2026 WNBA Model Performance Analysis
Comparing prediction accuracy across 0 games using multiple rating models.
Model Performance Leaderboard
Models ranked by AUC on 2026 games. Hover over column headers for explanations.
| # | Model | AUC | Acc | Brier | LogLoss | n | AUC 7d | Acc 7d | Brier 7d | n 7d |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Pure ELO | - | - | - | - | 0 | 0.745 | 66.7% | 0.192 | 21 |
| 2 | Bradley-Terry | - | - | - | - | 0 | 0.816 | 66.7% | 0.183 | 21 |
| 3 | Bradley-Terry Recency | - | - | - | - | 0 | 0.765 | 66.7% | 0.189 | 21 |
| 4 | Pythagorean Raw | - | - | - | - | 0 | 0.827 | 71.4% | 0.185 | 21 |
| 5 | Pythagorean Efficiency | - | - | - | - | 0 | - | - | - | 0 |
| 6 | Pythagorean Adjusted | - | - | - | - | 0 | 0.837 | 76.2% | 0.167 | 21 |
| 7 | Pythagorean Log | - | - | - | - | 0 | 0.837 | 76.2% | 0.167 | 21 |
| 8 | Margin Regression | - | - | - | - | 0 | 0.806 | 71.4% | 0.183 | 21 |
| 9 | Margin Recency | - | - | - | - | 0 | 0.827 | 71.4% | 0.186 | 21 |
| 10 | Points O/D | - | - | - | - | 0 | 0.806 | 71.4% | 0.183 | 21 |
| 11 | Points O/D Recency | - | - | - | - | 0 | 0.837 | 71.4% | 0.188 | 21 |
| 12 | Core Ensemble | - | - | - | - | 0 | 0.806 | 71.4% | 0.179 | 21 |
| 13 | Recency Ensemble | - | - | - | - | 0 | 0.816 | 71.4% | 0.179 | 21 |
| 14 | Adjusted Context Blend | - | - | - | - | 0 | 0.816 | 71.4% | 0.178 | 21 |
| 15 | Dynamic Bradley-Terry | - | - | - | - | 0 | 0.796 | 66.7% | 0.182 | 21 |
| 16 | Home Team (Baseline) | - | - | - | - | 0 | 0.500 | 52.4% | 0.255 | 21 |
| 17 | Avg Margin (Baseline) | - | - | - | - | 0 | 0.847 | 76.2% | 0.176 | 21 |
Methodology
ELO / Bradley-Terry
- ELO: Iterative updates, K=64, HCA=100
- BT: Static logistic regression on all games
- Both model win probability, not margin
- ELO updates after each game; BT fits once
Pythagorean Models
- Raw: Bill James formula (pts scored/allowed)
- Efficiency: Pace-adjusted (pts per possession)
- Adjusted: KenPom-style opponent adjustment
- Log: Log-linear multiplicative scale
Margin Regression
- Team-level ridge regression on point margin
- Linear Bradley-Terry (margin target)
- Alpha=0.05 (CV-tuned)
- Learns home advantage from data (~6 pts)
Baselines
- Home Team: Always predict home wins (60%)
- Avg Margin: Higher average margin wins
- Models should beat these to add value