HMM-CryptoStrategist

Introduction

This project explores the application of Hidden Markov Models (HMM) in developing a cryptocurrency trading strategy. Focusing on Bitcoin (BTC-USD), it utilizes historical data for model training and strategy testing.

Data Extraction

Historical BTC-USD data is fetched from Yahoo Finance for the period from '2022-01-01' to '2023-12-31', including features like Open, High, Low, Adjusted Close, and Volume.

Feature Engineering

• Daily returns and price ranges are calculated.
• 12-day and 21-day moving averages are computed.
• The dataset is split into training and testing sets.

Model Training

A Gaussian Hidden Markov Model with 4 components is trained on 'Returns' and 'Range', aimed at uncovering latent financial states for generating trading signals.

Signal Generation

• Trading signals are generated based on Moving Average and HMM predictions.
• A combined signal integrates both MA and HMM for decision-making.

Performance Metrics

• Cumulative returns for both the benchmark (Bitcoin) and the strategy are calculated.
• The Sharpe Ratio, assessing risk-adjusted returns, is computed for both.

Output Metrics

• Returns Benchmark: Total percentage return of Bitcoin.
• Returns Strategy: Total percentage return of the strategy.
• Sharpe Benchmark: Risk-adjusted return of Bitcoin.
• Sharpe Strategy: Risk-adjusted return of the trading strategy.

Conclusion

This project demonstrates the integration of machine learning in financial trading, particularly using HMM for cryptocurrency trading strategies.

Running the Project

1. Install Python packages: pandas, numpy, hmmlearn, yfinance, matplotlib.
2. Execute the script to perform data fetching, model training, and strategy evaluation.
3. Review the output for performance analysis.

Acknowledgments

• Data source: Yahoo Finance
• Libraries: pandas, numpy, hmmlearn, yfinance, matplotlib

License

This project is licensed under the MIT License - see the LICENSE.md file for details.