Deep Dive Into Tennis Analytics

Lukhas

Legend
Good read. I had noticed the lack of statistics, but I'm happy to see at least an attempt to answer why that is. Combined with the Match Charting Project from Tennis Abstract, you may want to take a look at the Tennis Profiler website.

 

Keizer

Hall of Fame
This is very good. And for those of you who are dubious because this was published by Harvard (counterintuitive, but okay...), Kovalchik is one of the more prolific tennis data analysts out there, right up with Jeff Sackmann of Tennis Abstract/Heavy Topspin. Nice seeing her get more exposure.
 

MeatTornado

G.O.A.T.
The hard truth is that tennis is just an extremely simple sport. Sure, it's good to know some tenancies of opponents, but at the end of the day a tennis court is really small and you need to be prepared to cover the entire court on every shot.

There's no room for a Moneyball-like revolution because it's an individual sport. There's no plugging in different players with different play styles to achieve more productive results. It's down to the individual and there's really only 2 options of play style in tennis; back of the court or coming to the net. And due to the physical technology revolution, only one of those strategies is realistic in the modern game. I guess analytics can help you figure out how often to do it, when a surprise attack can do the most damage. But again, tennis is a simple sport and most athletes can figure that out for themselves just by feeling the flow of the game.

The high tech cameras that taught us the advantages of spin rate in pitching or launch angle in hitting is also pretty irrelevant in tennis. Everyone already knows the advantages of putting as much topspin on the ball as you can, but it's easier said than done. And I'm sure there's also data out there for how you're better off flattening your shots and going for the winner early, or going big on the 2nd serve more often. But matches aren't played on paper. No player (other than an idiot like Kyrgios or Zverev) is going to try and hit 125 mph on a 2nd serve down break point because the chart says it increases your odds of winning the point. Same reason punting hasn't gone away in football.

Analytics have ruined enough sports. I don't want tennis players to suddenly start playing like chickens with their heads cut off because they're following what the chart says instead of figuring out what's working for them that day and when to pull the trigger vs when not to.
 

Lukhas

Legend
I just noticed the ATP website now has a serve and return tracker, allowing you to see where a player prefers to serve or return, and the percentage of points won for each zone. It's not much considering the limited scope of the data (Masters 1000s since 2011, Next-Gen Finals since 2017, ATP Cup since 2019), but I guess you need to start somewhere.
https://www.atptour.com/en/stats/player-tendencies

Looking at Nadal's return stats highlights something that stats did show but wasn't necessarily common sense: the Spaniard tends to be more solid when he returns with his backhand on the first serve than with his forehand. It's something the people behind Tennis Profiler highlighted in an interview with Eurosport FR. The stat pushed out was that after studying about 8,213 points, Nadal wins 36.2% of them returning with his backhand on the 1st serve and 29.3% of them returning with his forehand. Yet, he only sees 43% of first serves going to his forehand. It's particularly true if you manage to find the wide serve to his forehand on the ad side... but Nadal wins the most points on the return when you serve in the middle of the box on the ad side, so you better be accurate if you're going straight at his forehand. :whistle:

For example, Federer seldoms go wide to Nadal's forehand on the second serve on the ad side... even though he scores a not insignificant higher percentage of points hitting his favourite trademark wide kick serve compared to going down the T. That being said, they haven't played many matches in the aforementioned tournaments, so there are very few serves represented in the tracker.

 

jmnk

Hall of Fame
While you make many good points I'm going to argue a bit.
The hard truth is that tennis is just an extremely simple sport. Sure, it's good to know some tenancies of opponents, but at the end of the day a tennis court is really small and you need to be prepared to cover the entire court on every shot.

There's no room for a Moneyball-like revolution because it's an individual sport. There's no plugging in different players with different play styles to achieve more productive results. It's down to the individual and there's really only 2 options of play style in tennis; back of the court or coming to the net. And due to the physical technology revolution, only one of those strategies is realistic in the modern game. I guess analytics can help you figure out how often to do it, when a surprise attack can do the most damage. But again, tennis is a simple sport and most athletes can figure that out for themselves just by feeling the flow of the game.

The high tech cameras that taught us the advantages of spin rate in pitching or launch angle in hitting is also pretty irrelevant in tennis. Everyone already knows the advantages of putting as much topspin on the ball as you can, but it's easier said than done.
fully agree so far.

And I'm sure there's also data out there for how you're better off flattening your shots and going for the winner early, or going big on the 2nd serve more often. But matches aren't played on paper. No player (other than an idiot like Kyrgios or Zverev) is going to try and hit 125 mph on a 2nd serve down break point because the chart says it increases your odds of winning the point.
But they should. If, for a given player, based on serve percentage and point winning percentage, one _is_ more likely to win a point by going full out even on a second serve break point down he _should_ do it. The only reason why players do not do it is psychological one - it _feels_ somewhat less bad if you lose a point after a rally then on a double fault. But statistically speaking it is, for some players, a wrong choice to not go for it.

Same reason punting hasn't gone away in football.
but it is changing, isn't it? I do not have statistical data but it does seem teams go for it om fourth down more often, no?

Analytics have ruined enough sports.
like which one?

I don't want tennis players to suddenly start playing like chickens with their heads cut off because they're following what the chart says instead of figuring out what's working for them that day and when to pull the trigger vs when not to.
well, of course. But sometimes you have two seemingly equal choices and statistics can help in those cases.
 

Lukhas

Legend
All being said, there is a player who got the memo and doesn't hesitate to target Nadal's forehand on the first serve. It's also a player who does make use of statistical analysis: Daniil Medvedev. There is very little data on his second serves (about 33 second serves were charted), but on the deuce side he tends to spread his serves fairly evenly to the T or wide to the backhand, and serves out wide to the forehand on the ad court about 60% of the time, winning almost 75% of the points there. He also avoids the body serve like the plague at least against the Spaniard.



Medvedev is also a player who used biomechanical analysis of his service motion to improve it, on the advice of his coach Gilles Cervera. They spent some time at the Ker Lann campus in Brittany to seek potential improvements to his motion. L'Équipe had an interesting article about it about a year ago. Not sure if I can share the article online (you'd need to be able to read French anyway), but I'll put a picture below.

 
Top