4th & Data
EN ES
← Home

Track Record

How this model has actually done, checked only against games it never trained on. Not a feeling about any one pick. A record.

63.0% overall accuracy, 854 unseen games (2023, 2024, 2025)
55.6% always picking home, the naive baseline

Is "70%" Actually 70%?

Every game here was held out of training entirely, so this checks whether a stated confidence matches how often the model was actually right, not whether any single pick landed.

coin flip 50-55%
Stated
52.4%
Actual
51.0%

200 games · gap +1.4pp

slight edge 55-62%
Stated
58.4%
Actual
60.6%

246 games · gap -2.1pp

clear edge 62-70%
Stated
66.0%
Actual
67.2%

238 games · gap -1.2pp

strong edge 70%+
Stated
76.2%
Actual
74.7%

170 games · gap +1.5pp

By Season

2023 62.8% 285 games
2024 65.3% 285 games
2025 60.9% 284 games

Predicted Score vs. the Market

Scored the same way, on the same held-out 2025 games the market's own closing lines are compared against here.

Margin

10.2 pts

model · 9.7 pts market

Total

10.7 pts

model · 10.3 pts market

The model doesn't beat the market yet on either number. Shown here plainly rather than only in a JSON file, since an honest "not yet" is still worth being able to check.

Only seasons the model never trained on count above (2023–2025 for the win model, 2025 for the score model). A hit rate on training-season games would be flattering and meaningless: the model already learned from those outcomes.