Curve Fitting (Over-Optimization)

Curve fitting, also called over-optimization, is what happens when a strategy's parameters are tuned so tightly to one stretch of historical data that they describe its random noise instead of repeatable market behaviour. The symptom is a near-perfect backtest that falls apart on new data. It usually comes from testing dozens of parameter combinations and keeping only the best one, adding rules whose only purpose is to remove individual losing trades, or optimizing over too few trades to be statistically meaningful. Common defences are keeping the rule set simple and the parameter count small, checking that results stay stable when a parameter is nudged slightly, and validating on unseen data through walk-forward analysis or forward testing.