The system reads match video frame by frame and turns it into numbers a coach can use — where the player stood, where every serve landed, and on which shot the point was lost.
Every frame of the broadcast is mapped to a court model: players, ball and lines become coordinates. This is what the system sees on a single serve.
Not just stats — positions, directions and patterns, reconstructed for every point of the match.
Direction and landing zones for every serve — wide, body or T — split by first and second serve, deuce and ad court.
Who, from where, with which shot. An error map with court position and shot number in the rally for every unforced error.
Where the returner stands, in metres behind or inside the baseline — tracked separately against first and second serves.
No cameras to install, no sensors on court. If you have video, you have data.
Broadcast footage, a phone on a tripod or a live stream link. Any angle from behind the court works.
The model locks onto court lines, tracks both players and the ball, and reconstructs every point in coordinates.
A coach-ready report in your language: serve maps, error patterns, return positions and a plan for the next opponent.
A fragment of a real match report — serve direction in the second set, with win rates per zone.
| Direction | Share of serves | Serves | Win rate |
|---|---|---|---|
| Widepulls returner off court | 38% | 71% | |
| Bodyjams the return | 27% | 58% | |
| Tdown the centre line | 35% | 64% |
Pay per analysed match. Every plan includes all report languages and unlimited players.
analytics.tennis was developed side by side with a professional player from the WTA top 50 and her coaching team — every metric in the report exists because a coach asked for it on court, not because it was easy to measure.
Get demo