Genetic Programming for Stock and Futures Factor Discovery

Summary

This guide explains how to configure a genetic programming module to search for trading factors from supplied market data and input features. Users choose a market and data frequency, return field and horizon, time splits, fitness measure and thresholds, population size, generations, mutation probabilities, and parallel task count. The search evolves candidate expressions and outputs the selected factor formulas for later evaluation. The document describes separate handling for stocks and futures, including futures-specific fitness measures and contract selection. It says candidate factors face repeated fitness-threshold checks and cautions that strict thresholds, limited iterations, sparse futures data, inactive contracts, and position constraints can leave no usable output or produce missing fitness values. It recommends adjusting the search scope and compute effort as needed. Example workflows are referenced, but no performance results or validation evidence are reported. The guide labels this as an older implementation, so its settings and expression compatibility may not reflect current platform behavior.

Key ideas

• Genetic programming combines supplied features into candidate factor expressions through iterative search.
• Time splits and fitness thresholds are configurable parts of the factor selection process.
• The module supports stock data at daily frequency and futures data at daily or intraday frequencies.
• Futures-specific fitness measures include trading position constraints that can make qualifying candidates harder to find.
• The document describes an older platform implementation and provides no empirical performance results.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.