arrow_downward Pull to refresh
GW38 Finished

Predicted vs Actual

How Fanap's weekly point projections held up · 2025-26

Each gameweek we project points for every player with form and a fixture (PPG × fixture difficulty). After the GW we score those projections against real FPL points. MAE is the average gap between predicted and actual — lower is better. Figures cover the players we projected that week, not every player in FPL.

Season MAE

2.03

avg points off per player

RMSE

3.05

root mean squared error

Bias

-0.03

underpredicting

Coverage

GW27–38

12 GWs · 4314 scored

Headline figures exclude gameweeks with fewer than 30 scored predictions.

MAE by Gameweek

1.59
27
1.80
28
1.96
29
2.07
30
2.07
31
1.88
32
2.72
33
2.15
34
2.05
35
2.24
36
1.75
37
2.22
38

Gameweek Breakdown

GW Avg Predicted Avg Actual MAE Bias Sample
GW26 low n 0.50 0.50 0.20 0.00 2
GW27 2.29 2.27 1.59 +0.03 398
GW28 1.92 2.09 1.80 -0.17 377
GW29 2.34 2.28 1.96 +0.06 380
GW30 2.32 2.34 2.07 -0.02 388
GW31 2.32 2.30 2.07 +0.03 305
GW32 1.69 2.20 1.88 -0.51 381
GW33 3.54 2.71 2.72 +0.83 334
GW34 1.80 2.49 2.15 -0.69 232
GW35 1.93 2.30 2.05 -0.37 357
GW36 2.65 2.18 2.24 +0.46 382
GW37 1.86 2.16 1.75 -0.30 392
GW38 2.31 2.21 2.22 +0.10 388

Positive bias = overpredicting, negative = underpredicting. Greyed rows have fewer than 30 scored predictions and are excluded from the headline figures.