Editorial archive
Free editorial pieces on factor investing and feature selection. No paywall, no signup. Each article is written, fact-checked and corrected in the open.
Model evaluation
Leakage, lookahead and survivorship are quiet biases that inflate backtest results without announcing themselves — and most outsized returns trace back to a feature that saw the future.
Read the piece →Feature selection
Models rarely fail for lack of capacity; they fail because the wrong columns reached the training set. The selection step is where the real decision is made.
Read the piece →Factor investing
Two factors that share a label can sit on opposite sides of a trade. The sorting, weighting and neutralisation choices decide what a factor actually measures.
Read the piece →