The backtest has to guess. A bar records only its open, high, low and close, not the order in which the high and the low happened, so a trade whose stop and target both sit inside one bar has no known outcome.
If a bar's high reaches the target and its low reaches the stop, the data cannot say which was hit first. Some engines assume the target, some assume the stop, and some follow a fixed path through the bar. The choice is a setting, not a result, and on tight stops and targets it can decide the whole backtest. Count how many trades ended on such a bar, then score them both ways.
A strategy risks 1R with a target at 2R, so it breaks even at a win rate of 1 ÷ (1 + 2) = 33.3%. Over 100 trades, 30 reach the target cleanly, 50 reach the stop cleanly, and 20 end on a bar that touched both.
Same signals, same prices. One setting moves the result from a loss of 10R over the 100 trades to a gain of 50R.
Rescore the ambiguous trades as losses, then enter the new win rate and the payoff in the costs-in-R calculator to see where it sits against break-even. If only a few trades remain after that, the win-rate interval calculator shows how wide the uncertainty is.
The fill assumption. Look for a setting that says how the engine walks through a bar, or whether it uses intrabar data. If it is not stated, assume the optimistic order.
How tight the exits are. The closer the stop and target are to each other, relative to a typical bar's range, the more trades land on an ambiguous bar. A target smaller than one bar's range makes it common.
A lower timeframe rerun. Replaying the ambiguous bars on one-minute data, or on ticks, settles most of them. If the result holds only on the higher timeframe, it was the assumption that made the money.
The one-page Backtest Verification Checklist, free. Confirm the email and the PDF opens straight away; one worked check each Tuesday after that.
With a target of b R and a stop of 1R, break-even win rate is 1 ÷ (1 + b) and expectancy per trade is win rate × b − (1 − win rate), in R. Total result is wins × b − losses. The trade counts are an arithmetic example, not results of any strategy, and nothing here is investment advice.
Historical simulations are not live results. Nothing on this site is investment advice.