Factor Portfolio

This directory contains a Python-based factor investing portfolio system. See the factor_portfolio directory for detailed documentation and implementation.

Features

• Data Ingestion: Automated download and processing of S&P 500 stock data using yfinance
• Factor Models: Implementation of multiple factors:
  • Value (P/E and P/B ratios)
  • Momentum (trailing returns)
  • Volatility (rolling standard deviation)
  • Size (market capitalization)
• Alpha Modeling: Machine learning-based return prediction using Random Forest or Lasso regression
• Risk Modeling: Covariance estimation with shrinkage techniques
• Portfolio Optimization: Sharpe ratio maximization with constraints
• Backtesting: Rolling window backtesting framework with performance metrics
• Visualization: Interactive plots of returns, factor exposures, and portfolio weights

Dependencies

1. Factor Portfolio Website

cd factor-portfolio

2. Install dependencies:

conda env create -f environment.yml

Project Structure

factor_portfolio/
├── main.py                 # Main execution script
├── factors/               # Factor implementations
│   ├── base.py           # Base Factor class
│   ├── value.py          # Value factor
│   ├── momentum.py       # Momentum factor
│   ├── volatility.py     # Volatility factor
│   └── size.py           # Size factor
├── models/               # Machine learning models
│   └── alpha_model.py    # Alpha prediction model
├── optimization/         # Portfolio optimization
│   ├── portfolio_optimizer.py  # Portfolio optimization
│   └── risk_model.py     # Risk modeling
├── backtesting/         # Backtesting framework
│   └── backtester.py    # Backtesting implementation
└── utils/               # Utility functions
    ├── data_preparation.py  # Data processing
    └── visualization.py     # Plotting utilities

Usage

Run the main script:

python -m factor_portfolio.main

The script will:

1. Download S&P 500 data
2. Compute factor scores
3. Train the alpha model
4. Run the backtest
5. Display performance metrics
6. Generate visualizations

Performance Metrics

The system calculates and displays:

• Annual Return
• Annual Volatility
• Sharpe Ratio
• Maximum Drawdown

Extending the System

Adding New Factors

1. Create a new file in the factors directory
2. Inherit from the base Factor class
3. Implement the compute method

Adding New Models

1. Create a new file in the models directory
2. Implement the required interface (fit/predict methods)

Modifying Optimization

1. Edit portfolio_optimizer.py to add new optimization strategies
2. Modify constraints in the optimize_sharpe method

Dependencies

• numpy>=1.21.0
• pandas>=1.3.0
• yfinance>=0.1.70
• scikit-learn>=0.24.2
• matplotlib>=3.4.2
• seaborn>=0.11.1
• cvxpy>=1.1.7

License

MIT License