Marking structure after the fact: the habit that ruins backtests
Posted: Sun Oct 25, 2026 9:15 am
Looking at a finished chart makes everything seem obvious. The swing high is clear, the break is clean, and the order block jumps out. The problem is that in real time none of that is confirmed. A swing high is only a swing high after price has moved away from it. A break is only valid after the candle closes. When people mark structure on a historical chart with the future visible, they unconsciously choose the swing points that make the story work. I fix this by covering the right side of the chart and moving forward one candle at a time, marking only what I could have known at that moment. It is slow, but the results are far more realistic. I also freeze my rules, such as how many candles define a swing, before testing. If my results look much worse than my screenshots suggested, that is not a failure of the method, it is a correction of bias. A strategy you cannot mark in real time is not a strategy yet, it is a hindsight drawing exercise.