Journal of Behavioral Finance

ISSN: 1542-7560 (Print) 1542-7579 (Online) Journal homepage: https://www.tandfonline.com/loi/hbhf20

A Closer Look at the Disposition Effect in U.S.

Equity Option Markets

Kelley Bergsma, Andy Fodor & Emily Tedford

To cite this article: Kelley Bergsma, Andy Fodor & Emily Tedford (2019): A Closer Look at
the Disposition Effect in U.S. Equity Option Markets, Journal of Behavioral Finance, DOI:
10.1080/15427560.2019.1615913

To link to this article: https://doi.org/10.1080/15427560.2019.1615913

Published online: 07 Jun 2019.

Submit your article to this journal

Article views: 21

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=hbhf20
JOURNAL OF BEHAVIORAL FINANCE
https://doi.org/10.1080/15427560.2019.1615913

A Closer Look at the Disposition Effect in U.S. Equity Option Markets

Kelley Bergsmaa , Andy Fodora, and Emily Tedfordb
a
Ohio University; b84.51

ABSTRACT KEYWORDS
The authors explore whether the disposition effect occurs in U.S. equity option markets. The Disposition effect; Options;
disposition effect implies past winning securities will be undervalued and past losing Behavioral finance;
securities will be overvalued. By adapting Grinblatt and Han’s unrealized capital gains proxy Reference price
to the option markets, the authors document a significant relationship between option cap-
ital gains overhang and option returns. They also find open interest decreases as option
capital gains overhang increases, consistent with a disposition effect in U.S. equity options.
This evidence contributes to the emerging literature on behavioral finance in deriva-
tive securities.

Introduction trading of individual investors” (Barberis and Xiong

[2009]) and refers to the notion that investors sell win-
Behavioral finance is a new lens through which schol-
ners too soon and hold losers too long (Kaustia [2004],
ars and practitioners understand financial decision mak-
Odean [1998], Shefrin and Statman [1985]). A few stud-
ing. Kahneman and Tversky [1979], Shefrin and
ies connect the disposition effect to option trading.
Statman [1985], Thaler [1985], and Tversky and
Heath, Huddart, and Lang [1999] documented that
Kahneman [1992] fundamentally changed how the
executives exercise stock options in response to a stock
world views finance. Later studies develop behavioral
reaching its 52-week high. Poteshman and Serbin
asset pricing into a robust strand of literature, including
[2003] reported that option traders irrationally exercise
Benartzi and Thaler [1995], Odean [1998], Shefrin and call options early after the stock price reaches its yearly
Statman [2000], and Barberis, Huang, and Santos high. Chang, Solomon, and Westerfield [2016] found
[2001]. A large volume of behavioral finance research that investors at a large discount broker exhibit a
concentrates on equity markets, yet such studies are reverse disposition effect for options, where the effect is
comparatively less common in nonequity markets. For marginally significant for equity options and insignifi-
instance, an emerging literature examines behavioral cant for index options.1 Chiang, Chui, and Chou [2016]
finance in option markets. Option traders misreact to documented a disposition effect in index option returns
changes in instantaneous variance (Poteshman [2001]), using moneyness-based propensity to sell and adjusted
exercise options too early (Poteshman and Serbin capital gains overhang measures.2 Capital gains over-
[2003]), and buy options on growth stocks that subse- hang is a measure developed by Grinblatt and Han
quently underperform (Lakonishok, Lee, Pearson, and [2005; hereafter GH] to quantify unrealized gains/losses
Poteshman [2007], Mahani and Poteshman [2008]). In in the aggregate stock market. Based on prospect theory
addition, prior studies report option traders are unduly and mental accounting, GH describe how the dispos-
influenced by sentiment and fear (Beilis, Dash, and ition effect implies that past winning stocks will be
Wise [2014], Chang, Chen, and Fuh [2013], Chang, undervalued and past losing stocks will be overvalued.
Hsieh, and Wang [2015], Han [2008]) and exhibit Using capital gains overhang to approximate the magni-
anchoring bias and irrational probability weighting tude of gain or loss for the marginal investor, GH report
(DeLisle, Diavatopoulos, Fodor, and Kreiger [2017]). a significant, monotonic relationship between capital
In this study, we explore whether equity option trad- gains overhang and stock returns. Using a similar meth-
ers exhibit a disposition effect. The disposition effect odology to GH, we generate an option capital gains
has been called “one of the most robust facts about the overhang variable with OptionsMetrics data from 1996

CONTACT Kelley Bergsma bergsma@ohio.edu Ohio University, College of Business Annex 206, 1 Ohio University, Athens, OH 45701.
Color versions of one or more figures in the article can be found online at www.tandfonline.com/hbhf.
ß 2019 The Institute of Behavioral Finance
2 K. BERGSMA ET AL.

to 2015. In contrast to stock traders, option traders face [2005], Odean [1998], Shapira and Venezia [2001], Singal
a much shorter investing horizon and a set expiration and Xu [2011], Wulfmeyer [2016]). Our study tests for a
date in which options sometimes expire worthless. For disposition effect in options by repurposing GH’s meth-
that reason, we use daily option inputs over a 20-day odology to fit the unique features of U.S. equity option
holding period. We are the first to demonstrate a sig- markets. In doing so, we contribute to the growing litera-
nificant relationship between option capital gains over- ture on behavioral finance outside of equity markets by
hang (OCGO) and option returns. examining one of key components of behavioral asset
Our closer look at the disposition effect in U.S. equity pricing—the disposition effect—in U.S. equity options.
options yields surprising insights. Given that more retail We begin by describing our data and methodology.
option traders write options than purchase options Then, we analyze our results. We conclude with a sum-
(G^arleanu, Pedersen, and Poteshman [2009], mary of our findings and suggestions for future research.
Lakonishok, Lee, Pearson, and Poteshman [2007]), we
construct OCGO from the net short option position. The
Data and Methodology
relationship between OCGO and option returns (either
raw or delta-hedged) is positive, after controlling for Our main sample is comprised of daily individual
underlying stock characteristics or option-level charac- equity option data from OptionMetrics from 1996 to
teristics. This relationship is generally strongest for out- 2015. OptionMetrics provides bid and ask quotes, open
of-the-money options (i.e., more speculative instru- interest, trading volume, strike prices, and implied vola-
ments) and most robust during high investor sentiment tility for all U.S. exchange-traded options. First, we
periods, while it is weakest in December when investors exclude options that have missing volume or open
are more likely to trade for tax-motivated purposes interest. In addition, options are omitted from the sam-
(Odean [1998]). We also find novel evidence of a devi- ple that have, at the outset, less than 90 days or more
ation from a strictly monotonic pattern (between OCGO than 365 days to expiration. We merge options data
and option returns) for options in the top and bottom with stocks in the Center for Research in Security Prices
OCGO deciles. This deviation in the extremes could be database and include only options with underlying
connected to Hartzmark’s [2015] rank effect, whereby stocks that are traded on NYSE, AMEX, or NASDAQ.
investors sell top and bottom ranked securities in their Second, to test for the disposition effect, we need an
portfolios. Last, we investigate whether open interest (the option reference price. As we do not have individual
number of option contracts outstanding) declines as transaction level data with purchase prices, we use a
OCGO rises. A traditional disposition effect would pre- proxy for the average investor’s reference price as in GH.
dict that low unrealized capital gains overhang will be We define option reference price, Rt, as the weighted
associated with increased open interest (as traders refuse average option price over the past 20 trading days, where
to close their positions) and high overhang will be associ- the weights are determined based on option turnover.
ated with decreased open interest (as traders close their We calculate Rt using a similar iterative formula as
positions too soon).3 Our results are consistent with this described in Equation 9 on page 319 of GH:
!
prediction: there is a negative relationship between X20 Y
n1
OCGO and open interest. Taken together, the evidence Rt ¼ Vtn ½1Vtnþs  Ptn ð1Þ
n¼1 s¼1
suggests as OCGO increases, retail option traders are
more likely to close out their short positions, pushing up In our study, however, V is option turnover and
option prices and reducing open interest. P is option price. Option turnover is daily volume
Prior literature explains the disposition effect as aris- divided by open interest. Option price is the midpoint
ing from diminished sensitivity, loss aversion, mean of the bid and offer prices. In addition, we use the
reversion, prospect theory, realization utility, and regret past 20 trading days with nonzero volume to better
(Barberis and Xiong [2009, 2012], Dacey and Zielonka reflect an option holder’ time horizon.5 In contrast,
[2008], Fogel and Berry [2006], Jiao [2017], Kahneman GH used stock turnover and stock price over a 5-year
and Tversky [1979], Kohsaka et al. [2017], Li and Yang time horizon, reflecting stockholders’ longer invest-
[2013], Odean [1998], Shefrin and Statman [1985]).4 ment periods. The weights multiplied by each daily
Despite debate on the source of the disposition effect, option price are scaled to sum to 1. The weight on
there is strong empirical evidence of a disposition effect Pt–n represents the probability that an option that was
in stocks, futures, and mutual fund trading (Choe and last purchased at date t – n has not been traded since
Eom [2009], Coval and Shumway [2005], Das [2012], then. Thus, an option with high turnover has a more
Grinblatt and Keloharju [2001], Locke and Onayev updated reference price than one with low turnover.
 JOURNAL OF BEHAVIORAL FINANCE 3

Third, our proxy for option capital gains overhang calculated using option prices only, while delta-hedged
is derived from Equation 11 on page 320 of GH. option returns are the delta-hedged dollar returns on
However, we calculate option capital gains overhang day t scaled by stock prices on day t – 1.6 Delta meas-
from the perspective of the short position since retail ures how much the option price is related to price
option traders are more likely to write options than movements in the underlying stock. Our key independ-
buy options (G^arleanu, Pedersen, and Poteshman ent variable is OCGO as defined in Equation 2.
[2009], Lakonishok, Lee, Pearson, and Poteshman Following An [2016], we control for stock-level charac-
[2007]). Our equation is the following: teristics, including firm size, book-to-market, prior
Pt2 Rt1 month’s return, momentum, and turnover. Size is the
OCGO ¼  ð2Þ
Pt2 natural logarithm of the market capitalization at the
where reference price, Rt-1, is the prior day’s Rt as end of the prior month. Book-to-market (B/M) is
defined in Equation 1. We truncate OCGO at the 1% defined as in Fama and French [1992]. We obtain book
and 99% levels to minimize the influence of outliers. value from Compustat. LRet is the prior month’s stock
Fourth, for our final sample, we include only options return. Momentum is the cumulative monthly stock
with between 14 to 60 days to expiration (Bakshi and return for months t – 12 to t – 2. Turnover is stock
Kapadia [2003]). We exclude firms with stock prices less turnover of the prior month.
than $2 following An [2016], and with market capitaliza- To explore whether option-level characteristics
tion less than $2 million. Similar to Cao and Han [2013], affect the relationship between option returns and
we include only options with positive volume and bid OCGO, we add option moneyness, bid-ask spread,
quotes, and specify the bid price must be strictly less and implied volatility to our regressions. K/S is the
than the ask price and the midpoint of the bid and ask strike price divided by the underlying stock price
quotes must be at least $0.25. from the prior day and captures option moneyness.
To test for the relationship between capital gains BidAsk is the option bid-ask spread from 3 days
overhang and returns within option markets, we con- before. ImpVol is implied volatility from the prior day
duct daily regressions using the Fama and MacBeth from OptionMetrics.
[1973] procedure. Our dependent variables are raw and Table 1 presents summary statistics. We divide the
delta-hedged option returns. Raw option returns are sample by OCGO decile and present summary

Table 1. Summary statistics.

OCGO Decile N OCGO Size B/M LRet Momentum Turnover K/S BidAsk ImpVol
Panel A: Variable means for calls
1 380,561 –0.41 20.57 0.47 0.07 0.34 0.25 0.91 0.08 0.54
2 383,505 –0.26 27.71 0.45 0.05 0.36 0.25 0.93 0.08 0.53
3 383,494 –0.18 32.04 0.44 0.05 0.37 0.25 0.95 0.08 0.54
4 383,503 –0.11 34.64 0.44 0.04 0.37 0.25 0.96 0.08 0.54
5 381,558 –0.04 35.37 0.44 0.04 0.37 0.25 0.97 0.09 0.54
6 542,553 0.06 34.23 0.42 0.03 0.40 0.25 0.99 0.09 0.56
7 545,553 0.19 30.80 0.42 0.02 0.40 0.26 1.02 0.10 0.57
8 545,510 0.37 25.86 0.43 0.01 0.39 0.26 1.05 0.12 0.58
9 545,552 0.69 22.08 0.43 0.00 0.39 0.26 1.10 0.16 0.58
10 543,546 1.81 20.25 0.42 –0.01 0.42 0.27 1.18 0.24 0.60
All 4,635,335 0.29 28.06 0.43 0.03 0.38 0.26 1.02 0.12 0.56
Panel B: Variable means for puts
1 210,364 –0.41 28.55 0.43 –0.04 0.28 0.30 1.12 0.08 0.60
2 213,340 –0.26 34.79 0.43 –0.03 0.28 0.30 1.09 0.07 0.59
3 213,331 –0.17 37.70 0.43 –0.02 0.29 0.30 1.07 0.08 0.59
4 213,340 –0.10 38.73 0.42 –0.02 0.29 0.30 1.06 0.08 0.59
5 211,328 –0.04 38.48 0.42 –0.01 0.30 0.30 1.04 0.08 0.58
6 403,075 0.07 39.89 0.43 0.00 0.28 0.30 1.02 0.08 0.55
7 406,025 0.22 37.91 0.44 0.01 0.28 0.29 0.98 0.09 0.54
8 406,119 0.41 34.44 0.44 0.02 0.29 0.28 0.95 0.11 0.54
9 406,024 0.74 31.95 0.44 0.02 0.30 0.28 0.92 0.14 0.54
10 404,098 1.87 28.89 0.42 0.04 0.34 0.28 0.87 0.20 0.55
All 3,087,044 0.37 34.97 0.43 0.00 0.29 0.29 0.99 0.11 0.56
Note. This table reports the summary statistics for calls and puts sorted by option capital gains overhang (OCGO) decile. Each day, options are sorted into
deciles based on OCGO. OCGO is defined in Equation 2. Size is market capitalization in billions. B/M (book-to-market) is defined as in Fama and French
[1992]. LRet is the prior month’s stock return. Momentum is the cumulative monthly stock return from month t – 12 to t – 2 in percent. Turnover is the
prior month’s stock turnover. K/S is the option strike price divided by the underlying stock price of the prior day. ImplVol is implied volatility of the
prior day (annualized). Bid-Ask is the option bid-ask spread scaled by price from 3 days prior. Mean values of all variables are presented for calls (Panel
A) and for puts (Panel B). The sample period is from January 1996 to December 2015.
4 K. BERGSMA ET AL.

statistics for each decile as well as for the entire sam- documented that consumer sentiment and investor
ple. Options are first divided based on negative or sentiment are distinct. We suggest investors are in a
positive OCGO, and then into loss quintiles or gain more optimistic mood when investor sentiment is
quintiles by OCGO magnitude. For the remainder of high (Byun and Kim [2016]). Individuals in a positive
the paper, loss quintiles will be referred to as OCGO mood rely more on heuristic strategies rather than
deciles 1–5 and gain quintiles will be referred to as critical analysis associated with a negative mood
OCGO deciles 6–10. Panel A summarizes calls, while (Hirshleifer [2001], Schwarz [1990]). Therefore, during
Panel B summarizes puts. There is no monotonic rela- high sentiment periods, investors holding onto losing
tionship between OCGO and stock-level characteris- positions may be more likely to overoptimistically
tics, except for LRet, which is because an option’s expect that these securities will appreciate and thus be
value is tied to the underlying stock price. In contrast, more willing to continue to hold these losing posi-
there is a monotonic relationship with OCGO and tions, exacerbating the disposition effect. For that rea-
option characteristics. For calls, as OCGO rises, K/S, son, we expect that behavioral biases are magnified
BidAsk, and ImpVol all increase. For puts, as OCGO when investor sentiment is high and thus predict the
rises, K/S decreases, BidAsk increases, and disposition effect will be stronger in high investor sen-
ImpVol decreases. timent periods than low investor sentiment periods.
In later tests, we conduct subsample analyses. We split We split the sample by median monthly investor sen-
the sample based on option moneyness. Since retail timent, where the sample ends in October 2015 based
investors prefer securities with positive skewness (Han on data availability from Jeffrey Wurgler’s website
and Kumar [2013]), Choy [2015] suggested retail traders [http://people.stern.nyu.edu/jwurgler/].
prefer to trade out-of-the-money (OTM) options due to Moreover, we examine whether the relationship
their positively skewed payoffs7 and found evidence of between OCGO and option returns exhibits a V-shape
the strongest mispricing among OTM options. As retail or another deviation from a monotonic pattern for
traders are more prone to the disposition effect than pro- extreme gains and losses. Inspired by Ben-David and
fessional traders (Calvet, Campbell, and Sodini [2009], Hirshleifer [2012], An [2016] modified GH’s original
Dhar and Zhu [2006], Feng and Seasholes [2005], Hur, capital gains overhang variable to isolate gains and
Pritamani, and Sharma [2010], Locke and Mann [2005], losses and found a V-shaped disposition effect, rather
Shapira and Venezia [2001]), we expect that OTM than a monotonic disposition effect, in stock returns.
options—most preferred by retail traders—will exhibit Using data from a large retail brokerage, Hartzmark
the strongest relationship between OCGO and option [2015] documented a new rank effect, whereby invest-
returns. For calls, in-the-money (ITM) is defined as K/S ors sell top- and bottom-ranked securities in their port-
< 0.95, at-the-money (ATM) is defined as 0.95  K/S  folios. Moving across the spectrum from a large loss to
1.05, and OTM is defined as K/S > 1.05. For puts, OTM a large gain, it is still an open question whether invest-
is defined as K/S < 0.95, ATM is defined as 0.95  K/S ors’ propensity to sell decreases or increases in a mono-
 1.05, and ITM is defined as K/S > 1.05. tonic or nonmonotonic fashion based on returns. To
Moreover, we examine the relationship between explore whether a V-shaped relationship exists, we
OCGO and option returns separately for January, define Gain as equal to OCGO if OCGO is greater than
February-November, and December. Shefrin and zero and as equal to zero otherwise. Loss is equal to
Statman [1985] and Odean [1998] found the dispos- OCGO if OCGO is less than zero and zero otherwise.8
ition effect is weaker in December due to tax loss sell- To test the notion that extreme OCGO deciles exhibit
ing, while GH reported that the relationship between distinct relationships, we construct XLoss and XGain
capital gains overhang and stock returns is attenuated variables. XLoss is equal to OCGO if OCGO is in the
in January. lowest decile that day and zero otherwise. XGain is
In addition, we examine whether the relationship equal to OCGO if OCGO is in the highest decile that
between OCGO and option returns differs in high or day and zero otherwise.
low investor sentiment periods using Baker and Our final test explores whether rising OCGO is
Wurgler’s [2007] monthly investor sentiment index. associated with decreases in open interest. We calcu-
Byun and Kim [2016] found that option mispricing is late the percentage change in open interest (OI CHG)
more pronounced in high investor sentiment periods. as open interest of the current day minus open inter-
Kumar [2009] reported that the disposition effect is est of the prior day scaled by the prior day’s open
weaker in the stock market when consumer sentiment interest. Our control variables include option-level
is low; however, Kim, Kim, and Seo [2017] characteristics: K/S, BidAsk, ImpVol, and DTE. DTE is
 JOURNAL OF BEHAVIORAL FINANCE 5

defined as days to expiration, which is known to while Models 6–10 examine puts. In Model 1 of panel A,
influence levels of open interest. the OCGO coefficient is positive and significant with an
estimate of 6.03 (t ¼ 26.91), suggesting a direct relation-
ship between OCGO and raw call returns. The OCGO
Results
coefficient remains positive in sign and highly statistically
To explore the disposition effect in options, we begin significant after controlling for moneyness, bid-ask
by creating equal-weighted OCGO decile portfolios spread, and implied volatility in Models 2–4, respectively.
each day and plotting average option portfolio returns With all 3 option characteristics included in Model 5, the
in Figures 1A (calls) and 1B (puts). For both calls and OCGO coefficient is 3.75 (t ¼ 4.49). Similarly, for puts,
puts, option returns increase from OCGO decile 1 to the OCGO coefficient is positive and significant in
OCGO decile 10. This pattern is more pronounced for Models 6–9. With all control variables included in Model
raw option returns than delta-hedged option returns, 10, the OCGO coefficient is 3.09 (t ¼ 8.63). These findings
but it holds true in either case. Returns are lower for demonstrate a direct relationship between OCGO and
options with large paper losses and returns are higher raw option returns for both calls and puts.
for options for large paper gains. In panel B, we investigate the relationship between
Next, we formally test this idea by regressing raw and OCGO and delta-hedged option returns. Interestingly,
delta-hedged option returns on OCGO. Table 2 presents OCGO is positive and significant in 6 of 10 specifications,
the results. Raw returns are reported in panel A and delta- suggesting delta-hedged option returns increase as OCGO
hedged returns in panel B. Models 1–5 examine calls, increases, consistent with the patterns in Figure 1.

Figure 1. Option returns by OCGO decile.

These figures present average raw and delta-hedged option returns for calls and puts by OCGO decile. OCGO is option capital
gains overhang as defined in Equation 2. Panel A depicts the relationship for calls, while Panel B does so for puts. The sample
period is from January 1996 to December 2015.
6 K. BERGSMA ET AL.

Table 2. Exploring the disposition effect in options.

Call Put
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Panel A: Raw option returns
OCGO 6.03 4.09 3.75 5.29 3.75 4.03 3.76 2.56 4.84 3.09
(26.91) (9.35) (22.95) (2.62) (4.49) (26.06) (9.39) (19.73) (11.98) (8.63)
K/S 30.92 25.37 –16.71 –11.68
(8.74) (4.16) (–13.33) (–6.61)
Bid-Ask 31.67 27.23 18.71 13.96
(25.62) (10.25) (9.63) (3.90)
ImpVol 0.18 –5.43 –1.25 –4.14
(0.03) (–1.85) (–0.73) (–3.15)
Ln(Size) –0.13 0.21 0.61 –0.16 0.97 –0.01 –0.30 0.36 –0.43 –0.27
(–0.76) (1.32) (5.33) (–0.28) (1.65) (–0.23) (–2.28) (4.48) (–1.86) (–2.71)
B/M –0.21 1.67 0.33 6.66 0.08 0.04 –0.62 –0.42 –2.73 –1.03
(–0.30) (3.59) (1.65) (1.00) (0.16) (0.18) (–0.72) (–1.57) (–1.23) (–2.46)
LRet 5.34 2.85 2.98 4.35 6.74 –4.31 –9.04 –6.08 –9.44 –9.83
(2.18) (0.99) (4.36) (1.51) (7.32) (–5.58) (–4.82) (–5.75) (–2.38) (–5.22)
Momentum –1.22 0.38 –0.33 –0.24 0.37 –0.38 –0.75 –0.58 0.69 –0.61
(–3.49) (1.08) (–2.25) (–0.77) (1.75) (–2.58) (–1.28) (–3.02) (0.60) (–2.46)
Turnover –3.15 –5.28 –0.14 –3.03 –1.25 –0.40 –0.79 1.19 –0.18 2.28
(–8.40) (–7.86) (–0.40) (–2.60) (–2.07) (–1.01) (–1.15) (3.06) (–0.24) (3.16)
Intercept 4.79 –33.16 –16.46 3.66 –44.91 –0.15 23.31 –10.58 10.40 18.14
(1.22) (–5.85) (–6.47) (0.21) (–2.67) (–0.12) (7.80) (–6.16) (1.79) (5.02)
Adj. R2 6.72% 8.40% 8.52% 8.10% 10.55% 8.33% 10.49% 9.66% 10.68% 12.89%
308,354 120,540 308,354 120,467 120,467 215,732 65,902 215,732 65,842 65,842
Panel B: Delta-hedged option returns
OCGO 0.07 0.06 0.05 0.37 0.07 0.07 0.01 0.07 0.11 0.07
(5.90) (1.06) (5.92) (1.02) (1.49) (15.30) (0.37) (13.30) (3.96) (2.01)
K/S 1.20 1.35 –0.85 –1.14
(3.72) (2.60) (–11.14) (–7.58)
Bid-Ask 0.15 0.09 0.11 –0.74
(3.31) (0.39) (2.49) (–2.17)
ImpVol 0.37 –0.87 –0.71 –0.82
(0.53) (–3.99) (–9.93) (–9.55)
Ln(Size) 0.07 0.09 0.07 0.09 0.10 0.04 0.04 0.05 –0.01 –0.02
(14.10) (5.92) (9.54) (1.79) (2.02) (17.52) (9.37) (17.16) (–1.80) (–5.48)
B/M 0.07 0.06 0.09 0.44 –0.02 0.02 0.08 0.01 –0.10 –0.05
(1.25) (2.05) (1.06) (0.93) (–0.44) (1.72) (1.99) (0.95) (–3.05) (–2.03)
LRet 0.41 0.14 0.43 0.44 0.42 –0.35 –0.63 –0.40 –0.56 –0.75
(3.12) (0.57) (3.08) (1.93) (8.36) (–6.04) (–7.32) (–6.65) (–5.96) (–7.01)
Momentum –0.03 0.07 0.00 0.03 0.04 –0.02 –0.07 –0.02 –0.03 –0.06
(–0.80) (2.34) (0.60) (1.75) (4.53) (–2.87) (–6.37) (–3.41) (–2.19) (–5.86)
Turnover 0.02 0.02 0.03 0.26 0.21 0.02 –0.02 0.03 0.26 0.24
(1.45) (0.55) (2.16) (6.24) (8.58) (1.26) (–0.71) (1.50) (6.99) (6.49)
Intercept –1.83 –3.55 –1.85 –2.41 –3.51 –1.25 –0.31 –1.28 0.32 1.89
(–19.73) (–6.23) (–16.09) (–1.75) (–2.44) (–19.60) (–2.71) (–18.91) (2.06) (8.91)
Adj. R2 2.92% 5.48% 3.00% 4.53% 6.62% 3.80% 8.87% 3.84% 6.07% 10.62%
N 308,353 120,540 308,353 120,467 120,467 215,730 65,902 215,730 65,842 65,842
Note. This table explores the disposition effect in options by studying the relationship between option returns and a proxy for unrealized capital gains/
losses. Daily Fama-MacBeth regressions are used. The proxy for unrealized capital gains/losses is option capital gains overhang (OCGO), defined in
Equation 2. The dependent variable in Panel A is the raw option returns, while in Panel B, the dependent variable is dollar delta-hedged returns on
day t scaled by stock price on day t – 1. All control variables are defined in Table 1. All coefficients are in percent. The t statistics are reported in paren-
theses. , , and  indicate 10%, 5%, and 1% level of significance, respectively, using a 2-tailed test. N is the number of options in each specifica-
tion. The sample period is from January 1996 to December 2015.

For completeness, the notable control variable coef- Table 2 indicates a clear positive relationship
ficients are discussed subsequently. K/S enters posi- between OCGO and option returns. This evidence
tively for calls and negatively for puts, suggesting suggests GH’s methodology in the stock market can
OTM options are associated with a negative lottery be applied to the U.S. equity option markets with
premium (Barberis and Huang [2008], Choy [2015]). interesting results. As OCGO is calculated using the
The BidAsk coefficients are generally positive and sig- net short position (a more common position for retail
nificant, suggesting an illiquidity premium. The option traders), the evidence suggests larger unrealized
ImpVol coefficient is negative in sign, consistent with gains on short option positions are associated with
Cao and Han [2013]. The LRet coefficients are posi- higher next day option returns as traders close out
tive for calls and negative for puts. their short positions early, pushing up option prices.9
 JOURNAL OF BEHAVIORAL FINANCE 7

Table 3. Moneyness.
Calls Puts
ITM ATM OTM OTM ATM ITM
(1) (2) (3) (4) (5) (6)
Panel A: Raw option returns
OCGO 2.14 0.90 4.33 2.00 0.10 1.36
(2.49) (4.28) (2.79) (9.57) (0.22) (0.76)
Controls included
Adj. R2 11.85% 7.34% 13.33% 15.69% 10.00% 14.37%
N 64,818 67,029 67,551 39,489 36,143 31,098
Panel B: Delta-hedged option returns
OCGO 0.15 0.00 0.04 0.00 –0.04 0.20
(2.47) (–0.34) (0.96) (–1.15) (–1.44) (1.61)
Controls included
Adj. R2 10.42% 7.74% 13.20% 17.20% 10.78% 14.71%
N 64,818 67,029 67,551 39,489 36,143 31,098
Note. This table explores whether the relationship between option returns and a proxy for unrealized capital gains/losses differs by option moneyness.
Daily Fama-MacBeth regressions are used. For calls, in-the-money (ITM) is defined as K/S < 0.95, at-the-money (ATM) is defined as 0.95  K/S  1.05,
and out-of-the-money (OTM) is defined as K/S > 1.05. For puts, OTM is defined as K/S < 0.95, ATM is defined as 0.95  K/S  1.05, and ITM is defined
as K/S > 1.05. The dependent variable in Panel A is the raw option returns, while in Panel B, the dependent variable is dollar delta-hedged returns on
day t scaled by stock price on day t – 1. OCGO is option capital gains overhang as defined in Equation 2. All specifications include all option- and
stock-level control variables in Models 5 and 10 of Table 2. All coefficients are in percent. The t statistics are reported in parentheses. , , and 
indicate 10%, 5%, and 1% level of significance, respectively, using a 2-tailed test. N is the number of options in each specification. The sample period is
from January 1996 to December 2015.

We later explore the change in option contracts out- advantages, it is possible that tax loss selling could affect
standing in a subsequent table. option trading in December and thus mitigate the dis-
After establishing the relationships between OCGO position effect at this time.
and option returns, we investigate whether these pat- In addition, we investigate the role of investor sen-
terns depend on the options’ moneyness. Table 3 timent in the disposition effect in options. Table 5
presents the results. All option- and stock-level con- repeats our main specifications separately for low and
trol variables are included. For raw option returns in high investor sentiment periods. The OCGO coeffi-
panel A, the OCGO coefficients are positive and sig- cients have much stronger statistical significance dur-
nificant across moneyness groups for calls, but OCGO ing high sentiment periods than during low sentiment
has the greatest in economic and statistical magnitude periods, particularly for raw option returns. Therefore,
for OTM calls. For puts, OCGO is significant for our findings suggest the relationship between OCGO
OTM puts, while it is insignificant for ATM and ITM and option returns is strongest in high investor senti-
puts. Panel B reports the results for delta-hedged ment periods, consistent with the notion that behav-
option returns. For calls, OCGO is insignificant in all ioral biases are magnified when investor sentiment
but 1 specification (ITM calls). Taken together, the is high.
evidence indicates OCGO and raw option returns are Furthermore, we explore alternative shapes—other
most closely connected for OTM options, supporting than a purely monotonic shape—for the disposition
the idea that behavioral biases are stronger among effect in options. We first test whether a V-shaped rela-
more speculative securities. tionship between capital gains overhang and delta-
Moreover, GH found that the relationship between hedged option returns is more consistent with the data
stock-level capital gains overhang and stock returns dif- than a monotonic pattern. To this end, we replace
fers in January, while Odean [1998] reported a weaker OCGO with Loss and Gain (as defined previously) in
relationship in December. Thus, we explore whether the Table 6 in Models 1, 3, 5, and 7. If a V-shaped dispos-
disposition effect in option markets changes around the ition effect holds, then the Loss coefficient should be
turn of the year. In Table 4, we repeat our main specifi- negative and Gain coefficient should be positive.
cations separately for January, February-November, and Instead, the Loss coefficient is either insignificant
December. In contrast to GH, our results are not weaker (Models 1, 3, and 5) or marginally significant and posi-
for January, but rather for December. In 3 of 4 specifica- tive (Model 7). Gain coefficients are insignificant in all
tions for December only, the OCGO coefficient is insig- but 1 model (Model 5), where it is positive and signifi-
nificant, while it is only marginally significant in the cant. Thus, there is little evidence that OCGO has a V-
remaining specification. Although the wash-sale rule is shaped relationship with option returns.
meant to hinder market participants from selling stocks We then consider whether OCGO has a different
at a loss and buying options on those stocks for tax impact on option returns if options are in the top and
8 K. BERGSMA ET AL.

Table 4. Turn of the year.

Calls Puts
Jan Feb-Nov Dec Jan Feb-Nov Dec
(1) (2) (3) (4) (5) (6)
Panel A: Raw option returns
OCGO 3.36 2.83 12.13 3.18 3.18 2.27
(6.84) (10.36) (1.43) (5.41) (7.84) (1.83)
Controls Included
Adj. R2 9.74% 10.63% 10.53% 13.33% 12.92% 12.17%
N 18,653 111,468 24,594 9,071 60,962 12,586
Panel B: Delta-hedged option returns
OCGO –0.02 0.05 0.37 0.01 0.07 0.03
(–1.09) (1.30) (0.92) (0.52) (1.96) (0.38)
Controls Included
Adj. R2 6.10% 6.58% 7.36% 10.81% 10.57% 10.93%
N 18,653 111,468 24,594 9,071 60,962 12,586
Note. This table explores whether the relationship between option returns and a proxy for unrealized capital gains/losses differs around the turn of the
year. We estimate daily Fama-MacBeth regressions separately for the months of January (Models 1 and 4), February-November (Models 2 and 5), and
December (Models 3 and 6). The dependent variable in Panel A is the raw option returns, while in Panel B, the dependent variable is dollar delta-
hedged returns on day t scaled by stock price on day t – 1. OCGO is option capital gains overhang as defined in Equation 2. All specifications include
all option- and stock-level control variables in Models 5 and 10 of Table 2. All coefficients are in percent. The t statistics are reported in parentheses. ,
, and  indicate 10%, 5%, and 1% level of significance, respectively, using a 2-tailed test. N is the number of options in each specification. The
sample period is from January 1996 to December 2015.

Table 5. Investor sentiment.

Delta-hedged
Raw option returns option returns
Call Put Call Put
Low High Low High Low High Low High
(1) (2) (3) (4) (5) (6) (7) (8)
OCGO 4.81 3.03 3.70 2.71 0.17 0.00 0.12 0.02
(2.42) (21.20) (4.54) (14.01) (1.47) (0.43) (1.56) (3.86)
Controls included
Adj. R2 10.47% 10.30% 12.34% 13.00% 7.12% 6.10% 10.57% 10.39%
N 54,165 75,432 32,221 38,464 54,165 75,432 32,221 38,464
Note. This table explores whether the relationship between option returns and a proxy for unrealized capital gains/losses differs based on investor senti-
ment. We estimate daily Fama-MacBeth regressions separately for the months with high or low investor sentiment. A portfolio month is classified as
High Sentiment is if last month’s investor sentiment (Baker and Wurgler [2007]) is above the sample median; if otherwise, it is classified as Low
Sentiment. The dependent variable in Models 1–4 is the raw option returns, while in Models 5–8, the dependent variable is dollar delta-hedged returns
on day t scaled by stock price on day t – 1. OCGO is option capital gains overhang as defined in Equation 2. All specifications include all option- and
stock-level control variables in Models 5 and 10 of Table 2. All coefficients are in percent. The t statistics are reported in parentheses. , , and 
indicate 10%, 5%, and 1% level of significance, respectively, using a 2-tailed test. N is the number of options in each specification. The sample period
is from January 1996 to December 2015.

bottom OCGO deciles (i.e., extreme gains and losses). between OCGO and option returns. This deviation
In Models 2, 4, 6, and 8 of Table 6, option returns are could reflect investors’ disposition to trade the top and
regressed on OCGO, XLoss, XGain, and control varia- bottom securities in their portfolios differently
bles. If there is a deviation in the disposition effect for (Hartmark’s [2015] rank effect).
extreme gains and losses (Hartmark [2015]), the XLoss Last, we examine whether open interest decreases
and XGain coefficients should be significant. The as OCGO increases. We expect low unrealized capital
OCGO coefficient is positive and significant for raw call gains overhang is associated with greater open interest
returns (Model 2) yet is insignificant for all remaining (as option writers hold their positions too long) and
specifications. In contrast, the XLoss coefficients are sig- high overhang is associated with reduced open interest
nificant for raw put returns (0.34, t ¼ 4.09) and for (as traders close their positions too quickly). Our
delta-hedged put returns (0.26, t ¼ 2.03). Also, the dependent variable is the percentage change in open
XGain coefficients are positive and significant for raw interest (OI CHG) as compared with the prior day.
call returns (2.32, t ¼ 5.41) and for delta-hedged call Table 7 presents the results. In Model 1, the OCGO
returns (1.19, t ¼ 2.30). The relationship between coefficient is negative and significant for calls with an
OCGO and option returns demonstrates a deviation estimate of –1.92 (t ¼ –2.24). Thus, as our proxy for
from the general pattern for extreme losses and gains. unrealized capital gains increases, open interest
This deviation is consistently amplifying the original decreases. This evidence suggests retail traders who
pattern, suggesting a modified monotonic relationship have written calls are more likely to close these
 JOURNAL OF BEHAVIORAL FINANCE 9

Table 6. The shape of the disposition effect.

Delta-hedged
Raw option returns option returns
Call Put Call Put
(1) (2) (3) (4) (5) (6) (7) (8)
Loss 3.12 1.27 0.91 0.35
(0.60) (1.48) (0.16) (1.67)
Gain 0.11 –0.61 4.35 0.12
(0.03) (–1.21) (3.34) (1.36)
OCGO 2.05 0.05 0.65 –0.07
(2.23) (0.87) (0.49) (–0.68)
XLoss –0.82 0.34 0.36 0.26
(–1.38) (4.09) (0.27) (2.03)
XGain 2.32 –0.02 1.19 –0.01
(5.41) (–1.03) (2.30) (–0.47)
Controls Included
Adj. R2 10.69% 10.79% 7.02% 7.07% 13.17% 13.17% 11.45% 11.77%
N 120,467 120,467 65,842 65,842 120,467 120,467 65,842 65,842
Note. This table explores whether the relationship between option returns and a proxy for unrealized capital gains/losses is not strictly monotonic. Daily
Fama-MacBeth regressions are used. The dependent variable in Models 1–4 is the raw option returns, while in Models 5–8, the dependent variable is
dollar delta-hedged returns on day t scaled by stock price on day t – 1. OCGO is option capital gains overhang as defined in Equation 2. Loss is equal
to OCGO if OCGO is less than zero and zero otherwise. Gain is equal to OCGO if OCGO is greater than zero and zero otherwise. XLoss as equal to OCGO
if OCGO is in the lowest decile that day and zero otherwise. XGain is equal to OCGO if OCGO is in the highest decile that day and zero otherwise. All
specifications include all option- and stock-level control variables in Models 5 and 10 of Table 2. All coefficients are in percent. The t statistics are
reported in parentheses. , , and  indicate 10%, 5%, and 1% level of significance, respectively, using a 2-tailed test. N is the number of options
in each specification. The sample period is from January 1996 to December 2015.

Table 7. Open interest. Conclusions

Calls Puts Although option traders are considered to be more
(1) (2)
rational than stock traders (Ofek, Richardson, and
OCGO –1.92 –5.49
(–2.24) (–1.33) Whitelaw [2004]), option traders are still human and
K/S 18.27 –14.10 thus subject to cognitive biases. We investigate the
(3.68) (–1.89)
Bid-Ask –1.83 3.05 behavioral bias of the disposition effect in U.S. equity
(–0.58) (2.15) option markets. By adapting GH’s methodology to
ImpVol –0.01 –7.65
(0.00) (–1.14) options, we find a significant relationship between
DTE –0.15 0.03 OCGO and daily option returns. As retail traders are
(–1.42) (1.05)
Intercept –5.21 27.74 more likely to hold net short positions, OCGO is cal-
(–1.52) (1.84) culated from the option writer’s perspective. OCGO
Adj. R2 0.99% 1.49%
N 132,272 71,509 has a direct relationship with raw and delta-hedged
Note. This table explores the disposition effect in options by studying the option returns. These findings are stronger for OTM
relationship between option open interest and a proxy for unrealized options and during high investor sentiment periods.
capital gains/losses. Daily Fama-MacBeth regressions are used. The
dependent variable in the percentage change in open interest from the Retail investors are more likely to prefer OTM options
prior day to the current day. Option capital gains overhang (OCGO) is due to lottery-like payoffs and may be more prone to
defined in Equation 2. All option-level control variables are defined in
Table 1, except for days to expiration (DTE). All coefficients are in per- behavioral biases during high sentiment periods.
cent. The t statistics are reported in parentheses. , , and  indi- Moreover, open interest decreases as OCGO rises,
cate 10%, 5%, and 1% level of significance, respectively, using a
2-tailed test. N is the number of options in each specification. The sam- supporting the notion of a traditional disposition
ple period is from January 1996 to December 2015. effect as traders are more apt to close positions as
unrealized gains increase.
Our findings suggest promising avenues for future
positions as OCGO rises, consistent with a traditional research. Our analysis is limited to aggregate option
disposition effect. In Model 2, the OCGO coefficient volume data, yet future researchers should examine
is insignificant for puts, but negative in sign (–5.49, t similar questions using transaction-level options data.
¼ –1.33), again supporting the classic dispos- For instance, Chang et al. [2016] used brokerage data
ition effect. that includes options, but leave 3 potential areas unex-
Overall, our evidence suggests as option capital plored. First, it would be interesting to differentiate
gains overhang rises, retail option traders are more between the disposition effect in calls versus puts at
likely to close their short positions, pushing up option the transaction level and it would be worth examining
prices and reducing open interest. what percentage of options positions are unhedged,
10 K. BERGSMA ET AL.

hedged, or part of other strategies, such as covered 9. Beschwitz and Massa [2017] found the opposite result
calls (Hoffmann and Fischer [2012]; Shefrin and for short sellers in the stock market, such that stocks
Statman [1993]). Second, a researcher could replicate with high short sale capital gains overhang experience
more negative returns. However, their study examines
Ben-David and Hirshleifer’s [2012] stock-level analysis weekly stock returns from 2004 to 2010 in contrast to
using options to determine whether there is a V- our daily study of option returns from 1996 to 2015.
shaped disposition effect or whether there is a devi-
ation from a strictly monotonic relationship stemming
from selling top and bottom ranked securities in Acknowledgments
investor portfolios (Hartzmark [2015]). Third, it The authors appreciate helpful comments from Diego
would be useful to determine what percentage of Amaya, Bastian von Beschwitz, Sinan Gokkaya, Danling
option traders are disposition effect-prone relative to Jiang, Aristogenis Lazos, Xi Liu, Thomas Maurer, Andrea
Rossi, Jitendra Tayal, and Lee Wakeman, and seminar par-
the percentage of disposition effect-prone stock trad-
ticipants at the 2017 Financial Management Association,
ers, which, in turn, would quantify how an average 2017 Midwest Finance Association, and Ohio University. A
option trader is influenced by cognitive biases as com- previous version was circulated under the title, “A Strange
pared with an average stock trader. The previous Disposition? Option Trading, Reference Prices,
approach would further our understanding of behav- and Volatility.”
ioral finance in the new frontier of option markets.
ORCID
Notes Kelley Bergsma http://orcid.org/0000-0001-7391-8881
1. Schmitz and Weber [2012] reported a significant
disposition effect for call and put bank-issued warrants
among investors at a large German discount broker.
References
2. Lemmon and Ni [2014] reported that equity options An, L. “Asset Pricing When Traders Sell Extreme Winners
are mainly traded by individual investors, while index and Losers.” Review of Financial Studies, 29, (2016), pp.
options are less so. 823–861.
3. We thank the referee for suggesting this Baker, M., and J. Wurgler. “Investor Sentiment in the Stock
additional test. Market.” Journal of Economic Perspectives 21, (2007), pp.
4. Hens and Vleck [2011] found the standard prospect 129–151.
theory explanation is sound ex-poste, but not ex ante. Bakshi, G., and N. Kapadia. “Delta-Hedged Gains and the
Kadous et al. [2014] also reported that mean reversion Negative Market Volatility Risk Premium.” Review of
does not explain the disposition effect, but rather it is Financial Studies, 16, (2003), pp. 527–566.
influenced by self-regard and investing confidence in Barberis, N., M. Huang, and T. Santos. “Prospect Theory
an experimental market. and Asset Prices.” Quarterly Journal of Economics, 116,
5. Grinblatt and Keloharju [2001] found when a stock is
(2001), pp. 1–53.
at a monthly high or low, stockholders trade
Barberis, N., and M. Huang. “Stocks as Lotteries: The
differently than at other times, suggesting that a
Implications of Probability Weighting for Security
disposition effect holds over a 1-month horizon.
Prices.” American Economic Review, 98, (2008),
6. We scale by stock price for the following reason. If
option writers are delta-hedging covered calls (the 2066–2100.
Barberis, N., and W. Xiong. “What Drives the Disposition
most common retail option trading strategy according
to Lakonishok et al. [2007]), then they will incur the Effect? An Analysis of a Long-Standing Preference-Based
cost of the stock. Explanation.” Journal of Finance, 64, (2009), pp. 751–784.
7. Byun and Kim [2016] corroborated that OTM options Barberis, N., and W. Xiong. “Realization Utility.” Journal of
have lottery-like payoffs. Financial Economics, 104, (2012), pp. 251–271.
8. Choi, Hoyem, and Kim [2010] defined CGOHigh as Beilis, A., J. W. Dash, and J. V. Wise. “Psychology, Stock/
MAX(Normalized CGO – 0.5, 0) and CGOLow as FX Trading and Option Prices.” Journal of Behavioral
–MIN(Normalized CGO – 0.5, 0) to capture the Finance, 15, (2014), pp. 251–268.
U-shaped of capital gains overhang in the stock Ben-David, I., and D. A. Hirshleifer. “Are Investors Really
market. An [2016] used a different method to generate Reluctant to Realize Their Losses? Trading Responses to
Loss Overhang and Gain Overhang and also creates Past Returns and the Disposition Effect.” Review of
V-shaped Net Selling Propensity (VNSPt) ¼ Gaint – Financial Studies, 25, (2012), pp. 2485–2532.
0.23Losst. We do not use An’s method because option Benartzi, S., and R. Thaler. “Myopic Loss Aversion and the
prices are mechanically higher when there is more Equity Premium Puzzle.” Quarterly Journal of Economics,
time to expiration (due to greater time value) so Gain 110, (1995), pp. 73–92.
Overhang would likely contain option prices that Beschwitz, B., and M. Massa. “Biased Shorts: Short Sellers’
occurred further away in time than Loss Overhang’s Disposition Effect and Limits to Arbitrage.” INSEAD,
option prices. Working Paper (2017).
 JOURNAL OF BEHAVIORAL FINANCE 11

Byun, S.-J., and D.-H. Kim. “Gambling Preference and Grinblatt, M., and B. Han. “Prospect Theory, Mental
Individual Equity Option Returns.” Journal of Financial Accounting, and Momentum.” Journal of Financial
Economics, 122, (2016), pp. 155–174. Economics, 78, (2005), pp. 311–339.
Calvet, L. E., J. Y. Campbell, and P. Sodini. “Fight or Grinblatt, M., and M. Keloharju. “What Makes Investors
Flight? Portfolio Rebalancing by Individual Investors,” Trade?” Journal of Finance 56, (2001), pp. 589–616.
Quarterly Journal of Economics, 124, (2009), pp. 301–348. Han, B. “Investor Sentiment and Option Prices.” Review of
Cao, J., and B. Han. “Cross Section of Option Returns and Financial Studies, 21, (2008), pp. 387–414.
Idiosyncratic Stock Volatility.” Journal of Financial Han, B., and A. Kumar. “Speculative Retail Trading and
Economics, 108, (2013), pp. 231–249. Asset Prices.” Journal of Financial and Quantitative
Chang, C., L.-J. Chen, and C.-D. Fuh. “The Pricing of Risk Analysis, 48, (2013), pp. 377–404.
and Sentiment: A Study of Executive Stock Options.” Hartzmark, S. “The Worst, The Best, Ignoring All the Rest:
Financial Management, 42, (2013), pp. 79–99. The Rank Effect and Trading Behavior.” Review of
Chang, C-C., P.-F. Hsieh, and Y.-H. Wang. “Sophistication, Financial Studies, 28, (2015), pp. 1024–1059.
Sentiment, and Misreaction.” Journal of Financial and Heath, C., S. Huddart, and M. Lang. “Psychological Factors
Quantitative Analysis, 50, (2015), pp. 903–928. and Stock Option Exercise.” Quarterly Journal of
Chang, T., D. Solomon, and M. Westerfield. “Looking for Economics, 114, (1999), pp. 601–627.
Someone to Blame: Delegation, Cognitive Dissonance, Hens, T., and M. Vleck. “Does Prospect Theory Explain the
and the Disposition Effect.” Journal of Finance, 71, Disposition Effect?” Journal of Behavioral Finance, 12,
(2016), pp. 267–302. (2011), pp. 141–157.
Chiang, M.-H., H.-Y. Chiu, and R. K. Chou. “Measuring Hirshleifer, D. “Investor Psychology and Asset Pricing.”
the Disposition Effect on the Option Market: New Journal of Finance, 56, (2001), pp. 1533–1598.
Evidence.” Working Paper (2016). Hoffmann, A. O. I., and E. T. S. Fischer. “Behavioral
Choe, H., and Y. Eom. “The Disposition Effect and Aspects of Covered Call Writing: An Empirical
Investment Performance in the Futures Market.” Journal Investigation.” Journal of Behavioral Finance, 13, (2012),
of Futures Markets, 29, (2009), pp. 496–522. pp. 66–79.
Choi, W., K. Hoyem, and J.-W. Kim. “Capital Gains Hur, J., M. Pritamani, and V. Sharma. “Momentum and the
Overhang and the Earnings Announcement Premium.” Disposition Effect: The Role of Individual Investors.”
Financial Analysis Journal, 66, (2010), pp. 40–53.
Financial Management, 39, (2010), pp. 1155–1176.
Choy, S.-K. “Retail Clientele and Option Returns.” Journal
Jiao, P. “Belief in Mean Reversion and the Disposition
of Banking and Finance, 51, (2015), pp. 26–42.
Effect: An Experimental Test.” Journal of Behavioral
Coval, J., and T. Shumway. “Do Behavioral Biases Affect
Finance, 18, (2017), pp. 29–44.
Prices?” Journal of Finance, 60, (2005), pp. 1–34.
Kadous, K., W. B. Tayler, J. M. Thayer, and D. Young.
Dacey, R., and P. Zielonka. “A Detailed Prospect Theory
“Individual Characteristics and the Disposition Effect:
Explanation of the Disposition Effect.” Journal of
The Opposing Effects of Confidence and Self-Regard.”
Behavioral Finance, 9, (2008), pp. 43–50.
Journal of Behavioral Finance, 15, (2014), pp. 235–250.
Das, A. “Estimating the Loss from the Disposition Effect: A
Kahneman, D., and A. Tversky. “Prospect Theory: An
Simulation Study.” Journal of Behavioral Finance, 13,
(2012), pp. 1–10. Analysis of Decision Under Risk.” Econometrica, 47,
Dhar, R., and N. Zhu. “Up Close and Personal: Investor (1979), pp. 263–291.
Sophistication and the Disposition Effect.” Management Kaustia, M. “Market-Wide Impact of the Disposition Effect:
Science, 52, (2006), pp. 726–740. Evidence from IPO Trading Volume.” Journal of
DeLisle, J., D. Diavatopoulos, A. Fodor, and K. Kreiger. Financial Markets, (2004), pp. 207–235.
“Prospect Theory, Mental Accounting, and Option Kohsaka, Y., G. Mardyla, S. Takenaka, and Y. Tsutsui.
Prices.” Journal of Futures Markets, 37, (2017), pp. “Disposition Effect and Diminishing Sensitivity: An
614–638. Analysis Based on a Simulated Experimental Stock
Fama, E. F., and K. R. French. “The Cross-Section of Market.” Journal of Behavioral Finance, 18, (2017), pp.
Expected Stock Returns.” Journal of Finance, 47, (1992), 189–201.
pp. 427–465. Kim, J. S., D.-H. Kim, and S.W. Seo. “Investor Sentiment
Fama, E. F., and J. MacBeth. “Risk and Return: Some and Return Predictability of the Option to Stock Volume
Empirical Tests.” Journal of Political Economy, 81, (1973), Ratio.” Financial Management, 46, (2017), pp. 767–796.
pp. 607–636. Kumar, A. “Hard-to-Value Stocks, Behavioral Biases, and
Feng, L., and M. Seasholes. “Do Investor Sophistication and Informed Trading.” Journal of Financial and Quantitative
Trading Experience Eliminate Behavioral Biases in Analysis, 44, (2009), pp. 1375–1401.
Financial Markets?” Review of Finance, 9, (2005), pp. Lakonishok, J., I. Lee, N. D. Pearson, and A. M. Poteshman.
305–351. “Option Market Activity.” Review of Financial Studies, 20,
Fogel, S. O., and T. Berry. “The Disposition Effect and (2007), pp. 813–857.
Individual Investor Decisions: The Role of Regret and Lemmon, M., and S. Ni. “Differences in Trading and
Counterfactual Alternatives.” Journal of Behavioral Pricing between Stock and Index Options.” Management
Finance, 7, (2006), pp. 107–116. Science, 60, (2014), pp. 1985–2001.
G^arleanu, N., L. H. Pedersen, and A. M. Poteshman. Li, Y., and L. Yang. “Prospect Theory, the Disposition
“Demand-Based Option Pricing.” Review of Financial Effect, and Asset Prices.” Journal of Financial Economics,
Studies, 22, (2009), pp. 4259–4299. 107, (2013), pp. 715–739.
12 K. BERGSMA ET AL.

Locke, P. R., and S. C. Mann. “Professional Trader Motivation and Cognition: Foundations of Social
Discipline and Trade Disposition.” Journal of Financial Behavior, Volume 2, (pp. 527–561.) New York: Guilford
Economics, 76, (2005), pp. 401–444. Press, 1990.
Locke, P. R., and Z. Onayev. “Trade Duration: Information Shapira, Z., and I. Venezia. “Patterns of Behavior of
and Trade Disposition.” The Financial Review, 40, (2005), Professionally Managed and Independent Investors.”
pp. 113–129. Journal of Banking and Finance, 25, (2001), pp.
Mahani, R., and A. Poteshman. “Overreaction to Stock 1573–1587.
Market News and Misevaluation of Stock Prices by Shefrin, H., and M. Statman. “The Disposition to Sell
Unsophisticated Investors: Evidence from the Option Winners Too Early and Ride Losers Too Long: Theory
Market.” Journal of Empirical Finance, 15, (2008), pp. and Evidence.” Journal of Finance, 40, (1985), pp.
635–655. 777–790.
Odean, T. “Are Investors Reluctant to Realize Losses?” Shefrin, H., and M. Statman. “Behavioral Aspects of the
Journal of Finance, 53, (1998), pp. 1775–1798.
Design and Marketing of Financial Products.” Financial
Ofek, E., M. Richardson, and R. Whitelaw. “Limited
Management, 22, (1993), pp. 123–134.
Arbitrage and Short Sales Restrictions: Evidence from the
Shefrin, H., and M. Statman. “Behavioral Portfolio Theory.”
Options Markets. Journal of Financial Economics, 74,
Journal of Financial and Quantitative Analysis, 35,
(2004), pp. 305–342.
Poteshman, A. “Underreaction, Overreaction, and (2000), pp. 127–151.
Increasing Misreaction to Information in the Options Singal, V., and Z. Xu. “Selling Winners, Holding Losers:
Market.” Journal of Finance, 56, (2001), pp. 851–876. Effect on Fund Flows and Survival of Disposition-Prone
Poteshman, A., and V. Serbin. “Clearly Irrational Financial Mutual Funds.” Journal of Banking and Finance, 35,
Market Behavior: Evidence From the Early Exercise of (2011), pp. 2704–2718.
Exchange Traded Stock Options.” Journal of Finance, 58, Thaler, R. “Mental Accounting and Consumer Choice.”
(2003), pp. 37–70. Marketing Science, 4, (1985), pp. 199–214.
Schmitz, P., and M. Weber. “Buying and Selling Behavior of Tversky, A., and D. Kahneman. “Advances in Prospect
Retail Investors in Option-Like Securities.” DBW - Die Theory: Cumulative Representation of Uncertainty.”
Betriebswirtschaft, 72, (2012), pp. 409–426. Journal of Risk and Uncertainty, 5, (1992), pp. 297–323.
Schwarz, N. “Feelings as Information: Informational and Wulfmeyer, S. “Irrational Mutual Fund Managers:
Motivational Functions of Affective States.” In E. T. Explaining Differences in Their Behavior.” Journal of
Higgins and R. Sorrentino (eds.), Handbook of Behavioral Finance, 17, (2016), pp. 99–123.