Trading term

What is Backtesting?

Backtesting runs a precisely defined set of trading rules over historical data to see how they would have performed on it. It is the cheapest way there is to reject a bad idea, and a poor way to prove a good one.

The mechanics are simple: define entry, exit and sizing rules precisely enough to be applied mechanically, run them across past data, and record the results. What you get is an equity curve, a win rate, an expectancy and a maximum drawdown — a rough sense of whether the idea has any edge at all.

Its real strength is elimination. Most trading ideas are simply bad, and a backtest kills them in minutes rather than months of live losses. That alone justifies the exercise. What it cannot do is prove that an idea will keep working, because the future is not a resample of the past.

The discipline that separates useful backtesting from self-deception is holding data back. Develop on one period, then test once on a period you have never looked at. If performance holds up out-of-sample, you have weak evidence of an edge. If it collapses, you fitted the noise — and if you then go back and re-tune, the held-out data is contaminated and no longer tells you anything.