📊 Portfolio Risk and Return Analysis (Python) This project evaluates the risk and return characteristics of a financial portfolio composed of JPMorgan Chase (JPM), Morgan Stanley (MS), and Bank of America (BAC), benchmarked against the S&P 500 Index.
It implements a full return and risk analysis pipeline using Python, applying concepts from modern portfolio theory, the Capital Asset Pricing Model (CAPM), and Value at Risk (VaR) frameworks.
🎯 Objective To perform a comprehensive quantitative assessment of portfolio performance by:
Calculating daily and annualized returns
Measuring volatility and downside risk
Computing risk-adjusted performance ratios
Estimating Value at Risk (VaR) and Conditional VaR
Comparing the portfolio’s performance to the market benchmark (S&P 500)
🧰 Tools & Libraries Python
yfinance – for data collection
pandas, numpy – for data manipulation and calculations
matplotlib, plotly – for visualization
🔍 Methodology Overview
- Data Collection Historical daily closing prices for JPM, MS, BAC
Benchmark: S&P 500 (^GSPC)
- Return Calculations Daily simple returns and log returns for each stock
Portfolio returns using equal weighting (⅓ each)
- Annualization Annual return using compounding for simple returns
Annual volatility using scaled standard deviation
- CAPM Metrics Beta: Measures portfolio sensitivity to market
Alpha: Measures excess return beyond market expectations
- Performance Ratios Sharpe Ratio: Return per unit of total risk
Sortino Ratio: Return per unit of downside risk
Calmar Ratio: Return over maximum drawdown
Treynor Ratio: Return per unit of systematic risk (Beta)
- Risk Measures Historical Value at Risk (VaR) – 90% confidence level
Conditional VaR (Expected Shortfall) – Average of worst 5% returns
📊 Key Results Metric Value Annualized Return 44.96% Annual Volatility 21.31% Alpha (vs benchmark) 17.97% Beta 0.896 Sharpe Ratio 1.78 Sortino Ratio 2.76 Calmar Ratio 3.37 Treynor Ratio 0.42 VaR (90%) –$12,316 Conditional VaR (5%) –$27,814
Assumes an initial portfolio value of $1,000,000 and 252 trading days/year.
📌 Additional Notes The portfolio is equally weighted; you can modify weights for custom strategies.
Analysis is based on historical data only — no forecasting or ML models are used.
Code is cleanly structured for extension into multi-asset or sectoral portfolios.
👤 Author Amit Sharma M.Sc. Mathematics | Delhi Technological University (DTU) Placement Coordinator
📧 Feel free to reach out for research roles, internships, or quantitative finance opportunities. 🔗 Connect with me on LinkedIn