Noise vs event overfitting

Two concrete causes of overfitting in an optimization. Noise overfitting is having too many parameters, which lets the model fit random price wiggles instead of genuine structure. Event overfitting is a single lucky one-off move (for example a parameter set that happened to be open through a Non-Farm Payrolls print) inflating one configuration’s rank purely by chance.

First used in Lesson 2.5 · The bias catalog: hunting the ways a backtest lies — the lesson that makes this term real.