This is our current model's honest, continuously-updated performance across every sport — every settled bet since the 1 October 2026 model update, no cherry-picking. How the model thinks and why is on the model page.
| Sport | Bets | Win | ROI | Yield | CLV | Brier |
|---|---|---|---|---|---|---|
| Football(v2) | 159 | 51.9% | +9.5% | +2.5% | +0.9% | 0.205 |
| Hockey(v2) | 79 | 55.7% | +26.1% | +16.2% | +2.0% | 0.248 |
| Tennis(v2) | 66 | 48.9% | +14.7% | +18.5% | +0.9% | 0.233 |
| ncaaf | 45 | 57.8% | +11.4% |
If the model says 60%, about 60% of those bets should win. The closer to the diagonal, the better calibrated.
A hundred bets tell you almost nothing statistically; it gets meaningful around 500+. Judge the record over the long run, not a good or bad week.
Closing line value (CLV) is our most honest leading indicator for moneyline markets: consistently beating the sharpest books' closing line is long-term profitability, regardless of short-term variance. For totals and handicaps, realised return leads and CLV only corroborates.
The Brier score measures prediction quality (lower is better); the calibration curve shows whether 60% forecasts really land ~60% of the time. We show every settled bet — wins and losses. Past performance is not a guarantee of future results.
| +19.8% |
| +6.4% |
| 0.254 |
| mma | 40 | 37.1% | -9.5% | -20.9% | +1.4% | 0.231 |
| Basketball(v2) | 37 | 47.2% | -2.9% | -3.8% | -0.3% | 0.253 |
| nfl(v1) | 29 | 58.6% | +5.3% | +14.1% | +3.4% | 0.255 |
| Baseball(v2) | 9 | 44.4% | -3.3% | -25.8% | +0.8% | 0.276 |