Ridge, Lasso, and Elastic Net

A quant fitting a factor-exposure regression with 40 candidate factors and 200 observations will get a least-squares fit that looks great in-sample and falls apart out-of-sample — too many free parameters chasing too little data, the overfitting failure mode named back in Topic 1. This note gives the standard fix: shrinkage, penalising large coefficients directly in the objective being minimised, trading a small amount of bias for a often much larger reduction in variance.