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% | 8% | 7% | 14 |
| 20-40% | 29% | 15% | 13 |
- XI actual points (capt ×2)
- 39.0
- Captain
- Kylian Mbappé (1 pts)
- Captain rank in XI
- #11 of 11
- Best captain (hindsight)
- Pedro Porro
- Avg player (field)
- 2.84 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.
- DEFPedro Porro4 →14
DEF returned 14 vs 4.0 projected — clean-sheet/save/conceded points are systematically under-modelled for DEFs.
- MIDAnthony Gordon3.6 →9
Scored 9 largely through tackle/chance volume the per-90 rates didn't credit (model expected 0.4 pts from those routes).
- DEFReece James3.8 →9
DEF returned 9 vs 3.8 projected — clean-sheet/save/conceded points are systematically under-modelled for DEFs.
- MIDEnzo Fernández2.9 →8
Scored 8 largely through tackle/chance volume the per-90 rates didn't credit (model expected 0.3 pts from those routes).
- MIDMorgan Rogers2.2 →7
Returned 7 vs 2.2 projected — team/fixture strength for this matchup was likely underrated.
- GKUnai Simón3.2 →8
GK returned 8 vs 3.2 projected — clean-sheet/save/conceded points are systematically under-modelled for GKs.
- FWDKylian Mbappé7.6 →1
Ceiling pick projected 7.6 but returned 1. High-variance blank; the miss cost is amplified if captained.
- MIDOusmane Dembélé6.5 →2
Ceiling pick projected 6.5 but returned 2. High-variance blank; the miss cost is amplified if captained.
- MIDBradley Barcola5.2 →1
Played but blanked (1 vs 5.2 projected) — mostly variance.
- MIDJude Bellingham5.5 →2
Played but blanked (2 vs 5.5 projected) — mostly variance.
- DEFMarcos Llorente4.3 →1
Played but blanked (1 vs 4.3 projected) — mostly variance.
- FWDHarry Kane5.3 →2
Played but blanked (2 vs 5.3 projected) — mostly variance.
- 6×Variance blank — played, no returns; only act if it repeats
- 4×Team strength / fixture odds for this matchup
- 3×Defensive scoring calibration (clean sheets, saves, conceded)
- 2×Per-90 tackle / chance-created rates (volume routes)
- 2×Ceiling pick blanked — captaincy/exposure sizing, not projection
- 2×Start-probability model (predicted a start that didn't happen)
- 1×Start-probability model (appearances / rotation signal)
backtested 4 Aug, 08:15 UTC