Iterative Feature Elimination by Model Importance in Repeated Backtests

Summary

The discussion explains how to repeatedly reduce a model’s factor list using feature importance. The example starts with 15 factors, removes the three with the lowest importance after a training and backtest run, and repeats the process until five factors remain. The suggested implementation is a loop that retrains the model and drops low-importance features at each iteration.

This is a simple form of iterative feature selection, but the answer does not specify how importance is calculated, how the backtest is structured, or how to choose among the resulting models. Repeated selection based on the same evaluation data can overfit the backtest, and importance rankings may shift as factors are removed. The post offers no code, empirical comparison, or safeguards such as a separate validation period.

Key ideas

• A loop can automate repeated model training and factor removal.
• The example removes three low-importance factors per iteration until five remain.
• Feature importance should be recalculated after each retraining step.
• Importance rankings can change as the feature set changes.
• The discussion does not describe validation safeguards against backtest overfitting.

Tags

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