WWCFxWorld Cup Fantasy Advanced Model
Matchday12345678
Backtest · Matchday 2

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.

Model calibration · auto-tunedfrom rounds 1, 2, 3, 4, 5, 6, 7, 8

attacking-output multiplier the tuner learned per position (1.0 = no change)

GK×1.30
was under-predicting
DEF×1.30
was under-predicting
MID×1.14
was under-predicting
FWD×1.22
was under-predicting
Players scored
759
Mean abs error
2.39
pts/player
Rank correlation
0.36
predicted vs actual
Bias
-1.05
under-predict
Scouting-bonus calibration

predicted chance to clear threshold vs what actually happened

BucketPredictedObservedn
0-20%10%12%542
20-40%25%42%156
Recommended XI · actual result
XI actual points (capt ×2)
35.0
Captain
Marcus Thuram (1 pts)
Captain rank in XI
#8 of 11
Best captain (hindsight)
Lucas Digne
Avg player (field)
3.16 pts
Predicted vs actual · by ownership bucket

where the model flatters players — chasing differentials is only right if the low-owned buckets aren't inflated

<1% owned-0.71
pred 1.82 · actual 2.53 · n=564
1-5% owned-1.78
pred 2.76 · actual 4.54 · n=134
5-15% owned-2.00
pred 3.08 · actual 5.08 · n=40
15%+ owned-3.53
pred 4.18 · actual 7.71 · n=21
Model tuning hints
  • Under-predicting by 1.05 pts on average — scale up.
  • GK: under-predicting by 2.47 — adjust GK rates.
  • DEF: under-predicting by 1.53 — adjust DEF rates.
Biggest prediction misses
FWDJonathan David
pred 2.9 · actual 24
FWDCody Gakpo
pred 2.8 · actual 19
FWDAyase Ueda
pred 2.5 · actual 18
GKEloy Room
pred 0.6 · actual 14
FWDIsmaïla Sarr
pred 2 · actual 15
MIDJohan Manzambi
pred 2.4 · actual 15
FWDBrian Brobbey
pred 2.6 · actual 15
DEFAlexander Freeman
pred 3.6 · actual 16
Post-round autopsy · what actually happened

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.

Maximum Points
87
capt Kylian Mbappé (13) · best: Kylian Mbappé
Balanced
76
capt Kylian Mbappé (13) · best: Kylian Mbappé
Wildcard
74
capt Kylian Mbappé (13) · best: Kylian Mbappé
Recommended squad
35
capt Marcus Thuram (1) · best: Lucas Digne
Scouting Bonus
35
capt Marcus Thuram (1) · best: Lucas Digne
Best possible XI
201
formation-legal, whole pool
Best legal squad XI
201
budget + country limits respected
Field-average XI
~37.9
random valid team benchmark
Under-projected · what we missed
  • FWDJonathan David2.924

    Model gave him only 48% to start, but he played and returned 24. The appearance model missed his role.

  • FWDCody Gakpo2.819

    Model gave him only 54% to start, but he played and returned 19. The appearance model missed his role.

  • FWDAyase Ueda2.518

    Model gave him only 48% to start, but he played and returned 18. The appearance model missed his role.

  • GKEloy Room0.614

    Model gave him only 46% to start, but he played and returned 14. The appearance model missed his role.

  • FWDIsmaïla Sarr215

    Model gave him only 47% to start, but he played and returned 15. The appearance model missed his role.

  • MIDJohan Manzambi2.415

    Model gave him only 46% to start, but he played and returned 15. The appearance model missed his role.

Over-projected · what we chased
  • DEFNathaniel Brown6.82

    Ceiling pick projected 6.8 but returned 2. High-variance blank; the miss cost is amplified if captained.

  • DEFNico Schlotterbeck5.61

    Played but blanked (1 vs 5.6 projected) — mostly variance.

  • FWDMarcus Thuram5.21

    Played but blanked (1 vs 5.2 projected) — mostly variance.

  • DEFLadislav Krejcí4.81

    Played but blanked (1 vs 4.8 projected) — mostly variance.

  • DEFJoshua Kimmich5.72

    Played but blanked (2 vs 5.7 projected) — mostly variance.

  • FWDHarry Kane5.62

    Played but blanked (2 vs 5.6 projected) — mostly variance.

Model factors behind this round's misses
  • 10×Start-probability model (appearances / rotation signal)
  • 9×Variance blank — played, no returns; only act if it repeats
  • 1×Ceiling pick blanked — captaincy/exposure sizing, not projection

backtested 4 Aug, 08:15 UTC