What is look-ahead bias?

Imagine a signal that appears when a candle's high touches a level. During the candle, the high keeps changing until the bar closes. If the backtest assumes you acted on the final high — the one only known at the close — but in real time you'd have had to act mid-bar, the backtest is using future information. That's look-ahead bias.

Repainting is the usual culprit

The most common form is a repainting indicator: one whose past signals shift as new data arrives. On a historical chart it looks like the indicator called every turn perfectly. In live trading, the signal you saw vanishes or moves, because it was never final. Anything that repaints will produce a backtest you can never reproduce live.

How to avoid it

The fix is discipline about timing: act only on confirmed candle close, using only data that was final at the moment of decision. A non-repainting strategy generates the same signal in backtest and in real time, which is exactly why it can be reproduced on your own platform.

Why it matters for prop accounts

A look-ahead-biased strategy doesn't just underperform — it can blow a real account while its backtest still shows a flawless curve. On a prop evaluation with a hard drawdown limit, trusting a repainting backtest is how people fail a challenge they thought was a sure thing.