Model Performance
A probability model is judged by calibration over samples, not by single-match hits. Every engine change must move these numbers on out-of-sample matches — or it gets cut.
2026-27 Season · Cumulative (pregame predictions, no hindsight)
Lower log loss = better. Baseline = constant league base rates (home 43%, draw 26%, away 30%). Sample is still tiny (3 matches): treat every number here as noise until at least a half-season has accumulated.
Over 2.5 Calibration
Pregame Over 2.5 probability (same number as on each match analysis page) vs whether the final score actually had 3+ goals. Sample is still tiny (3 matches): treat band rates as noise until n ≥ 30.
| Model said | n | Model avg | Actual Over rate | Δ |
|---|---|---|---|---|
| 30–40% | 1 | 36.1% | 100.0% (1/1) | +63.9 pp |
| 40–50% | 1 | 47.6% | 100.0% (1/1) | +52.4 pp |
| 50–60% | 1 | 57.5% | 100.0% (1/1) | +42.5 pp |
Well-calibrated bands sit near Δ ≈ 0 (actual rate ≈ model average). Positive Δ = model under-called Over; negative = over-called.
Per Gameweek
| GW | Matches | Model log loss | Baseline | Δ | Picks |
|---|---|---|---|---|---|
| GW 1 | 1 | 0.8131 | 1.1819 | +0.3688 | 1/1 |
| GW 2 | 1 | 1.1009 | 1.1819 | +0.0810 | 0/1 |
| GW 3 | 1 | 0.7808 | 0.8363 | +0.0555 | 1/1 |