factor-models

To estimate Fama-French factor alphas, run an OLS time-series regression of your portfolio excess returns against downloaded factor returns. The model outputs annualized Jensen's alpha, factor betas, t-stats, and p-values using robust covariance matrices.

Yes, the Gibbons-Ross-Shanken (GRS) test evaluates joint pricing errors across multiple assets. It computes GRS statistics and p-values to determine if alphas from the factor regressions are collectively significantly different from zero.

Rolling factor loadings are computed by applying OLS time-series regression across a moving window over your aligned excess returns and factor data. This tracks how asset exposures to factors like momentum or size change dynamically.

Factor regressions require structured inputs: an aligned T×N matrix of portfolio excess returns and a T×K matrix of factor returns. Data is pulled via pandas-datareader or yfinance and aligned by date before estimating alpha and beta.

Yes, the OLS time-series regression computes alpha and factor betas with robust covariance estimation. This ensures reliable t-statistics and p-values for performance evaluation against academic factor models.

The best way to evaluate ETF performance is downloading Fama-French factors, aligning them with ETF excess returns, and running an OLS regression. This yields risk-adjusted alpha and factor betas to compare performance against benchmarks.