FF Rotation Signal

What this shows: for every setup that has ever produced an FF (original Fail + recovery Fail), the next N filled outcomes for that same setup are aggregated. Each FF event is treated independently. NoFill / Expired rows roll forward (the model didn't fire, so no trade). Score = (Post-FF WR − Baseline WR) × √Occurrences — penalises tiny samples.
Models with FF
Total FF Events
Total Occurrences
Avg Lift (sample-weighted)
Loading…