Building and Testing Equity Selection Rules Without Overfitting

Summary

This article outlines a process for developing equity selection rules, from finding market examples and forming a hypothesis to implementing factors, running an initial backtest, reviewing losses, and deciding whether to adjust the logic. It proposes several technical setups, including rebounds after declines, pullbacks following limit-up sessions, and rising volume paired with price gains and large-order buying. Example factors include limit-up identification, market breadth, recent strong returns, candle structure, and large-order flow.

The process emphasizes checking whether a rule still works on a separate two-year period after tuning; failure to persist is treated as evidence that the apparent result may come from overfitting. It also recommends weighing backtest results against how many stocks remain and combining validated rules into a portfolio strategy. The article refers to one successful and one failed case, but gives no detailed performance figures in the supplied text, so readers cannot independently assess the examples or conclusions.

Key ideas

• Develop candidate rules from observed market behavior, then express them as measurable factors.
• Review initial backtest losses to identify whether additional conditions have a sound rationale.
• Use separate data, including another two-year period, to check whether a tuned rule generalizes.
• Judge a rule using both its backtest behavior and the number of stocks it selects.
• The article names successful and failed cases but supplies no detailed performance results here.

Tags

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