Portfolio Optimization Report

A clean, reproducible pipeline to compare portfolio weighting strategies (EW, IV, ERC, GMV, MSR) with multiple covariance estimators (sample, constant correlation, shrinkage). It fetches data via yfinance, generates plots/tables, and compiles a LaTeX report.

Features

• Multiple Portfolio Strategies: EW (Equal Weight), IV (Inverse Volatility), ERC (Equal Risk Contribution), GMV (Global Minimum Variance), MSR (Maximum Sharpe Ratio)
• Covariance Estimation Methods: Sample, Constant Correlation, Shrinkage, and Lower Bound estimators
• Automated Report Generation: Fetches market data, generates tables and plots, compiles LaTeX PDF
• Backtesting & Analysis: Wealth evolution, drawdown analysis, risk contributions, CAPM regression
• Asset Management: Easy cleanup of generated files

Project Structure

portfolio_opt/           # Python package
  ├─ __init__.py
  ├─ data_loader.py      # Download prices and compute returns via yfinance
  ├─ portfolio.py       # Portfolio class and plotting functions
  ├─ kit.py              # Utilities (risk metrics, covariances, stats)
  └─ table_analyse.py   # Table analysis utilities

scripts/
  ├─ generate_report.py  # Entrypoint to build results and compile LaTeX
  └─ clean.py            # Cleanup script to remove generated files

data/
  └─ FF/                 # Fama-French CSVs (e.g., FF_Monthly_Data.csv)

notebooks/
  └─ FF_Data.ipynb

main.tex                 # LaTeX report (inputs from results/ and config/)
requirements.txt

Generated folders (ignored by Git):

• results/: LaTeX tables
• config/: figures (PDF)

Requirements

• Python 3.9+
• LaTeX distribution (for PDF compilation): pdflatex

Installation

1. Create a virtual environment:

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

2. Install dependencies:

pip install -r requirements.txt

Dependencies

• pandas: Data manipulation and analysis
• numpy: Numerical computations
• matplotlib: Plotting and visualization
• seaborn: Statistical visualizations
• yfinance: Yahoo Finance API for market data
• statsmodels: Statistical modeling
• scipy: Scientific computing

Data

• Fama-French Data: Ensure data/FF/FF_Monthly_Data.csv exists (Fama-French monthly factors). Adjust path in portfolio_opt/portfolio.py if you use a different filename.
• Market Data: Downloaded automatically via yfinance API

Run

Generate Report

From the project root:

python scripts/generate_report.py

This will:

• Download market data via yfinance
• Generate figures into config/ and tables into results/
• Try to compile main.tex using pdflatex (fallback path included). If it fails, run pdflatex main.tex manually.

Clean Generated Files

To remove all generated files (tables, plots, LaTeX auxiliaries):

python scripts/clean.py

This removes:

• All files in results/ directory
• All files in config/ directory
• LaTeX auxiliary files (.aux, .log, .out)
• Compiled PDF (main.pdf)

Portfolio Strategies Explained

The project implements several portfolio optimization strategies:

1. EW (Equal Weight): Equal allocation across all assets
2. IV (Inverse Volatility): Weight inversely proportional to volatility
3. ERC (Equal Risk Contribution): Optimal diversification by equalizing risk contributions
4. GMV (Global Minimum Variance): Minimizes portfolio variance
5. MSR (Maximum Sharpe Ratio): Maximizes risk-adjusted returns

Each strategy is tested with different covariance estimators:

• Sample: Standard historical covariance
• Constant Correlation: Assumes constant pairwise correlations
• Shrinkage: Ledoit-Wolf shrinkage estimator
• Lower Bound: Constrained optimization variant

Notes

• Outputs are ignored by Git via .gitignore.
• The script enforces running from project root so relative paths resolve correctly.
• Default tickers: SPY, QQQ, EFA, EEM, TLT, IEF, LQD, GLD (can be modified in generate_report.py)

License

MIT