I love watching tennis and making observations about what a player might be doing right or wrong, but I don’t trust my own eyes. I am a skeptic by nature. Whenever I hear a commentator make an assertion about a players’ strategy or tactics, I tend to question it when they don’t have numbers to back it up. After all, the eye test is often unreliable in determining underlying patterns.
It’s undeniable that tennis statistics are stuck in the dark ages. We can find first serve percentage, win-loss percentage, tiebreak record, and percentage of points won on serve and the return from the ATP and WTA. But those numbers tell us very little about the actual course of events that leads to those points being won. They also tell us little about a players’ strength and weakness beyond serving and returning.
Here are five stats I wish the ATP and WTA would track:
1. Serve placement.
Tennis players often have a go-to serve that they feel is most comfortable, or a serve that tends to win a high percentage of points. Yet, there is very little tracking data (aside from in-match Hawk-Eye information) on tennis players’ serve tendencies. Not only would it be interesting to know that information when watching top players, but it would also be valuable to players from a scouting perspective when playing against a certain opponent.
You could take it a step further and see whether a player goes for a different serve pattern on big points (break points, deuce points, etc.). It would reveal some fascinating tendencies, I think.
To make it simple, there would be three distinctions: out wide, down the T, and body (middle of the box). Service depth would be interesting as well, if we’re discussing pipe dreams.
2. Winners broken down by forehand and backhand.
Winners are a flawed stat by nature. Many times, the actual shot hit for a winner isn’t what leads to the point being won. A lot of the work comes before the winner. Maybe a winner is just a simple put-away up at the net after a well-hit groundstroke. But if we’re going to look at winners, it would be helpful to know the breakdown of forehand vs. backhand, since there are numerous players who lack weapons on one side or the other.
These numbers are occasionally tracked on tennis television broadcasts, but they are not consistently compiled anywhere.
3. Average return depth.
Getting a serve back into play but hitting it well within the service box is a recipe for giving up easy points to an opponent. It would be great to see what kind of depth the best returners (and worst returners) are getting on their return shots. Hitting a deep return is one of the best ways to neutralize the serve. Looking at this stat could be revealing, because instead of looking at whether the player wins a return point, which is subject to many different variables, it would attempt to gauge the quality of the return itself.
4. Break chances converted.
Break points are tracked by the ATP and WTA, but those numbers can often be misleading. A player can have 15 break points in one service game, and finally break, and their break point conversion rate would be 1/15. That tells you nothing about the actual match, because they could potentially have 15 break points in five different return games and have the same break point conversion rate.
With break chance conversion rate, you would take each game a player has a break point in, and count it as one chance. So if a player has 15 break points in one game but subsequently breaks, they would be 1/1. At least to me, this is a better reflection of the match because it indicates how many games in which a player has a chance to break. Perhaps there are benefits to looking at both together, but sometimes break point conversion rate fails to tell the full story.
5. Average time to hold serve (counting time the ball is in play only).
Excluding any pre-serve rituals, I would like to know how much time it takes each player to hold serve. In this situation, you can look at time as a measure of effort expended to hold serve. Maybe Roger Federer, a player who earns many free points off his serve, takes less time to hold serve than a player like Rafael Nadal, who doesn’t win a huge number of free points, but earns an extremely high percentage of service holds based on his prowess off the ground. This stat could tell us quite a bit about the level of effort required to hold serve for different players, and the physical toll a match might take on a players’ body.
What tennis stats do you wish were tracked? Feel free to share suggestions in the comments.
Terrific piece, Amy.
With respect to No. 4, I would define a “break chance” much more narrowly, even while agreeing with your larger focus on breaking within the service game even if some break points go begging earlier in that same game.
A break “chance,” to me, is when a player gets a realistic look on a break point. When an opponent throws down an ace or a service winner, it’s not as though a player “LOST” a break point. Tennis and tennis analysis should have the wisdom needed to distinguish between “break points saved by good play from the server” and “break points genuinely squandered by the receiver.”
Yes, this would make TV graphics a bit more complicated, but really, it shouldn’t be seen as rocket science.
Postscript to all of this: If any outlet was willing to pay me for keeping one of these stats during matches, sign me up. I think a lot of readers and maybe even the Changeover staff would do the same. 🙂
I have a computer game that I play called Tennis Elbow. One of the stats that it shows is the skill level of your computer animated player. I would like a stat that shows the skill level of each player. For example, rather than showing that Serena has a great serve, assign a percentage to it, as well as her backhand, movement, forehand, net skills etc. Doing the same to her opponent will allow fans to see from the outset why a match up is the way it is, whether it is mismatched or not.
I recall at one point they had a power ranking showing the amount of power that was being generated by a player. I have seen instances where a lot of so-called power players are out hit by opponents who you would not consider to be power players. It would be good if that ranking/stat was published as well.
One of my numerous pet peeves with WTA performance statistics is that, AFAIK, only the top ten performers in each category are published on line (on the WTA Press Center page). I know internet bandwith is kind of a scarce commodity, but player’s weaknesses are just as important as their strengths. I’m sure they have the data — why not share it (at least for the top 25, 50, 100 or whatever) with their fans, as they do the rankings, which extend to upwards of a thousand players?
As to specific statistics, aces per service game would be a more meaningful statistic than total aces. Last year, going into Istanbul, Serena had 445 aces for the year, roughly twice as many as Sharapova (221); but more meaningful is the fact that Serena had accumulated her aces in 481 service games, while it had taken Sharapova 632 games. In other words, Serena had .925 aces per service game, while Sharapova only had .350 aces per service game. Of the Istanbul Eight, Kvitova was also at .35, with Radwanska next best at .248 A/SG. The WTA aces stats (for the top ten ace-hitters only) does show the number of matches, which takes us partway in the right direction, but obviously some players play a lot more games per match than others.
If they can ever up with good, consistent numbers on winners, I’d be interested in seeing the number of winners given up by the top players. I have a strong sense that Serena and Vika would be far ahead of the pack in this regard, but it would be nice to be able to quantify that feeling.
Patrick, I may be able to help you on the last point, if you’d like, with some stats of my own. 😉
Now, they are not perfect, being drawn from whatever matches I find info on (aside from the AO, Slam stats are mostly available). Also, the nature of them means they are very dependant on the opponents’ playing styles. But still, for your edification, the ratio of opponents’ winners per total points played, for the Top 10 players in the year 2013, in descending order.
Errani(24,67%)
Radwanska(22,45%)
Kerber(20,90%)
Wozniacki(18,53%)
Azarenka(15,55%)
Stosur(12,87%)
Williams(11,45%)
Kvitova(10,22%)
Sharapova(9,45%)
Li(8,28%)
Thanks, Emmanuel, for the interesting stats. But to draw meaningful conclusions one would have to know which and how many matches of each player they represent. For example, I’m surprised to find Kvitova and and Sharapova giving up fewer winners than Serena, but if the totals are drawn largely from the AO, where Serena was hobbled by a bad ankle, the numbers become more explicable.
Obviously the attacking style of Kvitove and Sharapova, as with Serena, would partly explain their strong showing (their matches are usually on their racquet), but I’m surprised to see Azarenka down in the middle of the pack and very surprised to see Radwanska badly trailing Stosur, for example. That seems very counter-intuitive. I suspect the explanation is that to get a more meaningful number we should subtract the player’s own ufe’s from the total pts, before doing the rest of the calculation. It’s difficult for the opponents of Stosur or Kvitova, for example, to accumulate a lot of winners if Sam or Petra is pre-empting that opportunity by making several dozen ufe’s per match. 🙂
Obviously, the slam totals are readily available, but where can one consistently find winners and ufe’s for other tournaments? Is it just a hit and miss proposition where you come across the stats in an article about player A’s third round match, but are not necessarily able to find data about player B’s match?
I don’t want to put you to a lot of trouble, but if you have an EASY way to send copies of your files to capa0848@aol.com I would greatly appreciate it. And if I can reciprocate in any way, please let me know. I maintain a number of files of various types, focusing primarily on won-loss records of active players, grand slam records of leading players past and present, Serena Williams career etc etc. Virtually all of my data is available on line, of course, but usually in bits and pieces.
Thanks so much,
Pat Finley
Patrick, I think for the winners stats, I think it would probably be Serena and Sharapova and perhaps Kvitova. Vika does not hit a lot of winners. She draws errors from her opponents, but even in her last match where she beat Serena, she had less winners and UFEs than serena. She was outhit 2-1 in winners I think but still won the match. Go figure.
I have been annoyed by the “breaks of serve/break points” stat since I was old enough to understand what that was (so, at least the last 20 years). It’s long been a very misleading stat. I like Matt’s idea, as it would make that particular statistic even more illustrative. Would it make tracking the statistic more intensive? Sure, but so what? Make those statisticians earn their money. 😛
I would also love to see some sort of statistic regarding winners/unforced errors hit at important points. For instance, how often did the server hit an ace on break point or set point? How often did the returner hit a winner when up break point? Or down set point? Conversely, did the server hit any DFs when faced with a break point? Did the returner make a really stupid error into the net on a very makeable forehand when they had a break point opportunity? It doesn’t have to get that specific, of course, but something like “Winners on Important Points” and “UFEs on Important Points” would help show which player was holding their nerve and which one was letting the occasion get to him or her.
I would like to see just the basic winners and unforced errors for every match, whether it really tell us something or not but may have an idea, during the season not just big events. Wish the ATP/WTA would fit that into their scoreboard. As someone posted earlier, make the stat persons earn their money.
Fantastic article. I think another stat that should be included is forced errors. Some players have style that doesn’t make them hit winners (aggressive) or wait for UEs (defensive). Rafa, for example, forces the error most points that he plays. That’s why when the stats come up he has a very low number of winners and unforced errors. Where did all his points come from then?
@ Karen Karen, you may have misunderstood me. Obviously I’d like to see winners and unforced errors for every match, but my point was that I’d also like to see winners given up (or allowed). Defense, which is based largely on anticipation and mobility, is a very important commodity in tennis, and is largely unrecognized statistically except for return of serve. A winners given up stat would, I think, go a long way toward explaining, in terms of metrics, why Azarenka and Radwanska are among the top players.