Traders often open a chart, spot a familiar pattern, and scroll backwards until they find enough winners to feel reassured. That scroll is not a backtest. It is confirmation shopping.
In our cohorts we insist on a fixed look-back window decided before the first bar is marked. If the rule set is meant for London open ranges on the FTSE 100 future, the window covers both calm summers and the autumn volatility spikes — not only the months that flatter the idea.
Invalidation must be written with the same care as entry. A breakout that never defines what kills the thesis will produce endless re-entries on the same chart. We treat every ambiguous case as a fail during training, then revisit the wording until the rule is boringly clear.
Sample size matters less than sample honesty. Twenty carefully logged instances with every miss recorded beat a hundred cherry-picked screenshots. The review clinics exist precisely to hear someone else ask: would you have taken that trade if you had not already known the outcome?
If you leave a study session feeling slightly less certain than when you arrived, the work probably succeeded. Certainty that arrives too quickly is usually the chart telling you what you wanted to hear.