Benchmarks
How models compare on backtest accuracy.
Published models use the same scoring contract: Brier score, hit rate, calibration, against-the-spread rate, and drawdown. These rows are pulled from the live model registry and graded prediction results.
Registry-backed leaderboard
Last generated: Oct 31, 2026, 3:26 AM
Sport
Benchmark summary
Leaderboard rows
12
Published model and sport rows ranked in this view.
Scored picks
12,679
Rows carrying a usable probability for benchmark scoring.
Best Brier
0.200
Lower Brier score is better. It measures probabilistic forecast error.
Best ATS
71.4%
Highest against-the-spread win rate in this filtered view; small samples vary a lot.
Sports
4
Sports represented by ranked benchmark rows.
Window
365d
Rolling lookback window used for the benchmark summary.
| Rank | Model | Sport | Brier | Spread (ATS) | Scored |
|---|---|---|---|---|---|
| #1 | Unnamed model | NFL | 0.200 | 71.4% | 42 |
| #2 | Unnamed model | NFL | 0.232 | 63.2% | 38 |
| #3 | Unnamed model | NHL | 0.237 | 63.2% | 190 |
| #4 | NHL Moneyline Classifier (Random Forest) | NHL | 0.241 | 60.2% | 244 |
| #5 | MLB Moneyline (Ridge Classifier) | MLB | 0.241 | 61.4% | 53 |
| #6 | NHL Moneyline Classifier (Random Forest) | NHL | 0.241 | 59.8% | 244 |
| #7 | Unnamed model | NBA | 0.242 | 51.1% | 66 |
| #8 | MLB Moneyline Classifier (Random Forest) | MLB | 0.243 | 58.4% | 166 |
| #9 | MLB Moneyline (TabPFN) | MLB | 0.243 | 58.4% | 166 |
| #10 | MLB Moneyline (Randomized trees) | MLB | 0.245 | 58.4% | 166 |
| #11 | MLB Moneyline Predictor (Gradient boosted trees) | MLB | 0.246 | 56.6% | 166 |
| #12 | NHL Moneyline (CatBoost) | NHL | 0.247 | 58.6% | 244 |
Metric glossary
- Brier score
- Mean squared error of probabilistic forecasts. Lower is better.
- Log-loss
- Penalizes overconfident wrong answers harder than Brier. Used for binary classifiers.
- Hit rate
- Win percentage excluding pushes. Breakeven against -110 juice is about 52.4%.
- ATS
- Against-the-spread win rate for scored spread picks.
- CLV
- Closing-line value: average edge captured against the final market price.
- Max drawdown
- Worst peak-to-trough loss in units across the scored run.
What gets audited next
- Replayable runs: every benchmark row links back to immutable model versions and scored prediction windows.
- Leakage checks: validation rejects runs that train on the same games they claim to evaluate.
- Verified badges: signed benchmark runs can be surfaced on model cards and leaderboard rows.