Overfitting (curve fitting)

Overfitting is tuning a strategy's settings until they fit past data, noise included. An overfit strategy looks excellent in its backtest and fails on new data. Warning signs are many settings, results that collapse when a setting moves slightly, and few trades.

Two habits guard against it. Keep the number of settings small and check that nearby values also work. Then hold back a stretch of history you never tune on, and test on it once at the end.