CAPM Analysis Project

A comprehensive Capital Asset Pricing Model (CAPM) analysis tool that performs rolling regression analysis on multiple stocks to calculate beta coefficients, R-squared values, and expected returns.

Overview

This project analyzes 10 major stocks using the CAPM framework:

• AAPL (Apple)
• JPM (JPMorgan Chase)
• XOM (Exxon Mobil)
• JNJ (Johnson & Johnson)
• WMT (Walmart)
• NEE (NextEra Energy)
• TSLA (Tesla)
• CAT (Caterpillar)
• NFLX (Netflix)
• LMT (Lockheed Martin)

Features

• Rolling CAPM Regression: Calculates beta and R² using a 126-day (6-month) rolling window
• Static CAPM Analysis: Full-sample regression for each stock
• Visualization:
  • Individual stock regression plots
  • Rolling beta and R² time series
  • COVID-19 period comparison (crisis vs recovery vs normal regimes)
  • Collective plots for all stocks
• Statistical Analysis:
  • Alpha and beta coefficients with p-values
  • R-squared values
  • Standard errors and confidence intervals
  • Expected vs realized returns

Requirements

pip install yfinance pandas numpy scipy matplotlib

Usage

Simply run the main script:

python main.py

The script will:

1. Download 10 years of historical price data for all stocks and the S&P 500
2. Calculate daily returns
3. Perform rolling CAPM regressions
4. Generate all visualizations in the plots/ directory
5. Print summary statistics and expected return comparisons

Output

The script generates three types of plots:

1. Regression Plots (plots/regression/): Linear regression scatter plots for the first 126-day window
2. Rolling Analysis (plots/rolling/): Time series of rolling beta and R² values
3. Comparison Plots (plots/comparison/): Bar charts comparing beta and R² across different market regimes

Market Regimes Analyzed

• COVID Crisis Regime: February 24, 2020 - April 30, 2020
• Post Crisis Recovery Regime: May 1, 2020 - June 30, 2021
• Normal Market Regime: All other periods

Methodology

The CAPM regression model used is:

R_i = α + β * R_m + ε

Where:

• R_i = Stock return
• R_m = Market return (S&P 500)
• α = Alpha (excess return)
• β = Beta (systematic risk)
• ε = Error term

The rolling window approach uses 126 trading days (approximately 6 months) to calculate time-varying beta and R² values.