Splitting results by hour exposes where the edge actually lives
Posted: Sat Oct 31, 2026 7:16 am
An overall profit factor of 1.3 might hide the fact that two hours produce all the gains and four hours lose money. Aggregate numbers blur the structure.
Breaking down results by hour of the day, by day of the week, and by session reveals concentration. Many strategies turn out to work only in specific windows, and trading outside them drags down the total.
Once the pattern is clear, the fix is simple: stop trading the losing hours. That sounds obvious, but it requires enough trades in each bucket to avoid reading noise. A few hundred trades per bucket would be ideal, though that is often unavailable.
Testing the filtered version on out-of-sample data confirms whether the pattern holds. If the best hours remain the best, the finding is more credible.
Time-based filters also reduce costs, since fewer trades mean less spread paid, and they free up attention for the times that matter.
Which hours produce the best results, and was trading restricted to them?
Breaking down results by hour of the day, by day of the week, and by session reveals concentration. Many strategies turn out to work only in specific windows, and trading outside them drags down the total.
Once the pattern is clear, the fix is simple: stop trading the losing hours. That sounds obvious, but it requires enough trades in each bucket to avoid reading noise. A few hundred trades per bucket would be ideal, though that is often unavailable.
Testing the filtered version on out-of-sample data confirms whether the pattern holds. If the best hours remain the best, the finding is more credible.
Time-based filters also reduce costs, since fewer trades mean less spread paid, and they free up attention for the times that matter.
Which hours produce the best results, and was trading restricted to them?