Tennis stats
Objective averages and our prognosis accuracy per competition. The percentages are PROGNOSIS ACCURACY, not profitability or betting value.
The % is the model's prognosis accuracy, not a return. The history is short (since late June) and grows over time, so leagues with few matches don't yet show a percentage.
Swipe the table sideways →
| League | Matches | Games/match | Winner accuracy | Games accuracy |
|---|---|---|---|---|
| ITF / Futures / UTR | 1952 | 20.7 | 72% (70–74%)n=1952 · large sample | n=2insufficient sample |
| ATP · WTA · Challenger | 1157 | 24.4 | 68% (65–71%)n=1157 · large sample | 53% (47–59%)n=297 · large sample |
Frequently asked questions
What is surface Elo?
A rating computed separately for clay, hard, grass and carpet: a player can be much stronger on one surface than another.
What do aces and double faults measure?
Aces are winning serves (serve power); double faults are two serve misses in a row (consistency). Both help estimate games and sets.
How does BetsTalent calculate probabilities?
With an in-house model: opponent-adjusted Elo and, in football, Poisson/Dixon-Coles with per-team attack and defence. In tennis, Elo by surface. No raw averages.