Why Djokovic Didn’t Overuse His Best Serve Against Sinner

I can’t trust my eyes. Not only because I have astigmatism and without glasses the shapes on the television look like Super Mario Bros, but patterns only really made sense to me if I can type them into a spreadsheet and play with them for a bit. While watching the second men’s semifinal of the Australian Open last Friday, there was something about Djokovic’s serve on the ad court that nagged at me. A few times he pounded a first serve down the T on that side, and Sinner seemed to just flail at it. I’m used to lefties like Ben Shelton using that T slice (on deuce court for him) to great effect, but you don’t see righties do that much. 

So I wanted to use a recent analysis that we did at Next Level Stats on recreational and college players’ serves to try to test out the theory that the crappy old GPUs in my brain couldn’t validate for me. And luckily, one of the fanatics who contributes to the Match Charting Project had the shot-by-shot match stats up within a day. So I did it. And I was right…and then I found something even more interesting in the men’s final.

An underrated aspect of serving is serve placement. But it’s not enough to just mix it up 50/50 since there can be very good reasons to serve more to your opponent’s backhand or maybe serve wide to open up a serve +1 winner. So it is extremely illustrative to compare if a server has a roughly equal chance of winning a point when serving to different locations. In Djokovic’s semifinal for example, if he had served 50/50 wide and down the T (let’s ignore body serves for the moment), but he won 80% of serves out wide and 0% down the T, you might tell him to stop doing that. These margins are never so clear cut in reality, and if you’re just comparing the counts from one match, often there just isn’t enough data to say anything with certainty. But that’s why there are hunches…and statistics. So I crunched the numbers.

Djokovic hit 159 total first serves (the number of second serves were too small to run these tests) against Sinner. He came pretty close to winning an equal number of his first serves on the deuce court. On the deuce court, Djokovic won pretty close to the same percent of points he served wide or T (not just service winners but won the point in any fashion). Those are the 64% and 54% in the figures above. Those differences weren’t actually significant when put through the ringer. But that ad court that I had a hunch about while watching? Hitting out wide, Djokovic won 65% of those first serves. Hitting down the T, he won 84%! Significant? Yessir. 

Funny thing is that despite a higher winning percentage hitting down the T, Djokovic hit wide more often on both courts. You and I might not notice the difference between a 65% and 84% winning percentage to different directions on just one side of the court. But I’m guessing the man who spent more than 8 freaking years! as the ATP #1 has a pretty good internal gauge for this kind of thing. So why not go to the T more on ad serves?

Assuming he knew that was working for him, maybe Sinner knew as well and was protecting that part of the serve box, so Djokovic only took those chances when he had them. Or maybe (I think more likely) Djokovic was saving those “free” points for when he needed them most. Think of Agassi’s famous story of exploiting Becker’s service tell. 

There’s some interesting work defining what are known as “clutch” points in tennis in order to highlight a player’s stats on the most important points during a match. This is intuitive–you try harder when the score is 6-6 in the third set than when it’s love-all in the first. But actually, ranking points based on a rule gives us a window into how a player handles pressure. 

I looked at clutch and break points (faced and opportunities) and found that Djokovic went down the T on the ad court in 9 out of 16 chances on clutch/break points (56%) compared to 22 out of 58 chances (38%) on non-clutch points. He won 8 of those 9 down the T, by the way. Those are too few points to run any statistical tests on. But they sure imply that he knew what he was doing.

I promised a surprise, remember? We ran the same tests on the men's final against Alcaraz, and guest who flipped the script…twice. I’ll explain that in my next post.