Track record
Tennis: what the model predicted, against what actually happened
What you are looking at
Historical uses the raw model probability rather than the calibrated one, because calibration is itself derived from this history — scoring against it would be marking its own homework.
Historical — out of sample
A gap near zero means the model's probabilities mean what they say: things it calls 70% happen about 70% of the time. That is the single most important number on this site.
Calibration by confidence
| Model said | Selections | Predicted | Actual | Gap | Predicted vs actual |
|---|---|---|---|---|---|
| 0.05–0.10 | 25,336 | 7.6% | 11.9% | +4.32 | |
| 0.10–0.15 | 22,710 | 12.8% | 17.2% | +4.44 | |
| 0.15–0.20 | 18,428 | 17.6% | 23.3% | +5.73 | |
| 0.20–0.25 | 15,842 | 22.5% | 26.9% | +4.38 | |
| 0.25–0.30 | 20,314 | 27.6% | 28.4% | +0.84 | |
| 0.30–0.35 | 17,608 | 32.6% | 36.2% | +3.58 | |
| 0.35–0.40 | 18,375 | 37.6% | 39.6% | +1.99 | |
| 0.40–0.45 | 18,855 | 42.7% | 42.4% | -0.33 | |
| 0.45–0.50 | 32,458 | 48.2% | 46.0% | -2.16 | |
| 0.50–0.55 | 32,067 | 51.8% | 52.9% | +1.07 | |
| 0.55–0.60 | 17,785 | 57.2% | 56.2% | -1.04 | |
| 0.60–0.65 | 16,698 | 62.3% | 59.5% | -2.81 | |
| 0.65–0.70 | 15,218 | 67.4% | 63.8% | -3.60 | |
| 0.70–0.75 | 13,158 | 72.5% | 68.8% | -3.75 | |
| 0.75–0.80 | 10,796 | 77.6% | 71.0% | -6.53 | |
| 0.80–0.85 | 14,129 | 82.4% | 76.2% | -6.21 | |
| 0.85–0.90 | 13,069 | 87.4% | 80.4% | -7.05 | |
| 0.90–0.95 | 21,096 | 92.5% | 87.9% | -4.63 | |
| 0.95–1.00 | 36,754 | 98.4% | 96.3% | -2.16 |
By market
| Market family | Selections | Predicted | Actual | Gap |
|---|---|---|---|---|
| Games handicap | 137,637 | 59.2% | 58.7% | -0.44 |
| Total games | 117,491 | 50.4% | 50.3% | -0.03 |
| Set handicap | 26,984 | 67.5% | 66.9% | -0.56 |
| Straight sets | 17,876 | 39.3% | 40.2% | +0.82 |
| Set score | 17,721 | 25.6% | 23.7% | -1.89 |
| First set / match | 17,532 | 25.9% | 25.8% | -0.10 |
| Deciding set | 9,108 | 50.0% | 48.0% | -2.04 |
| Odd/Even | 9,108 | 50.0% | 50.0% | +0.00 |
| Total sets | 9,108 | 50.0% | 48.0% | -2.04 |
| First set | 9,107 | 50.0% | 50.0% | +0.00 |
| Match winner | 9,024 | 50.4% | 50.4% | -0.02 |
A family with a large negative gap is one the model oversells. The radar's calibration already corrects for this before ranking, but it is worth knowing which markets to trust least.
By league
| Competition | Selections | Predicted | Actual | Gap |
|---|---|---|---|---|
| WTA results-only | 238,147 | 51.7% | 51.7% | -0.05 |
| ATP results-only | 142,549 | 52.3% | 51.4% | -0.87 |
Live record
Profit at fair odds
This is not a profit claim. Every bet is settled at the model's own fair odds, where a perfectly calibrated model scores exactly zero by construction. It is a second calibration measure, and it is more sensitive than the headline gap because a longshot priced at 7% pays about 13× — so small errors on unlikely outcomes cost far more than the same error on a favourite. A negative figure alongside a near-zero overall gap points to the familiar favourite–longshot pattern: the model slightly oversells long odds and slightly undersells short ones. Real bookmaker prices are shorter than fair odds, so actual betting would do worse than this.