Data Science Intern Assignment: Algorithmic Trading
Objective:
To assess the candidate's ability to analyze financial data, develop predictive models, and
implement basic algorithmic trading strategies using data science techniques.
Task 1: Data Analysis and Feature Engineering
Dataset:
You are provided with historical stock price data for a company (or use a publicly available
dataset like the Yahoo Finance API to get data for any stock of your choice).
1. Download the data (e.g., 3 years of daily stock prices: open, close, high, low,
volume).
2. Preprocess the data:
○ Handle missing values, outliers, and anomalies.
○ Perform feature engineering by creating technical indicators (e.g., Moving
Averages, RSI, MACD) that could be useful for algorithmic trading.
3. Data Visualization:
○ Plot the stock’s price data with relevant technical indicators to visualize trends
and potential trading signals.
○ Plot the distribution of returns and volume over time.
Deliverables for Task 1:
● A Python script or Jupyter notebook demonstrating the steps of data preprocessing
and feature engineering.
● Data visualizations highlighting trends, patterns, and key indicators.
Task 2: Model Building
Now that you have engineered some features, the next task is to develop a predictive model
to forecast whether the price will go up or down the next day (binary classification).
1. Train a predictive model to forecast the direction of price movement (up/down) for
the next trading day using the engineered features.
○ Use machine learning algorithms such as Logistic Regression, Random
Forest, or XGBoost.
○ Split the data into training and testing sets.
○ Optimize the model with cross-validation and tuning parameters (if
applicable).
2. Evaluation:
○ Evaluate the model using accuracy, precision, recall, and F1-score.
○ Also, compute the confusion matrix and ROC-AUC curve to assess
performance.
Deliverables for Task 2:
● Code for training and evaluating the model.
● A report summarizing the model’s performance, and any improvements you would
suggest.
Task 3: Backtesting a Simple Trading Strategy
Using the predictive model built in Task 2, you need to create a simple algorithmic trading
strategy. The strategy will place a "buy" order if the model predicts the price will go up the
next day, and a "sell" order if the model predicts the price will go down.
1. Backtesting the strategy:
○ Backtest the trading strategy on historical data and calculate its performance
metrics (e.g., cumulative returns, Sharpe ratio).
○ Simulate trade execution with an assumed transaction cost (e.g., 0.1% per
trade) and slippage.
2. Evaluate the performance:
○ Plot the strategy’s performance over time, comparing it against the
buy-and-hold strategy.
○ Calculate performance metrics such as cumulative returns, maximum
drawdown, and Sharpe ratio.
Deliverables for Task 3:
● Python code for the backtesting process.
● A report with the backtest results, strategy performance, and insights.
Task 4: Optimization and Refinement
The next step in improving the algorithm is to optimize it.
1. Parameter Optimization:
○ Use techniques like grid search or random search to optimize
hyperparameters of the predictive model (e.g., number of trees in Random
Forest, learning rate in XGBoost).
2. Strategy Refinement:
○ Modify the trading strategy to include risk management (e.g., stop-loss,
take-profit).
○ Assess the effect of adding more features or improving the model on strategy
performance.
Deliverables for Task 4:
● Code for parameter optimization and strategy refinement.
● A brief report discussing the improvements made and the final results after
optimization.
Bonus Task: (Optional)
Alternative Models:
If time permits, experiment with more advanced models (e.g., Recurrent Neural Networks or
LSTM for time-series forecasting). Compare their performance with the traditional machine
learning models used in Task 2.
Deliverables:
● Code and a comparison of results between models.
Submission Guidelines:
● Submit all code in a Jupyter notebook or Python script format.
● Ensure the code is well-documented, and provide clear explanations for each step.
● Include a brief report summarizing your findings, challenges faced, and
improvements you would suggest for future work.
Evaluation Criteria:
1. Technical Skills: Ability to preprocess, analyze, and model financial data effectively.
2. Creativity: Innovative feature engineering and use of appropriate models and
techniques.
3. Clarity of Code and Reports: Well-structured code with clear explanations and
well-documented reports.
4. Backtesting and Strategy Performance: Understanding of trading strategies, and
ability to backtest and assess performance metrics.
5. Communication: Clear and concise communication of your approach, results, and
insights.
Good luck, and we look forward to seeing your work!
