1-1 in sets. Who wins the 3rd?

Yesterday I posted a simple stat on X. I wanted to know what happens in ATP best-of-3 matches when the score is 1-1 in sets. Who wins the 3rd set more often: the player who won the 1st set, or the player who won the 2nd set?

On X I forgot to include the matches with a retirement in the 3rd set. They should count (for betting, the player who goes through wins), so here are the updated numbers:

Set 1 winner: 47.0%
Set 2 winner: 53.0%

The sample: 11,993 ATP matches that went to a 3rd set, since 2010 (best-of-3).

Btw, I made the same analysis for the WTA (10,529 matches), and the numbers are nearly identical, 46.9% and 53.1%.

So the player who won the last set has a small edge. Not huge, but with almost 12,000 matches it is not luck.

Some of you asked me a good question: what if we split it by favourite and underdog? Because maybe the set 2 winner wins more just because he is usually the better player.

Fair point. So today I posted the split. These are the updated numbers:

Set 1 winner: 47.0% overall
↑ 58.1% if he was the pre-match fav
↓ 36.4% if he was the pre-match dog

Set 2 winner: 53.0% overall
↑ 63.6% if he was the pre-match fav
↓ 41.9% if he was the pre-match dog

Fav and dog are defined by closing odds before the match.

And here is the same split for the WTA:

Set 1 winner: 46.9% overall
↑ 56.4% if she was the pre-match fav
↓ 37.2% if she was the pre-match dog

Set 2 winner: 53.1% overall
↑ 62.8% if she was the pre-match fav
↓ 43.6% if she was the pre-match dog

Same pattern as the ATP. If anything, the set 2 effect is a bit stronger in the WTA.

What does this tell us?

First: winning the 2nd set does not turn the dog into the favourite. In the ATP, when the fav wins set 1 and the dog wins set 2, the fav still wins 58.1% of the time (56.4% in the WTA). Quality matters more than "momentum".

Second, and this is the interesting part: the same type of player does better when they won the 2nd set. In the ATP, the fav wins 63.6% when he won set 2, but only 58.1% when he won set 1. More than 5 points of difference. In the WTA it is even a bit bigger: 62.8% vs 56.4%. Same for the dog, just the other way around.

And here is the key check. I looked at the pre-match odds in both groups. They are almost the same: in the ATP, the average fav had a 66.5% implied chance in one group and 66.1% in the other (65.6% and 65.5% in the WTA). So the favourites were equally strong before the match. The difference is not about who was better. It is about who won the last set.

You can call it momentum, fitness, or tactics. The numbers do not tell us why. But the effect is there, in both tours.

Of course, these numbers change a lot with the pre-match odds. For example, a fav priced 1.00-1.20 who lost set 1 should win the 3rd set much more often than these averages.

One last thing: I only have pre-match odds, not live odds. So this is not a betting system, and the live market may already price it in. But next time you see a match at 1-1, you know a bit more about what usually happens.