Why Most Backtests Lie (and How to Read an Honest One)

By Mohamed Amine Sououdi, Founder · 5 min read

A backtest is only as honest as its assumptions. Change a few quietly and you can make almost any strategy look profitable. Knowing the tricks protects you from your own optimism and from anyone selling a strategy.

Trick 1: look-ahead bias

Using data that would not have been available at the trade — entering on a candle’s close but at a price from earlier in the candle, or reacting to news before it dropped. It is the most common and most damaging distortion.

Trick 2: ignoring costs

Spread, commission and slippage hit every single trade. A strategy with a tiny per-trade edge and thousands of trades can look great gross and lose money net. High-frequency backtests that ignore costs are almost always fantasies.

Trick 3: curve fitting

Tuning parameters until the past looks perfect. The result describes history, not the future. The antidote is out-of-sample testing on data you never tuned on.

Trick 4: survivorship and cherry-picking

Showing the one market or period where it worked and hiding the rest. An honest backtest shows every year, good and bad, not just the highlight reel.

What an honest backtest looks like

Realistic costs, no look-ahead, a large sample, a year-by-year breakdown, and an out-of-sample check. Bulk Quant is built on exactly these rules — deterministic logic on real data, so the numbers cannot be inflated. See it free.

Track this automatically with NEXALIONE

Auto-sync your broker, and your profit factor, expectancy and win rate compute themselves — with an AI coach on your own setups. Free to start.

Start free

Keep reading