Scoring Trading Strategies with In-Sample and Out-of-Sample Tests

Summary

The document describes a custom MetaTrader 5 Strategy Tester criterion for ranking strategies. It divides the test period into an earlier in-sample segment and a later out-of-sample segment, then calculates return, drawdown, risk-adjusted performance, trade profitability, and distribution statistics for both. It also proposes comparing the two segments with a Kolmogorov-Smirnov test and checking distribution normality with a Jarque-Bera test.

A weighted composite score combines profitability, consistency, risk-adjusted performance, and statistical quality. The stated purpose is to help compare optimization results and identify strategies whose results weaken outside the fitting period. The method requires at least 50 trades and enough history for the split; strategies that fail minimum requirements receive a rejecting score. The document gives no backtest results or evidence that the composite score predicts live performance. Its thresholds and weights are design choices, and an in-sample/out-of-sample split alone does not remove all sources of overfitting.

Key ideas

• The proposed tester criterion evaluates strategy results separately in earlier and later portions of the test period.
• It combines return, drawdown, risk-adjusted measures, profitability, and distribution statistics.
• The method uses distribution comparison and normality tests as additional checks.
• A weighted score ranks strategies across profitability, consistency, risk-adjusted results, and statistical quality.
• The approach requires a minimum trade count and sufficient history, and it provides no live-performance evidence.

Tags

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