How Many Trades Make a Statistically Significant Test?

Category: Strategy Backtesting & Evaluation • Article #13 • Reading time: 5 minutes

Overview

Backtest results vary by chance. Small samples (10 trades) are unreliable. How many trades needed for significance? Statistical power analysis: with 100 trades, detect 50-bps edge with 80% power. Fewer trades, less reliable. Use power curves to assess significance of backtest results.