HISTORICAL ANALYSIS · 2026
Backtest results
Compare the model’s historical ATS rate with the break-even threshold below. Historical performance does not establish this week’s win probability.
ATS measures covering the spread. MAE is mean absolute margin error in points; lower is better. Binomial p measures evidence against a 50% cover-rate baseline, not the probability a pick will win.
Backtest results
ATS ACCURACY
—
MODEL MAE
—
MARKET MAE
—
BINOM P
—
break-even at -110: 52.4% · n= 0 graded (ATS) · n=0 projected (edges)
| Bucket | N | Accuracy |
|---|---|---|
| 0-3 | 0 | — |
| 3-7 | 0 | — |
| 7-14 | 0 | — |
| 14+ | 0 | — |
| Bucket | N | Accuracy |
|---|---|---|
| 0-3 | 0 | — |
| 3-7 | 0 | — |
| 7-14 | 0 | — |
| 14+ | 0 | — |
| Week | Matchup | Proj | Actual |
|---|