📊 Portfolio Optimization using Genetic Algorithm (GA)

🧠 Overview

This project implements a Portfolio Optimization model using a Genetic Algorithm (GA).
It identifies the optimal allocation of assets in a portfolio to maximize return and minimize risk under realistic investment constraints.

The notebook (portfolio_optimization.ipynb) automates stock data loading, log-return computation, GA-based optimization using DEAP, and visualization of results.

---

🚀 Features

✅ Automatic CSV Data Loading — Reads all stock price files from the Stocks_Data/ folder.
✅ Realistic Log Returns — Uses logarithmic returns for stability and accuracy.
✅ Long-only Constraint — No short selling (weights ≥ 0).
✅ Max Allocation Limit — Caps maximum allocation per stock at 40%.
✅ DEAP-based GA Optimization — Modular, efficient, and customizable.
✅ Rich Visualizations:

• Portfolio Allocation (Pie Chart)
• Efficient Frontier (Return vs Volatility)
• GA Convergence Curve (Sharpe improvement over generations)
  ✅ Performance Comparison — Compares Optimized vs Equal-Weight portfolio.
  ✅ Markdown Summary — Auto-generated professional report at the end.

---

🧩 Project Structure

Portfolio_Optimization/
│
├── portfolio_optimization.ipynb   # Main Jupyter notebook
├── Stocks_Data/                   # Folder with stock CSV files
│   ├── hdfc.csv
│   ├── itc.csv
│   ├── l&t.csv
│   ├── m&m.csv
│   ├── sunpha.csv
│   └── tcs.csv
└── README.md                      # This documentation

---

⚙️ Requirements

Install required Python libraries (preferably in a virtual environment):

pip install numpy pandas matplotlib deap ipykernel

---

🧾 Workflow Summary

1. Load stock price data from CSV files.
2. Compute daily log returns and annualized mean & covariance matrix.
3. Configure GA parameters (population, generations, mutation, crossover).
4. Run Genetic Algorithm to maximize the Sharpe Ratio.
5. Apply constraints:
   • No short selling
   • Maximum 40% allocation per stock
6. Visualize results:
   • Pie chart (allocation)
   • Efficient frontier (return vs volatility)
   • GA convergence curve
7. Compare Optimized vs Equal-Weight portfolios.
8. Generate Final Markdown Summary.

---

📈 Key Outputs

• Expected Annual Return (%)
• Annual Volatility (%)
• Sharpe Ratio
• Top 3 Holdings
• Comparison Table: Optimized vs Equal-weight portfolio

---

🧠 Insights

• The optimized portfolio provides better risk-adjusted returns than equal-weight allocation.
• The Genetic Algorithm effectively enforces diversification constraints.
• Using log returns prevents unrealistic compounding errors and improves accuracy.

---

💡 Future Enhancements

• Add risk preference (λ) parameter for dynamic risk-return balancing.
• Integrate interactive sliders using ipywidgets.
• Extend to multi-objective optimization for Pareto-efficient portfolios.

---

👩‍💻 Authors

Ritu Pal

📧 Email: ritupal1626@gmail.com
🌐 GitHub: ritup04

Priyanshi Solanki

📧 Email: priyanshissolanki7@gmail.com
🌐 GitHub: Priyanshi-Solanki