of these issues do not exist when looking at VIX futures (VX). Additionally, VIX futures have a longer history. And as shown
below, it is not just the ETFs and ETNs, but also the futures, that have shown a long-term downside tendency.

Figure 19

This continuous futures contract was created with data available directly from the CBOE website. VIX futures expire on Wednesday
mornings, so the roll from one month to the next was done at the close on Tuesday just before the Wednesday expiration. The price of
the future declined from a roll-adjusted 229.9 on October 31, 2006 to 13.05 at the end of 2023. While there were sharp counter-trend
moves along the way, the persistency was impressive. The largest and longest countertrend move occurred during 2008. Next we will
zoom in on the 2007-2008 bear market as well as the 2022 bear market.

IFTA.ORG PAGE 101

 IFTA JOURNAL 2025 EDITION

This chart shows the S&P 500 during the 2007 to 2009 bear market.

Figure 20

As you can see the S&P topped in October of 2007. The bear market bottomed in 2009, but it wasn't until May 2013 that SPX managed
to make a new high. So a buy-and-hold S&P 500 strategy would have spent about 5 1/2 years in a drawdown during this bear market.
But let's look and see what happened to VIX futures during this same period.

Figure 21

We see here that the bottom for the VIX futures did not come until May of 2008, a full 7 months after the initial S&P market top.
There was a big spike during the bear market, but the recovery was much quicker. VIX futures made a new low in February of 2010.
So rather than a 5 ½ year drawdown, a VX “short and hold” strategy would have only endured about a 1 ½ year drawdown. That's a
massive difference, and it makes the idea of trading VIX-based products very appealing. When market shocks occur, you don't need a
large SPX rally for a short-volatility strategy to make a new high. You just need the market to calm down enough to allow the downside
tendency of VIX futures to reassert itself. Next let's look at the 2022 bear market.

PAGE 102 IFTA.ORG

 IFTA JOURNAL 2025 EDITION

Figure 22

The top panel is the S&P 500 from the beginning of 2022 through 2023. While SPX bottomed in October of 2022, it took until early
2024 to get back to a new high. It was a 2-year drawdown. The bottom panel shows VIX futures. VX barely went six months without
making a new low. By the end of July 2022 it was already at new lows, and SPX had not even bottomed yet! When the SPX finally
bottomed in October 2022, VX was just weeks away from another round of new lows. This persistent downside tendency creates a
massive edge, and the shortened drawdowns make the thought of trading VIX-based products quite appealing vs trading the SPX.

Filtering VX Movement Using SPX/VIX Action

I established earlier that SPX is a better indicator than VIX for anticipating SPX movement. But what about VX movement? Next is a
series of studies that utilize SPX readings and VIX readings and measure VX movement based on the same filters we used in evaluating
SPX edges. It is important to keep in mind that since the long-term VX trend is down, these studies look at results of going SHORT VX.
So a positive result means that VX has exhibited a downside edge under these circumstances. A negative result would indicate times
the VX has risen.

Figure 23

These results are intriguing. We see that short-term oversold readings show favorable results and short-term overbought readings
are unfavorable for the “short VX” trade. Without any filters, for the full period measured, VX declined 213.52 points (top row). By
simply shorting VX when SPX posted an RSI(2) <= 40, and sitting out the rest of the time, more than all of the 213.52 downside points
would have been realized. And that would have been accomplished by only being in the market 34% of the time! Also impressive is
the fact that RSIs over 80 (bottom row) showed a 45.57 point gain in VX over time. This suggests that periods when SPX is short-term
overbought are dangerous times to short VX, and they could even provide opportunities for taking long VX positions.
What if we use the RSI of VIX instead of SPX?

IFTA.ORG PAGE 103

 IFTA JOURNAL 2025 EDITION

Figure 24

As with the SPX tests we ran earlier, the edges here are not as consistent, nor as strong. An overbought VIX reading ( > 80) suggests
a favorable condition for “short VX”. But the Avg Profit/Loss and Net Profits are about half of what is seen in the SPX RSI <= 20 results
from the previous table. Also notable is that while an overbought SPX reading suggests a long VX edge, an oversold VIX reading does
NOT. Using RSI of VIX as a filter does not appear nearly as valuable as using RSI of SPX.
I also broke out the returns by whether SPX closed above or below its 200ma. Here is the RSI of SPX breakdown.

Figure 25

No surprises. Oversold in an uptrend is more consistent than oversold in a downtrend, but the instances below the 200ma show
larger average moves.
Also notable is that the “upside VX” edge almost entirely plays out when SPX is in a downtrend (bottom row). Overbought in an
uptrend stats only show small losses when shorting VX.
Next is the 200ma filter breakdown using RSI of VIX:

Figure 26

The only strong edge here appears to be overbought VIX readings when SPX is in a long-term downtrend (blue outlined row). In
looking at high VIX readings when SPX is above the 200ma (top row), the Avg Profit/Loss is the same as without any RSI filter above
the 200ma. And the Drawdown and Profit Factor stats are worse.
This all suggests that RSI of SPX is a much better indicator of forward VX performance than RSI of VIX.
This last RSI table shows returns using 2, 3, and 4-period RSIs of SPX.

PAGE 104 IFTA.ORG

 IFTA JOURNAL 2025 EDITION

Figure 27

The stats change some depending on the RSI length you use, but the implications remain the same. When SPX is short-term oversold,
it is a favorable condition for shorting VX. And when SPX is overbought, VX shorting becomes dangerous, and long VX positions could
even be considered.
Next let’s look at times SPX closed at a high, a low, or in between. This series of tests is set up similar to the series shown earlier that
examined SPX performance based on where SPX/VIX closed in relation to a new high, low, or in between. The first table below looks at
SPX finishing hi-low-mid using a 10-day measure.

Figure 28

Consistent with our other tests, a low SPX suggests a favorable time for shorting VX. A high SPX suggests an unfavorable time for
shorting VX. Readings in between “highest” and “lowest” are not far from average returns. With middle quadrants not showing much
edge (anywhere), the next several tests will focus just on “highest” and “lowest” readings.
The two tables below look at “short VX” performance broken down by “highest” and “lowest” close in the last “X Days”. The first
table uses VIX highs and lows, and the 2nd table uses SPX highs and lows.

Figure 29 & 30

IFTA.ORG PAGE 105

 IFTA JOURNAL 2025 EDITION

The delineation in the bottom table is clearly better. It is all red near the top and all green near the bottom of the table. High SPX
closes are a bad time to short VX (see the negative Avg Profit/Loss in the top 5 rows of the bottom table). Low SPX closes create
opportune times for shorting VX. This can be seen by the strong Avg Profit/Loss numbers and other stats in the bottom 5 rows of the
2nd table.
Delineation in the top table was not nearly as impressive. High VIX readings showed a moderate edge for shorting the VIX. Low VIX
readings did not provide a discernable edge. We keep seeing the same theme repeated.
Lastly, I broke down the SPX numbers to show how they looked above vs below the 200ma. This can be found in the table below.

Figure 31

Like with RSI, 1) strong closes above the 200ma are basically neutral. 2) strong closes below the 200ma show a strong “upside VX”
edge, 3) low closes above the 200ma show a high % Winners and muted drawdowns, while 4) low closes below the 200ma show the best
Avg Profit/Loss statistics.

A View of a Simple Model

It is clear that there is ample opportunity in shorting VX (and other VIX-based products). The return breakdowns show that SPX
action can be used to identify opportune times to take short (and long) VX positions. But while the data tables show net numbers and
demonstrate the overall edges quite nicely, the consistency of the edges cannot be seen by simply looking at a data table. For that, a
profit curve is helpful. I created a sample model based on the RSI results that were examined earlier. The rules are simple:

1) If the RSI(2) of SPX closes <= 30, short VX on close and hold the next day.
2) If the RSI(2) of SPX closes > 90, buy VX on close and hold the next day.
3) If neither of the above is true, then the position is flat.

Figure 32
The curve is not perfect, but it is certainly impressive. This
was accomplished with just 1 measure (RSI of SPX). There was
no filtering for uptrend/downtrend, even though our data
suggested that could be used as well. The 225 points of profit are
greater than the 213 points that VX lost over the period tested.
And these results were accomplished despite only carrying
VX exposure 45% of the time. The rest of the time the position
would have been flat. Also notable is that the Max Drawdown
in the model is 28.06 points. That is much lower than the 68.86
point drawdown for a “short and hold” approach. In fact, it
represents a drawdown reduction of 59.25%!
None of this represents a complete system. But it is a
solid start, and should help traders and managers consider
implications of SPX movement on VIX-based instruments. Of
course there are other factors that traders and investment
managers will want to consider. These may include, but are
not limited to, the VIX futures term structure, upcoming
known event risks (elections, economic releases, etc.), and

PAGE 106 IFTA.ORG

 IFTA JOURNAL 2025 EDITION

potential seasonality influences, as well as trading costs and tax provides.

consequences. Further, position sizing and risk management
will need to be taken into account. But it is clear the SPX action References
can provide a quantifiable edge for trading VX. And it provides Connors, Larry & Sen, Conor How Markets Really Work. USA, Connors
even better information than VIX action itself does for trading Research Group, 2004
VX. Utilizing this concept should help traders and investment Fassas, Athanasios and Hourvouliades, Nikolaos L., VIX Futures
managers to develop winning strategies. As a Market Timing Indicator (June 3, 2018). Available at SSRN:
https://ssrn.com/abstract=3189502 or http://dx.doi.org/10.2139/
Conclusion ssrn.3189502
Lots of energy has been used over the years trying to Gayed, Michael, Actively Using Passive Sectors to Generate Alpha
understand what edges the VIX might provide in predicting Using the VIX (November 30, 2019). 2020 NAAIM Founders Award
SPX movement. While VIX readings can be used to identify & 2020 Dow Award Finalist, Available at SSRN: https://ssrn.com/
compelling setups in many cases, it appears that overbought/ abstract=3718824 or http://dx.doi.org/10.2139/ssrn.3718824
oversold measures of the VIX are less predictive than using Dondoni, Alberto and Montagna, Dennis Marco and Maggi, Mario,
similar measures of SPX. Additionally, while most traders want VIX Index Strategies: Shorting Volatility As a Portfolio Enhancing
Strategy (January 18, 2018). Available at SSRN: https://ssrn.com/
to know where the SPX is headed, the easier question is “Where
abstract=3104407 or http://dx.doi.org/10.2139/ssrn.3104407
is VX headed?”. The downside persistency of VIX-based products
R. H. Morris, D. F. Hessels & R. J. Bishop (1968) The relationship
make them highly attractive for trading purposes. When a
between hatching egg weight and subsequent performance
bear market arrives, drawdown lengths are often substantially of broiler chickens, British Poultry Science, 9:4, 305-315, DOI:
briefer for a short-VIX position than a long-SPX position. In 10.1080/00071666808415726
examining VX behavior, I again find that overbought/oversold
D. Narahari, K. AbdulMujeer, A. Thangavel, N. Ramamurthy, S.
measures of SPX provide better filtering than measures of VIX. Viswanathan, B. Mohan, B. Muruganandan & V. Sundararasu
I believe most traders have been getting it wrong for years. (1988) Traits influencing the hatching performance of
Rather than using the VIX to predict SPX movement, more Japanese quail eggs, British Poultry Science, 29:1, 101-112, DOI:
substantial edges lie in using SPX to predict movement of VIX- 10.1080/00071668808417031
based products.
According to Morris, Hessels, and Bishops 1968 paper The
relationship between hatching egg weight and subsequent
performance of broiler chickens, “Body weight to 12 weeks was
found to be strongly related to egg weight, in a linear fashion,
though this influence declined with age.” So yeah…bigger eggs
get you bigger chickens.
In their paper Traits influencing the hatching performance of
Japanese quail eggs, Narahari, et al. state “Japanese quail eggs
…from moderately heavier dams hatched slightly better than
the eggs from lighter dams.” They also noted that “Fertility and
hatchability were directly proportional to the egg weight.” So
it appears bigger quails are going to give you a better chance at
bigger quail eggs.
If bigger quails make bigger quail eggs, then bigger chickens
will most likely make bigger chicken eggs. But thanks to
Morris, I am more confident in saying that bigger eggs get
you bigger chickens than I am saying bigger chickens get you
bigger eggs. Based on my research, I am also more confident in
saying that overbought/oversold SPX readings provide a sizable
edge in anticipating VIX futures movement than I am saying
overbought/oversold VIX readings provide a sizable edge in
anticipating SPX movement.
While proper risk management needs to be implemented in
any strategy, the techniques shared in this paper can be used
to design profitable VX models, and even potentially greatly
reduce drawdowns versus “short and hold”. When utilizing
measures of overbought/oversold on SPX, models can be created
that effectively filter unfavorable conditions for attempting
short-volatility trades, and can even be used to identify
favorable opportunities for taking temporary long VX positions.
By turning conventional techniques around, VIX traders can
take advantage of the quantifiable edges that SPX movement

IFTA.ORG PAGE 107

 IFTA JOURNAL 2025 EDITION

Technical Analysis for the Trading Professional, 2nd Edition

By Constance M. Brown, McGraw Hill, 2012
Review by: John Gajewski, Past President, Australian Professional Technical Analysts, Inc.

It doesn’t take long for traders to realise that the one indicator rounds off her methodology with a detailed description of her
or method they love, neither produces the consistent results custom Composite Index, and insights on other useful aspects of
they were expecting, nor gives them a proper grip on the market the market, like volatility and depth of perception.
to generate consistent profits. All the techniques discussed in the book are considered in
Constance Brown addresses these issues in her book every instance of analysis. Constance herself provides the best
Technical Analysis for the Trading Professional, with this summary of her approach when she writes:
second edition incorporating her learning experience since “The Elliott Wave Principle gave me a sense of what was
the first edition 14 years earlier. She is generous in sharing her coming and the size of the move. Gann analysis gave me a sense
extensive knowledge and experience into the analysis of market of when a move would happen. Fibonacci targets taught me
behaviour, throwing in many insightful observations along the price targets, correct leverage, and where I was wrong before I
way. On key techniques she also gives you the formulae to alter was stopped out. Oscillators added to the probability of being
your charting system to display the adjustments she introduces. right and gave the needed permission to execute at a Fibonacci
As the title suggests, the book is aimed at the trading price target. But it was maturity that taught me not to be in the
professional who is seeking greater precision and better timing markets all the time. It was also experience that taught me we
in their trades. Novice traders should not be discouraged as see things faster in charts than we will see them unfold in real
there are still many takeaways for them. However, Constance life.”2
does bring an enormous amount of knowledge and experience This book is a valuable resource for anyone looking to deepen
into each market analysis so without some broad experience of their understanding of the market, improve their entries and
the market the discussion may feel a little overwhelming. But hold risk to a minimum.
that is the price of precision and better timing – a lot of detail,
history and hard work
Constance looks unfavourably upon the “Stochastic Default Notes
Club”1 – those traders who unthinkingly accept software 1
Constance M Brown, Technical Analysis for the Trading
vendors’ default settings for their indicators – and those Professional, 2nd Edition, 2012, p63.
traders who settle for approximations rather than aiming 2
ibid. p375.
for precision. To stay on the right side of the trend, maximise
profit potential and avoid the costs of stop-outs, hard, detailed,
intelligent work is needed.
The book is organised in three parts. In Part One, Constance
Brown addresses some misunderstandings of key indicators and
methods and introduces the reader to a number of her unique
parameter adjustments and overlays, especially with the use of
oscillators. She introduces her custom Composite Index which
is a variation on the popular RSI and explains how she uses the
two together to sift through market opportunities and better
identify the profitable trades, low-risk entry points and when
to take profit. There are valuable insights on trendlines, cycles,
moving averages, which are all directed to improving the quality
of the trade entry.
Part Two is about price objectives. Considerable amount of
time in this part is spent on Gann timing and target projection
techniques, and even more time on Elliott Wave analysis and
Fibonacci price projections. It is the proper understanding and
exact calculations and projections from relevant points – which
generally are not the price extremes – that will give the trader
the edge he needs to turn opportunities into profit.
In Part Three, the shortest section of the book, Constance

PAGE 108 IFTA.ORG

 IFTA JOURNAL 2025 EDITION

Author Profiles

Nashwan Mohammed Al-Thawr, MFTA David Linton, MFTA

Nashwan has more than 23 years of experience David is one of the UK’s leading technical
in financial markets analysis, technical analysis, analysts and a well-known market
options hedging strategies, trading equities, commentator. He is the founder and CEO of
ETFs, futures, futures options, and crypto. Updata, based in London. He is the author of
Cloud Charts—Trading Success with the Ichimoku
He is a former global stocks and derivatives Technique. He teaches the Ichimoku module for
strategist and one of the founders of the Investment Academy in the UK Society of Technical Analysts Diploma Course, and he is a
2005. Additionally, he has worked for the General Corporation member of the American Association of Professional Technical
of Social Security Funds as a finance and investment advisor Analysts. He is a partner at Betagroup, leaders in technical
and established the Investo4life fund as its CEO. analysis training.

Loïc Bellina, CFTe, MFTA Robert Hanna

Loïc is a financial executive at SNCF Voyageurs, Rob Hanna is a Registered Investment Adviser
the passenger division of France's national Representative of Eastsound Capital Advisors,
railway company. He specializes in management LLC d.b.a. Capital Advisors 360, where he
control, data analysis, and Business Intelligence, manages numerous short-term quantitative
with a focus on digital transformation and stock and bond models, as well as VIX-based
finance modernization. In his role, he manages models and VIX option overlay strategies. He
IT budgets exceeding €110 million, overseeing CAPEX and OPEX has worked as a full-time market professional since 2001,
within Agile frameworks. managing a private investment fund prior to joining CA360.
Certified as a Project Management Professional (PMP) and Hanna also runs QuantifiableEdges.com, focusing on
European Engineer (EUR ING), Loïc has 9 years of experience quantitative market analysis. Quantifiable Edges provides
in railway engineering and 6 years in information systems and research for individuals and institutions via a website and
Agile methodologies. His MFTA research introduces innovative subscriber letter, published five nights a week, and hosts
methods for quantifying trend strength and market sentiment, courses on VIX trading, market timing, and short-term trading.
refining Renko chart analysis with volatility filters and financial Hanna is a graduate of the Boston College Carroll School of
ratios. Management. He lives and works in Massachusetts, serving
Certified by the AMF (French Financial Markets Authority), institutional and individual clients across the US and abroad.
Loïc is passionate about financial markets and investment
advisory. He also serves as President of the Savoie Section John Gajewski
of MIF – Mutuelle d’Ivry, a mutual insurance organization. Mr. Gajewski is a retired market analyst having
Additionally, he is a certified first responder and a licensed radio spent over 30 years analysing currency, bond,
amateur, reflecting his commitment to community service and commodity and stock markets in Sydney for
technology. local banks and for The Chart Manager Report.
For many years he has been involved in
Nobuaki Kakimoto, CFTe, MFTA technical analysis education through the
Nobuaki has long been a prominent member of Securities Institute of Australia and remains involved in his
the Nippon Technical Analysts Association and local technical analysis society.
an individual investor. His main interest has
been in the field of Forex where he has used
various technical analysis tools including his
own creation since 2007.
As he has continuously traded using technical indicators, he has
learned that a deep understanding of what the technical tools
mean is essential to making trading decisions.
His MFTA thesis is a culmination of what he has learned through
wide range of trading using technical analysis tools. He calls
himself a newcomer in the field of technical analysis but desires
to continue his research which would help traders break free
from misconceptions and superficial understanding of technical
indicators. His MFTA paper would be the first presentation of
what he is aiming at.

IFTA.ORG PAGE 109

 IFTA JOURNAL 2025 EDITION

Davide Pandini, PhD, CMT, MFTA, CFTe, CSTA Dr. Patrick Winter
Davide Pandini holds a Ph.D. in Electrical and Dr. Winter is a Berlin-based AI entrepreneur. He
Computer Engineering from Carnegie Mellon holds a doctoral degree and two B.Sc. degrees in
University in Pittsburgh, Pennsylvania. He was Information Systems and Business
a research intern at Philips Research Labs in Administration, all with distinction, from the
Eindhoven, the Netherlands, and at Digital Universities of Osnabrück and Marburg. His
Equipment Corp., Western Research Labs, in long-standing interest in methodological
Palo Alto, California. He joined STMicroelectronics in Agrate research and experimentation frequently finds application in
Brianza, Italy, in 1995, where he is a Technical Director and a trading. His contributions to the field have earned him five
Fellow of STMicroelectronics Technical Staff. consecutive awards from the Association of Technical Analysts
Davide has authored and co-authored more than 50 papers in in Germany (VTAD). Despite this, he usually prefers not to trade
international journals and conference proceedings and served himself because he believes that it is more efficient to create
on the program committee of several premiere international value than to trade it, especially since investors then have an
conferences. He received the STMicroelectronics Corporate incentive to work with, rather than against, each other.
STAR Gold Award in 2008 and 2020, and the STRIVE Gold Award
in 2022, for R&D excellence. Since June 2015, he has been the
chairman of the ST Italy Technical Staff Steering Committee.
In the field of technical analysis, Dr. Pandini is a Certified
Financial Technician (CFTe) and a Master of Financial
Technical Analysis (MFTA) of IFTA, holds the Chartered Market
Technician (CMT) designation from the CMT Association, and
is a professional member of SIAT (Società’ Italiana Analisi
Tecnica). In 2021, he was the recipient of the XII SIAT Technical
Analyst Award in the Open category. He was a speaker at SIAT
Trading Campus in 2022, in Milano, Italy, and at the Investing
and Trading Forum in 2022 and 2023, in Rimini, Italy. He was
the winner of the prestigious IFTA 2021 John Brooks Memorial
Award and was a speaker at the IFTA 2023 annual conference in
Jakarta, Indonesia, at Bogu Investment Forum 2024, in China,
and will be at the IFTA 2024 annual conference in Boao, Hainan,
China.
Davide served as volunteer at the Universal Exhibition
Expo2015—Feeding the Planet, Energy for Life—in Milano, Italy.

PAGE 110 IFTA.ORG

 IFTA JOURNAL 2025 EDITION

IFTA Staff IFTA Board of Directors

Executive Director IFTA HEADQUARTERS
Linda Bernetich, CAE International Federation of Technical Analysts
Production Manager 1300 Piccard Drive, Suite LL 14
Jansen Vera Rockville, MD 20850 USA
Phone: +1 (240) 404-6508
Managing Editor Fax: +1 (301) 990-9771
Heather Rigby Email: admin@ifta.org | Web: www.ifta.org
Director of Accounting Services
Dawn Rosenfeld President
Wieland Arlt, CFTe (VTAD)
wieland.arlt@ifta.org
Treasurer and Secretary
Oliver Reiss, CFTe, MFTA (VTAD)
dr.oliver.reiss@gmail.com
Membership Director
Indrawijaya Rangkuti, MBA, CFTe (AATI)
indra.senna@gmail.com
Vice President Europe, Website Director
David Watts, BSc (Hons) CEng MICE MIWEM, FSTA (STA)
Dwatts360@gmail.com
Vice President Asia-Pacific
Akihiro Niimi, MFTA, CFTe (NTAA)
akihiro_niimi@ntaa.or.jp
Vice President The Americas, Webinar Director
Bruce Fraser (AAPTA)
rdwyckoff@me.com
Vice President Middle East and Africa
Ron William, CFTe, MSTA, CFTe (SAMT)
roniwilliam@gmail.com
Education Director
Saleh Nasser, CMT, CFTe (ESTA)
snasser72@gmail.com
Examination Director
Gregor Bauer, Ph.D., CFTe, (VTAD)
gregor.bauer@vtad.de
Marketing Director
Eddie Tofpik, MSTA (STA)
Eddie.Tofpik@admisi.com
Quant Director
Giovanni Trombetta, CFTA, Electronic Engineer (SIAT)
giovanni.trombetta@gandalfproject.com
Non-Institutional Director
Shinji Okada, CMTA, CFTe, MFTA (NTAA)
shinjiokadaresearch@gmail.com
Digital Media Director
Anisah Ozleen Othman, MSTA, CFTe (MATA)
remisieranis@gmail.com

IFTA.ORG PAGE 111