Factor Search: Why Optimizing Factor Combinations Is Difficult

Summary

This discussion addresses whether a trader should use hyperparameter search to find an optimal combination of existing factors. The response says that “optimal” needs a clear definition: it could mean the best backtest result or the best factor-analysis outcome. Before searching, each factor's direction must be established, including whether larger or smaller values are favorable, and the factor expressions should be consistent across candidate combinations.

The response argues that searching combinations can become impractical when each candidate uses different expressions, because comparable parallel evaluation depends on shared structure. It suggests parallel search for the optimal parameters of an individual factor as a more tractable alternative and points to an external example. The document does not provide that example's method, test results, validation protocol, or evidence that single-factor optimization generalizes. It also does not explain safeguards against selection bias or overfitting, so the recommendation is a starting point rather than a complete factor research workflow.

Key ideas

• Define whether optimality means portfolio backtest performance or factor-analysis quality.
• Establish whether each factor favors high or low values before searching.
• Consistent factor expressions make parallel comparisons more manageable.
• The response presents single-factor parameter search as more practical than searching arbitrary combinations.
• No validation results or safeguards against overfitting are supplied in the discussion.

Tags

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