Evaluation of Trading Strategies Using Python

The following basic trading strategies are implemented and evaluated as part of this analysis:

• Bollinger Bands
• Simple Moving Average Cross Over
• Buy and Hold

Market-Data-EDA.ipynb

We start by analysing the dataset downloaded from Kaggle that contains stock and ETF OHLC (Open, High, Low, Close) prices. This dataset consists of a large number of individual files with each containing market data for one ticker. We first develop a market data API that abstracts away access to ticker files before proceeding with an EDA. We also showcase usage of Dask given that the market data is spread across a number of files and all of the data cannot fit into memory on my laptop.

Basic-Strategies.ipynb

We then implement a simple backtester for running trading strategies. As part of it we develop a number of Python classes: Strategy, Backtest, ConfigSearch, etc. The backtester allows us to search for the best perforing ticker & strategy combinations. We run a backtest for all 3 strategies using 5 year's worth of market data.

Getting Started

Install prerequisites:

$ conda env create -f environment.yml 

Jupyter notebooks are located under notebooks directory:

• Market-Data-EDA.ipynb
• Basic-Strategies.ipynb

Acknowledgements

• The strategy backtesting notebook is inspired by the following Kaggle analysis:
  • https://www.kaggle.com/malyvsen/algotrading-analysis
• Some of the ideas around backtesting are inspired by quantstart.com
  • https://www.quantstart.com/articles/Should-You-Build-Your-Own-Backtester/
  • https://www.quantstart.com/articles/Event-Driven-Backtesting-with-Python-Part-I/

Potential Future Improvements

• Implement and evaluate the Mean Reversion Pairs Trading strategy

  • https://www.quantstart.com/articles/Backtesting-An-Intraday-Mean-Reversion-Pairs-Strategy-Between-SPY-And-IWM/
  • https://www.pythonforfinance.net/2018/07/04/mean-reversion-pairs-trading-with-inclusion-of-a-kalman-filter/
  • https://medium.com/auquan/pairs-trading-data-science-7dbedafcfe5a

• Develop a more sophisticated backtester

References

• Kaggle dataset:
  • https://www.kaggle.com/borismarjanovic/price-volume-data-for-all-us-stocks-etfs
• Quantopian lectures