Feature Engineering for Quantitative Trading: Extraction, Selection, and Design
Summary
This overview explains feature engineering as the work of turning raw market and company data into useful inputs for quantitative strategies and machine-learning models. It uses a dual moving-average crossover to show that a trading rule can be understood through a constructed feature: the difference between short- and long-window averages. It also connects factor selection to strategy development and argues that the quality and presentation of inputs constrain model performance.
The practical guidance covers three tasks: extracting features from raw data, selecting features with useful variation and relevance to the target, and constructing new features from existing data. Examples include adjusting prices for corporate actions, using dimensionality reduction, comparing short- and long-term volume or price averages, and deriving volatility measures from closing prices or daily ranges. These are conceptual illustrations rather than reported trading results. The discussion emphasizes domain knowledge, but does not lay out a rigorous validation protocol; predictive usefulness and any apparent relationships still require out-of-sample testing and careful data handling.
Key ideas
- Feature engineering transforms raw data into inputs that better represent patterns a strategy can use.
- A moving-average crossover can be encoded as the difference between short- and long-term averages.
- Feature extraction, feature selection, and feature construction address distinct parts of the workflow.
- Selection should consider both variation in a feature and its relevance to the prediction target.
- Financial expertise can guide feature design, while examples in the article do not establish predictive performance.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.