A profitable backtest proves one thing: the rules generated profit on data the rules were tuned against. That's it. Forward testing — and an honest journal — is where strategies actually graduate.
What backtests are good for
Killing bad ideas fast. If a setup loses on 5 years of historical data with optimistic fills, it won't survive live. Backtests filter the obviously broken. They do not validate the surviving few.
The three lies backtests tell
- Perfect fills. You always got the exact entry price. Live, you didn't.
- No slippage on stops. Gaps don't exist in the backtest. They do in your account.
- Curve fit. Tuned to the dataset. Useless out of sample.
Forward testing in 4 steps
- Paper trade for 30 sessions. Log every trade in your journal as if real.
- Micro size live for 30 sessions at 0.1R. Real fills, real emotions, negligible damage.
- Full size only after expectancy holds across at least 50 forward trades.
- Quarterly review. Re-check expectancy by setup tag. Edge decays.
The journal closes the loop
Backtest expectancy and forward expectancy should match within a reasonable band. A 1.2R backtest expectancy that becomes 0.3R live isn't unlucky — it's curve fit, slippage, or execution gap. Your journal analytics are how you spot the gap in time to stop trading the strategy.
Backtests propose. Forward tests dispose. Journals decide.