Prediction Accuracy
After a round finishes, we score the predictions against actual fantasy points (FIFA feed) — error, ranking, scouting-bonus calibration, and how the recommended XI did. The hints feed back into tuning the model.
attacking-output multiplier the tuner learned per position (1.0 = no change)
predicted chance to clear threshold vs what actually happened
| Bucket | Predicted | Observed | n |
|---|---|---|---|
| 0-20% | 11% | 8% | 108 |
| 20-40% | 29% | 27% | 77 |
| 40-60% | 48% | 46% | 11 |
- XI actual points (capt ×2)
- 72.0
- Captain
- Ousmane Dembélé (3 pts)
- Captain rank in XI
- #8 of 11
- Best captain (hindsight)
- Lionel Messi
- Avg player (field)
- 3.22 pts
where the model flatters players — chasing differentials is only right if the low-owned buckets aren't inflated
Every squad's real score vs the hindsight-best XIs, plus every big miss classified to the model factor that caused it — so each round makes the next projection better.
- MIDCharles De Ketelaere2.9 →19
Scored 19 largely through tackle/chance volume the per-90 rates didn't credit (model expected 0.6 pts from those routes).
- MIDAzzedine Ounahi3 →16
Scored 16 largely through tackle/chance volume the per-90 rates didn't credit (model expected 0.5 pts from those routes).
- MIDJude Bellingham5 →15
Low goal prior (24%) but hauled 15 — likely a finishing spike. Check role support before chasing.
- MIDHans Vanaken4.2 →13
Low goal prior (3%) but hauled 13 — likely a finishing spike. Check role support before chasing.
- FWDErling Haaland5.1 →13
Returned 13 vs 5.1 projected — team/fixture strength for this matchup was likely underrated.
- MIDBrahim Díaz4.5 →12
Scored 12 largely through tackle/chance volume the per-90 rates didn't credit (model expected 0.5 pts from those routes).
- MIDOusmane Dembélé11.6 →3
Played but blanked (3 vs 11.6 projected) — mostly variance.
- MIDBradley Barcola8.6 →2
Ceiling pick projected 8.6 but returned 2. High-variance blank; the miss cost is amplified if captained.
- MIDVinícius Júnior8.4 →2
Ceiling pick projected 8.4 but returned 2. High-variance blank; the miss cost is amplified if captained.
- DEFAlexander Freeman5 →-1
Projected 73 minutes but recorded 0 points — benched, unused or the round hadn't scored him. Start model called this wrong.
- FWDCyle Larin5.3 →0
Projected 73 minutes but recorded 0 points — benched, unused or the round hadn't scored him. Start model called this wrong.
- DEFJorge Sánchez4 →-1
Projected 74 minutes but recorded 0 points — benched, unused or the round hadn't scored him. Start model called this wrong.
- 5×Per-90 tackle / chance-created rates (volume routes)
- 4×Start-probability model (predicted a start that didn't happen)
- 3×Variance blank — played, no returns; only act if it repeats
- 3×Ceiling pick blanked — captaincy/exposure sizing, not projection
- 2×One-off finishing spike — do NOT over-correct, check role support
- 2×Defensive scoring calibration (clean sheets, saves, conceded)
- 1×Team strength / fixture odds for this matchup
backtested 4 Aug, 08:15 UTC