The intelligence behind the oddsWeekly · 2026-09-28 to 2026-10-04
Week 2026-W40: model vs market
4 finished matches with a forecast published before kickoff. Lower Brier score and log loss are better. oddsxi-goal-model p6-v5.
Matches4
Most likely outcome right0
1X2 Brier: model / market0.958 / 0.916
1X2 log loss: model / market1.471 / 1.427
Where the model did best
| Match | League | Score | Model gave the result | Market gave the result |
|---|---|---|---|---|
| Castellon - Ceuta | Segunda Division | 1-1 | 16.8% | 14.0% |
Where the model did worst
| Match | League | Score | Model gave the result | Market gave the result |
|---|---|---|---|---|
| Las Palmas - Valladolid | Segunda Division | 2-2 | 24.0% | 29.0% |
| Sabadell - Andorra | Segunda Division | 0-0 | 25.4% | 28.4% |
| Girona - Mallorca | Segunda Division | 0-0 | 27.1% | 28.9% |
Research record
5 cases where the model disagreed with the market by 3 points or more: 2 came true; the market expected 2.0. One week says little; see the Performance Lab for the full record.
Football analysis, not betting advice.