Method
How to backtest an SMC strategy (without fooling yourself)
7 min read · by Jérôme Le Menn
SMC is full of convincing concepts. The problem: convincing doesn't mean profitable. Here's how to validate an SMC strategy honestly, to know whether it has a real edge — or overfits the past.
1. Formalize unambiguous rules
“I enter on a nice order block” is not testable. Specify everything: which timeframe for structure, which order-block definition, which confirmation, which stop, which target, which risk per trade.
A rule a stranger could execute identically is a testable rule. That's the starting condition.
2. Replay without cheating with the future
The most destructive bias is look-ahead: deciding by “seeing” what happened afterward. A good manual backtest reveals candles one at a time, never showing the future.
Replay bar by bar — or tick by tick for realistic fills — and log each trade as in live: entry, stop, exit, emotion.
3. Demand a sufficient sample
Ten trades prove nothing. You need dozens, ideally hundreds of trades, across several market conditions (trend, range, high and low volatility).
An edge that only exists in a strong trend isn't a robust edge: it's an observation that may not repeat.
4. Test robustness, not just profit
A rising equity curve can be luck. The deflated Sharpe (DSR), the Monte-Carlo of ruin and walk-forward measure whether your performance survives out-of-sample.
If your SMC strategy passes these tests, you may have a real edge. If not, better to know before risking a real account — or the fee of a prop firm challenge.
Don't believe it — prove it
Backtest this concept on real data, tick by tick, and get a robustness verdict. 7 days of Pro free, no card.
Start for freeFrequently asked questions
How many trades to validate an SMC strategy?
As many as possible: a few dozen at minimum, ideally hundreds, spread across different market conditions. The deflated Sharpe tells you whether the sample is conclusive.
How do I avoid overfitting?
Test on data you didn't use to build the strategy (out-of-sample / walk-forward), and beware of rules added just to make the curve look pretty.
Where can I backtest an SMC strategy?
On GetBacktest: tick-by-tick replay on real data, guaranteed anti-look-ahead, and a robustness verdict (DSR, Monte-Carlo, walk-forward) to decide objectively.
Read next