There is a habit in watching anything competitive. You judge the opening, then you stop paying attention until the end.
A dataset of 116,382 professional tennis matches says that is backwards. The opening tells you comparatively little. A narrow window in the middle tells you nearly everything, and almost nobody is watching it.
The Opening Is Weak Information
Lose the first set and you go on to win 16.52% of the time. That is the number everyone quotes.
Split it by how the set was lost and it falls apart. Lose it in a tiebreak and you still win 21.92% of the time, across 14,265 matches. Lose it 6-0 and you win 7.18%.
So the opening result on its own is close to useless. Even the margin, which helps, only moves the answer by fifteen points.
The Middle Is Overwhelming
Now follow the same player into the second set. These are the chances of them winning the match, by the score in that set:
- Level at 1-1: 17.66%, from 63,405 matches
- One game behind at 1-2: 10.42%
- 3-1 ahead: 39.02%, from 18,412 matches
- 6-1 ahead: 58.53%
From 17.66% to 58.53% inside a single set. Eight games, and the question moves from almost settled against you to better than even.
Surviving is worth nothing. At 1-1 the player is exactly where they started. What changes everything is taking the lead, and the data says it changes fast.
Why the Mistake Is So Easy
The first set takes about forty minutes and gets all the attention. The window that actually decides things is eight games long and looks like nothing while it happens.
That shape is familiar outside sport. The opening is visible, dramatic and weakly predictive. The decisive stretch is quiet, short and arrives while everyone is still discussing the opening.
Two Things That Turn Out Not to Matter
The useful half of any dataset is knowing which splits are noise.
Surface barely moves the outcome. Grass gives 17.20%, clay 16.79%, hard courts 16.47%, a spread of three quarters of one point.
Tour hardly does better. The ATP sits at 18.21% over 10,945 matches and ITF men at 15.18% over 27,415, with four other tours in between.
Surface changes how points are won. Players hold serve 74.43% of the time on grass against 65.98% on clay, over 2,239,440 service games. It does not change who wins.
Where the Numbers Come From
Completed best-of-three singles on the ATP, WTA, Challenger and ITF circuits, 1 January 2023 to 14 September 2026. Retirements and walkovers excluded, and every figure counted from the point-by-point record, never estimated from a final score.
Best-of-five is its own population. On the ATP the comeback rate rises to 23.98% over 1,597 matches. Two more sets is more room for the better player to prove it.
The tables are at blog.livetennisapi.com/studies.json under CC BY 4.0, free to reuse with attribution. They come from the Live Tennis API, which serves live scores, fixtures and match events over JSON. The free tier does not ask for a card.
Figures as of 14 September 2026. Attribution: Live Tennis API.
