Rolling Multi-Factor Loadings (Betas)
Point-in-time rolling factor betas — how much each stock loads on market, size, value, momentum, profitability, and investment over a moving window.
Category: Factor Research
From one beta to many
A factor loading is the regression slope of a stock's excess return on a factor's return — its sensitivity to that style. Where CAPM gives a single market beta, the multi-factor model estimates a vector: R_i - R_f = alpha + beta_erp ERP + beta_smb SMB + beta_hml HML + beta_wml WML + beta_rmw RMW + beta_cma CMA + beta_liq LIQ + error.
Rolling and point-in-time
SurgeFlow re-fits the loadings on a six-month trailing daily window anchored on each month-end, across several model specifications from CAPM up to FF6/FF7. Only the investable point-in-time universe as of each anchor is included, so a loading as-of an old date uses only data available then. - Six-month rolling window - Monthly anchors - CAPM through FF7 specs - Point-in-time universe per anchor
How to read it
A loading says how much the stock historically moved per unit of factor move; high momentum and profitability loadings, for example, describe a quality-momentum stock. Betas drift, so each number is a snapshot — read it with fit quality and sample size, since a short or low-fit window is less reliable.