Factor Regression
Factor Regression for return attribution and risk decomposition
Estimate asset or portfolio exposure to common return factors and measure the variance the model leaves unexplained.
Choose a factor model and historical window to estimate exposures such as market beta, size, value, momentum, and quality.
What the regression shows
- Factor exposures (betas) for individual assets or portfolios.
- Return variance explained by the selected factors and variance left unexplained.
- Concentration in particular factor exposures.
When to use factor regression
Compare estimated factor exposures before and after changing portfolio weights.
What to review
- The factor model and regression window.
- Whether the asset set covers the full portfolio or a subset.
- Low R-squared and unstable factor loadings.
FAQ
- Does a high R-squared mean the model is correct?
- Not necessarily. A high R-squared means the factors explain historical variance well. It does not guarantee that the same exposures persist or that the factor model is complete.