Model Predictions
| Model | Winthrop | Marshall | Spread | Result |
|---|---|---|---|---|
| Elo | 21.6% | 78.4% | - | โ |
| Bradley-Terry | 38.6% | 61.4% | - | โ |
| Bradley-Terry Recency | 38.7% | 61.3% | - | โ |
| Margin | 30.9% | 69.1% | - | โ |
| Margin Recency | 30.7% | 69.3% | - | โ |
| Pythagorean | 49.1% | 50.9% | - | โ |
| Adjusted Efficiency | 42.0% | 58.0% | - | โ |
| Log Adjusted | 41.9% | 58.1% | - | โ |
| Points Off/Def | 32.8% | 67.2% | - | โ |
| Points Off/Def Recency | 32.5% | 67.5% | - | โ |
| Core Ensemble | 37.1% | 62.9% | - | โ |
| Recency Ensemble | 37.0% | 63.0% | - | โ |
| Dynamic Bradley-Terry | 25.1% | 74.9% | - | โ |
| Adjusted Context Blend | 6.0% | 94.0% | - | โ |
Actual: Marshall +23 (73-50). Win probabilities: out-of-sample forecasts from the weekly fit before the game, the same forecasts as the model pages.