The tennis model projects match winners on the ATP Tour, covering Grand Slams/Masters 1000/500-level events (The Odds API's coverage floor — smaller ATP 250 slates won't appear here even on days they're played). Held-out 2023-25 backtest: 61.8% accuracy against a 61.3% always-favorite baseline (AUC 0.612) — a thin edge, reported honestly.
Player form/rank is as of each player's last completed match in the source data, not a real-time feed. Market odds shown favorite / underdog, converted from the book's real decimal price. Informational only, not betting advice.