Glossary term
Feature Engineering
Turning raw data into useful signals a model can understand.
Feature engineering transforms raw observations into variables that may help predict returns or risk. Examples include momentum over different horizons, volatility, valuation ratios, earnings growth, liquidity measures, and sector-relative scores. Good features express a market hypothesis in a measurable way.
Example: Instead of feeding raw prices into a model, a quant might create a 12-month momentum feature, a 1-month reversal feature, and a volatility feature.