The accepted story about a two-set comeback is a story about endurance. The player who lost the first set digs in, survives, waits for the opponent to tire, and takes the match late. It is a good story. The record does not support it.
Across 116,382 completed professional singles matches, the player who loses the opening set wins 16.52% of the time. Follow that player game by game into the second set and it becomes clear where the match is really settled.
Surviving is worth almost nothing
Here is the chance of winning the match, for the player who lost the first set, from each position in the second:
- Level at one set each: 17.66%, from 63,405 matches
- 3-3 in the second set: 19.47%, from 33,832
- 5-5 in the second: 21.13%, from 22,506
Look at the first of those. A player who loses the first set and then fights all the way back to one set each wins 17.66% of the time. Their chance before the match started to go wrong was 16.52%.
An entire set of work, by any measure the hardest set they will play, buys them barely more than a point of improvement. Reaching 5-5 in the second is worth 21.13%, still under a quarter.
Taking charge is worth a great deal
Now the same player, ahead in the second set:
- 2-0 up: 36.80%, from 16,660 matches
- 3-1 up: 39.02%, from 18,412
- 3-0 up: 45.22%, from 8,755
- 6-4 up, taking the set: 50.16%, from 8,132
- 6-1 up: 58.53%, from 3,682
A single break of serve early in the second set roughly doubles their chances, from 16.52% to 36.80%. Two breaks, at 3-0, puts them close to even money in a match they were losing twenty minutes earlier.
And losing ground is fatal
- 1-2 down in the second: 10.42%, from 47,102
- 2-3 down: 10.19%, from 37,473
- 3-4 down: 9.51%, from 31,243
- 4-5 down: 8.17%, from 26,514
A player a break down in the second set, having already lost the first, is at roughly one chance in ten. By 4-5 it is one in twelve.
What the numbers actually say
The match turns inside about eight games of the second set, and it turns on breaking serve. Lasting longer has little to do with it. Everything after that is the scoreboard catching up with a decision already taken.
That makes a familiar claim true for the wrong reason. The second set is indeed where comebacks happen. The cause is not the opponent fading across it. The early games of that set are where the chance either arrives or disappears, and by the middle of the set it is largely settled.
A note on format
One place where endurance does show up is the longer format. In best-of-five matches on the men’s professional tour, the comeback rate is 23.98%, from 1,597 matches. The best-of-three corpus gives 16.52%. More sets mean more chances, which is exactly what you would expect. It is the only part of the endurance story the data supports.
Why these positions can be counted at all
A scoreline cannot produce any of this. It needs the state of the match recorded at every game, kept as it happened and not summarised at the end. Each state is counted once per match, so a single long set does not vote twice.
These match states were read from the Live Tennis API. It gives developers JSON access to live scores, fixtures, players and match events, with completed results and point-by-point history behind them. The figures are open at blog.livetennisapi.com/studies.json, CC BY 4.0, free to reuse with attribution.
Method: Professional singles matches completed from 1 January 2023 to 14 September 2026, best-of-three unless the best-of-five figure is named. Retirements and walkovers excluded. Each state counted once per match.
