Last week we debuted skill ratings to capture the trends in a player’s ability to win points on serve and return. These serve and return skill ratings can tell us a lot about a player’s broad technical strengths and help to reveal how a player achieves success on tour. But we can go even further…
We all know that a variety of skills can be called upon in any given point on serve – from nailing a kick serve out wide to executing a driving forehand volley. A huge factor that will influence which skills are going to be needed on any particular point is whether the point is played on a first or second serve.
On a first serve, the server can push the limits of their power, whereas the receiver is generally on the backfoot with a more limited set of shots to choose from. A second serve point is a whole other ball game. On these points, servers have to maximize control and it is the receiver who can gain an upper hand quickly if they are willing to attack.
Given these different dynamics, it make sense to assess player skills on first and second serve separately. In this post, we are going to do just that by introducing first and second serve skill ratings.
Our ratings approach is the same as for overall serve and return skill, which we described last week. The key difference is that ratings on first serve (and first serve return) are based solely on point outcomes on first serve, while second serve ratings only consider points on second serve. In this way, servers earn points towards their first serve rating by outperforming expectation on first serve, while, when they don’t, their opponent gets credit on the first serve return. And the same applies to the second serve.
Having four new dimensions to evaluate serve and return skill opens up a lot of interesting possibilities for analysis. One that we at You Cannot Be Serious Stats were especially interested in was identifying serve and return “types” – that is, common patterns in the relative abilities on serve and return among pro players.
To do this, we’ve applied a clustering algorithm to the differences of each player’s rating against the average across all four skill ratings. For example, a player with ratings of 1900, 2000, 2100, and 2000 on first serve, second serve, first return, and second return, respectively, would have a mean rating of 2000 and the numbers that would enter into the clustering are -100, 0, +100, and 0. This within-player centering enables us to separate overall strength from the similarity in within-player patterns so we can identify, for example, players who all share a first serve as their best skill.
When we look at the skill clusters among the top 100 rated male players, we find six distinct types. Below is a description of each type and the percentage of top ATP players who fall into each category.
Stronger Receivers: 13%
Weaker Return: 19%
Stronger 1st Serve: 8%
Stronger 2nd Serve: 17%
Weaker 2nd Return: 29%
Stronger 1st Serve & Return: 14%
Interestingly, relative weaknesses on the return define two of the more common categories. While the group of players whose first serve is their predominant weapon are the least common. Together, this suggests that a solid first and second serve may be the most common way ATP players succeed on tour.
Figure 1. ATP player skill types based on current relative strengths on first and second serve and return.
In Figure 1, we’ve selected a sample of current players in each skill type and show their skill rating on each of the four serve and return dimensions. Again, the type clusters are based on the skill differences relative to a player’s own average, which puts the focus on relative rather than absolute ability. A number of the results are what we would expect – players like Rafael Nadal and Andy Murray are distinctive for their relative strengths on the return whereas players like Stefanos Tsitsipas, Matteo Berrettini, and Maxime Cressy stand out for their massive first serve skill.
Other results are less obvious and what makes the skill type especially insightful. This includes a relative second serve weakness for players like Novak Djokovic and Jannik Sinner. While Djokovic is above a rating of 2100 on all of his first serve and both return skills, his second serve rating is closer to 2000. For Sinner, the difference is even more stark as his current second serve rating is close to 300 points below the skill of his first serve.
Another fascinating finding is the top players who demonstrate a relative deficit on the second serve return. This group includes Denis Shapovalov and Andrey Rublev, both of whom have an approximate 100-point skill gap between their first and second return performance. That pattern could indicate a lack of effectiveness against kick serves or a suboptimal choice of aggression when receiving second serves.
On the WTA side we find a very similar set of skill types but the prevalence of the types contrasts sharply from the ATP players. Below is a description of each skill type and its frequency among the top 100 rated WTA players.
Stronger Return: 15%
Stronger 1st Serve: 19%
Stronger Serve: 17%
Stronger 1st Return: 21%
Stronger 2nd Return: 15%
Stronger 2nd Serve: 13%
The main difference that jumps out is the greater prevalence of strong return skills relative to serve. There are three categories that are defined by strengths in the return game and these collectively account for 51% of top players.
A selection of players of each skill type is shown in Figure 2. As with the men, these results are a mix of the obvious and the more surprising. Observing even a few matches would likely show you that Victoria Azarenka is especially strong on the return game and Karolina Pliskova has a vey effective first serve. But one would probably have to have a very discerning eye to observe Madison Key’s superior strength on second serve return, Shelby Rogers’s strong edge on the second serve, or Ons Jabeur’s relative advantage on first serve return.
Figure 2. WTA player skill types based on current relative strengths on first and second serve and return.
This post has highlighted just a few of the insights we can gain from skill ratings and the corresponding skill types they reveal. Many more takeaways can be found in the plots above and each finding is testament to what basic match stats can show when analyzed with a compelling method.
We are excited for what we can continue to learn as we delve further into these skill ratings and their trends over the course of the season.
I was linked to this blog by a friend who recommended it to me since I dabble in tennis stats myself (https://medium.com/@sid.ravichandran1). I must say that this is really good analysis both technically and in terms of the the topic chosen. Returns of serve kinda sail under the radar for the casual fans when compared to serves, so it's good to see the significance of the return reflected in the data. And it's a great feeling when the analysis agrees with what you observe as a tennis watching fan! Cheers