Overfitting is when a strategy is tuned so tightly to past data that it captures noise instead of a real, repeatable edge. The backtest looks incredible; live trading is a disaster. It is the single biggest reason retail backtests mislead.
How overfitting happens
You add a rule, the curve improves; you tweak a number, it improves again. After enough tweaks the strategy has effectively memorised the exact history you tested on. It is no longer describing how markets behave — it is describing what already happened.
The tell-tale signs
Too many rules for too few trades. Oddly specific parameters (a 47-bar lookback). A gorgeous curve on old data. And the big one: the edge disappears on data you did not tune on.
The fix: out-of-sample and simplicity
Always hold back recent data the strategy never saw, and judge it there. Prefer fewer, robust rules over many fragile ones. Ask what real market behaviour you are capturing — if you cannot explain why the edge exists, be suspicious.
Let the test catch it for you
Bulk Quant automatically splits your data 70/30 and tells you plainly whether your edge holds out of sample or shows overfit-risk — so you find out before the market does. Test your strategy free.