What Is Backtesting?

beginner7 min read

Replaying your rules across years of data to see how they would have done. The promise and the peril.

Backtesting is replaying your trading rules across historical data to see how they would have performed. You define precise rules, run them over years of past prices, and get a simulated track record — returns, drawdowns, win rate — without risking a rupee.

Backtesting is the closest thing trading has to a laboratory — it lets you test an idea against reality before betting money, which is genuinely powerful. But it comes with a seductive peril: *it’s trivially easy to produce a beautiful backtest that means nothing.* Because you already know how history unfolded, you can (consciously or not) tune rules to fit the past perfectly — and that curve-fitted “strategy” promptly fails on the future it never saw. So a backtest is a hypothesis-testing tool, not a profit machine: a good one builds justified confidence, a naively-built one builds dangerous delusion. The entire rest of this module — and the whole next module on biases — exists because the gap between a real backtest and a fooling-yourself backtest is where almost all systematic traders lose money. Respect the tool; distrust the pretty equity curve.

The promise — test an idea on years of data, risk-free, and measure its edge before going live.
The peril — knowing the past lets you (accidentally) fit rules to it, producing great history and useless future.
The mindset — treat a backtest as evidence for a hypothesis, not proof of profit; the pretty curve is the easy part.

ExampleYou code a rule, run it over 10 years of Nifty data, and get a smooth 25% CAGR with tiny drawdowns. Exciting — but did the rule have a real reason to work, or did you tweak it until the chart looked perfect? The same data that produced the result can’t validate it; that takes the out-of-sample discipline coming next.

Key takeawayBacktesting replays your rules over historical data to estimate their edge risk-free — a powerful lab. But because you know the past, it’s easy to curve-fit a gorgeous backtest that fails live. Treat it as hypothesis-testing, not a profit guarantee; the beautiful equity curve is the easy and least trustworthy part.

FAQs

If backtests can be misleading, are they worth doing?

Absolutely — a rigorously-built backtest is essential evidence; you should never trade an untested system. The key is *how* you build it: out-of-sample testing, realistic costs, bias avoidance and robustness checks (the rest of this track) separate a meaningful backtest from a self-deluding one. The tool is invaluable; misusing it is what’s dangerous.