Overfitting (curve fitting)

Tuning rules so tightly to past data that they describe history instead of the market.

Trading Conceptscurve fittingover-optimisation

Full explanation

Overfitting looks like success. The equity curve of the backtest is smooth, the win rate is high, and the parameters are oddly specific — a 37-period average, entries only between 08:14 and 09:02.

The giveaway is fragility. Shift the parameters slightly, or run the rules on a different period or pair, and the performance collapses. Live results then look nothing like the test.

The defences are simple: few parameters, a clear reason for each rule, testing on data you did not tune on, and forward testing before real money.

Why traders watch it

Overfitting is the most common reason a strategy that tested beautifully loses money from the first week it goes live.

Trading considerations

  • Be suspicious of very specific parameter values.
  • Hold back a period of data the strategy was never tuned on.
  • Fewer rules with obvious logic beat many rules with perfect history.

Educational guidance only — never a trading signal or recommendation.

Related indicators