Process
How to Backtest an ICT Trading Strategy
Backtesting is the process of applying a defined trading rule set to historical market data and recording the outcomes. For ICT setups, the biggest challenge is turning discretionary chart language into rules that can be applied consistently.
1. Write the setup as rules
Specify instrument, timeframe, session, directional filter, liquidity condition, entry trigger, stop placement, target and trade-management rules. If a rule contains words like “looks strong,” define what that means.
2. Fix the sample
Choose a date range and test every eligible setup in sequence. Do not stop after a few attractive winners. A larger consecutive sample gives you more information about variation in outcomes.
3. Track more than win rate
Record number of trades, wins, losses, average R, largest losing streak, drawdown and results by session or setup type. Win rate alone does not determine whether a strategy has positive expectancy.
4. Separate development from validation
Use one historical period to develop the rules and a different period to test them. Changing rules after seeing validation results weakens the test.
5. Forward test before scaling
A historical result is not a promise of future performance. Demo or very small-risk forward testing can reveal execution, spread, slippage and discipline problems that a clean chart review misses.
Worked example (hypothetical)
Separate strategy discovery from evaluation. Use an initial sample to write the rule and resolve ambiguities, then freeze the definition before collecting a fresh out-of-sample sample. Include spread, commissions, plausible slippage, missed entries and skipped trades. Report expectancy in R, win rate, average win/loss, maximum drawdown and the number of observations. If results deteriorate on the fresh sample, record that honestly rather than adding filters until the historical chart looks better. A backtest estimates how a defined rule behaved in that sample; it does not guarantee future performance.
FAQ: How many trades are enough for a backtest?
There is no universal trade-count threshold that proves an edge. The needed sample depends on how variable the outcomes are and how narrow the strategy is. Report sample size and uncertainty rather than treating a small run as conclusive.
FAQ: Should I change rules when the backtest starts losing?
Record the failure and finish the planned test before changing rules. If you revise a setup after seeing losses, treat it as a new hypothesis and evaluate it on fresh data to reduce overfitting.
FREE LEARNING PATH
Turn this concept into a complete trading framework.
Create a free account to track your progress through the ICT Flow curriculum, then continue from the foundations into 38 modules and 203+ lessons.