Serve statistics – Heavy Topspin

The Riddle of the Ruud-Rublev Reversal

The story of the the 2024 ATP Tour Finals was the dominance of Jannik Sinner. I’ll refer you to what I wrote after the Australian Open: Yes, Jannik Sinner Really Is This Good. He just passed the 2,300 Elo threshold, becoming only the 12th man since Rod Laver to do so. When I update the Tennis 128 in a few weeks, he’ll be on it.

For all that, I’m preoccupied with something else. On the final day of round robin play, Casper Ruud beat Andrey Rublev to secure a place in the semi-finals. It was their eighth meeting and the Norwegian’s third victory:

Notice anything odd? Take away the unfinished Australian Open tilt, and the head-to-head breaks down precisely on surface lines. Rublev has won all four encounters on clay, while Ruud has run the table on hard courts. Not just any hard courts: indoors, at the Tour Finals.

We can look to external factors to explain some of the individual results. Casper had more at stake on Friday than the Russian did, with a chance to qualify for the final four. Rublev is older and broke through correspondingly earlier, so he was the natural favorite in their early meetings. Injuries and illness may have influenced another outcome or two, even aside from the retirement in Melbourne.

Still: 0-3 in completed matches for the ball-basher on hard courts, and 0-4 for the Roland Garros finalist on clay. Something’s going on here.

Not so fast

Rublev, to be fair, is hardly a fast-court specialist. His first tour-level title came on dirt in Umag, and he picked up a Masters crown last year in Monte Carlo, on one of the circuit’s slowest surfaces. The Russian’s forehand is a weapon in any conditions, and slow courts can disguise some of his weaknesses.

On the other hand, however much Rublev likes the dirt, Ruud likes it more. Earlier this year, I quantified the notion of “surface sensitivity,” the degree to which a player’s results are influenced by court speed. Rublev scored at -2.2, indicating that he does better on slower surfaces, a bit more so than the typical tour regular. Casper was considerably further down the list, at -5.0. The rating tells us that he’s more receptive to slow courts than Pablo Carreno Busta or Jaume Munar. He’s grades about the same as Diego Schwartzman.

Maybe the oddball head-to-head is a quirk of when the pair have met? Not all hard courts play fast, and not all clay behaves like the crushed brick at Roland Garros. Here are the venues for the seven completed meetings, along with my ace-based surface speed rating for each. Ratings above 1 are faster than average, below 1 are slower:

Year  Tournament   Winner  Speed  
2024  Tour Finals  Ruud     1.36  
2023  Bastad       Rublev   0.86  
2022  Tour Finals  Ruud     1.50  
2021  Tour Finals  Ruud     1.51  
2021  Monte Carlo  Rublev   0.54  
2020  Hamburg      Rublev   0.52  
2019  Hamburg      Rublev   0.74

Hypothesis denied! The 2021 and 2022 Tour Finals were the fastest conditions of their respective years, while Monte Carlo was the slowest of the entire 2021 season. Last year’s Bastad surface was fairly neutral for a clay court, but the rest of the Ruud-Rublev showdowns took place on fast hard courts or slow clay.

We could always mark down a string of seven surface-confounding results to luck, especially when both players are capable in all conditions. But it would be far more satisfying to find an explanation that tells us something about the players and their particular skills.

Stoppable

We don’t have to look far. Here are Casper’s win rates on first and second serve points against Rublev, separated by surface:

Surface  1st W%  2nd W%  
Clay      56.8%   48.4%  
Hard      72.7%   50.7%

Everybody wins fewer first serve points on clay than on hard courts, but not like this. The average gap for top-50 players in 2024 is four percentage points–not sixteen. At tour level against the entire field, Ruud has shown an even smaller difference, winning 73.1% of hard-court first-serve points against 71.2% of those on clay.

Rublev is not a brilliant returner. He’s a serviceable one with tactics to match. He often struggles to get first serves back in play. On second serve, he’s unafraid to unleash his weapons, accepting some errors in exchange for tilting other points in his favor. On clay, he’s able to turn a few more first serves into rally openers. In 2024, his gap between clay and hard-court first-serve return points won was bigger than average. But not nearly as wide as it is against Casper.

When it works…

Ruud, like many men who have developed into strong clay-courters, doesn’t have a monster serve. He can place it, he can disguise it, and he knows how to play behind it. On a fast hard court, those skills–combined with a bit more risk-taking–can result in numbers that look more like those of a big server. Against Rublev last week, he won just shy of 80% of his serve points, supported by 15 aces.

When the Norwegian is hitting corners and the ball is skimming off a court like the speedy one in Turin, Rublev is helpless. Over his career, according to the nearly 200 matches logged by the Match Charting Project, he puts 58% of first serves back in play. Against Casper on Friday, he didn’t manage 50%–a repeat of his performance on the same court two years earlier.

Rublev’s first-serve-return struggles on hard court contrast with how he feasts on Ruud’s serve on clay. The next table shows the rate at which the Russian puts Casper’s first serves in play, as well as his win percentage when he does so:

Match                Result  1st: RiP%  RiP W%  
2024 Tour Finals RR       L      46.9%   43.3%  
2023 Bastad F             W      64.6%   61.3%  
2022 Tour Finals SF       L      45.7%   50.0%  
2021 Tour Finals RR       L      67.5%   51.9%  
2021 Monte Carlo SF       W      84.4%   59.3%  
2020 Hamburg SF           W      91.4%   59.4%

The exception to the rule here is the 2021 Tour Finals match, which ended in a third-set tiebreak. It was the closest either man has come to securing one of the matches he “should” have won. Rublev won 52.9% of total points, but Ruud served his way out of just enough jams to come through.

On the very slow Monte Carlo and Hamburg courts, Rublev’s ability to handle the Norwegian’s first serve meant that Ruud was left with no edge whatsoever. Casper might be the superior baseliner, but he started too many points at a disadvantage. The picture might look different if they played on slow clay today, since Ruud is stronger and tactically savvier on serve. But I imagine it would still be a struggle, one in which few of Casper’s service games would sail by quickly.

Pundits like to say that tennis is a game of matchups. They often overstate their case: The better player (by ranking, or Elo rating, or whatever) usually wins. When they don’t, it isn’t always because of some quirk in the head-to-head. With Rublev and Ruud, though, such a quirk dictates the results. Few men are able to erase Rublev’s advantage on a hard court as much as the Norwegian can. Casper’s first serve is rarely so ineffectual as when the Russian is waiting for it on dirt. Ruud is set for another big European clay swing next year–so long as his buddy Andrey lands in the other half of the draw.

* * *

Subscribe to the blog to receive each new post by email:

Qinwen Zheng’s Rising Tide

Since a first-round exit to Lulu Sun at Wimbledon this year, Qinwen Zheng has transformed herself from a rising prospect to a force at the top of the women’s game. The 22-year-old has won 30 of 35 matches, picking up three titles including an Olympic gold. In her first appearance at the tour finals this week, she has defeated two top-five players and earned a place in the semi-finals.

Zheng currently sits at 7th in the official rankings, equal to her career best. Her performance in Riyadh will move her up to at least sixth. Elo, a leading indicator as usual, already considers her the third-best player in the world, behind only Aryna Sabalenka and Iga Swiatek. The Chinese player is riding an astonishing 30-2 streak against everyone not named Sabalenka, so it’s hard to argue.

What has changed? Zheng has long been ticketed for big things. Her January run to the the Australian Open final indicated that she was reaching her potential. But she made only two quarter-finals in the next ten events, losing both. There were clear weaknesses in her game then. Has Qinwen 2.0 plugged those gaps?

Lifting all metrics

When I last wrote about Zheng, I referred to her serve as “under construction.” Her first serves were (and are) among the very best in the game. But she missed often, and her second serve was below average for a top-50 player.

I proposed an admittedly theoretical solution, that she could play somewhat more conservatively on the first serve, still winning plenty of points. Then she could go (relatively) bigger on seconds, trading a few more double faults for better results. The bottom line, at least according to the algorithm, was that the shift would increase her serve points won from a good 60.1% to a great 61.7%.

She hasn’t done any of that. Yet since leaving Wimbledon, she has won 63.3% of service points. That’s better than the full-season mark of anyone except Swiatek.

Qinwen found a blunter solution: She just got better at everything. Here’s an overview of her serve and return results for the two halves of 2024–up to and after Wimbledon–as well as her hard court results in 2023:

Time Span        W-L  1stIn%  1stW%  2ndW%    SPW    RPW  
2023 Hard      26-12   51.9%  74.3%  45.7%  60.5%  43.6%  
2024 1st half  19-12   51.5%  74.9%  45.5%  60.6%  42.9%  
2024 2nd half   30-5   53.8%  76.5%  47.9%  63.3%  45.9%

First serves in? Up two percentage points. First serves won? Two points. Second serves won? Two points. Return points won? Three points from the first half of 2024 to the second, even though the average surface is faster.

These are enormous shifts. 54% of first serves in still leaves her near the bottom of the table, but moving from 60.6% to 63.3% serve points won is the difference between the edge of the top ten and, as noted, number two. Key to the move is the rate of second serves won, which improved from the bottom third of tour players to the top half. On return, 42.9% to 45.9% is a jump from the bottom quartile of tour regulars to the top.

In short, Zheng went from having weaknesses to not having weaknesses. It’s never easy to divvy up the credit between player and coach, but if Pere Riba doesn’t win coach of the year, we might as well quit giving out the award.

Ratioing the tour

One of the goals of my research is to help us be more specific when we analyze players. Zheng’s various points-won rates give us a clearer view than just going goggle-eyed at a 30-5 record. But it’s tough to pinpoint a player’s improvement when she suddenly does everything better.

Qinwen’s rates of winners and unforced errors leave us in the same conundrum: She’s just gotten better by every conceivable metric. Still, I have to share. Sometimes it’s worth going goggle-eyed.

I have winner and unforced error stats for a limited subset of matches–77, in Zheng’s case–from a combination of grand slam data and the Match Charting Project. For the 58 matches through the loss at Wimbledon, she hit winners on 16.9% of points, versus UFEs on 19.2%. That works out to a ratio of 0.88, which is quite good. Commentators like to point to a 1:1 ratio as a goal, but that’s relatively rare on the women’s tour. 0.85 is usually sufficient to win a match.

Since July, the Chinese player’s W/UFE rates have basically flipped. In 18 matches worth of data, she’s hit winners on 19.3% of points, against a 17.0% unforced error rate. Those numbers are good for a ratio of 1.14. Here’s a complete list of the women who have posted better ratios this year:

1. Aryna Sabalenka

That’s it. With more complete data, it’s possible that Zheng would outscore Sabalenka, too. We have W/UFE for four of Qinwen’s five second-half losses, but only 14 of 30 wins. The sample is probably a bit biased against her.

The backhand complement

When I wrote about Zheng in January, her forehand–assessed by my Forehand Potency (FHP) metric–already ranked in the top ten among tour regulars. Her backhand remained a question mark.

If there are any specifics we can glean about the 22-year-old’s improvement, it is here. Until the beginning of this season, her backhand was, more or less, a neutral shot. The Backhand Potency (BHP) stat measures how often a shot ends the point for or against the player, as well as how often it sets up a point-ending shot shortly thereafter. In 2024, Qinwen’s backhand has been five times more effective that it was before:

Time Span  BHP/100  Negative Matches  
2021-23        1.2          13 of 32  
2024           6.3           6 of 34

In the past, the Zheng backhand cost her points–that is, it rated a negative BHP–nearly half the time, in 13 of 32 charted matches. This year, it has rarely done so. While BHP per 100 backhands (BHP/100) can’t be directly converted to a number of points per match, it’s safe to estimate that her backhand is now worth at least two or three points per match that she wasn’t winning before.

A few points per contest are enough to separate a good player from a great one. The backhand alone accounts for a big chunk of the gap between early Qinwen and the current unbeatable model.

We can even see the connection between BHP and return points won. This year, Zheng has gotten more returns in play: about one percentage point more, despite the fact that she has played Sabalenka so many times. Often, pros increase that metric by playing more conservatively. They send more balls back but do so weakly, losing most of those points. The Chinese woman, on the other hand, has also improved her win rate when she puts the return in play. That number–at least in the sample of charted matches–has risen from 53.8% to 56.8%.

Return stats aren’t all about the backhand, but when someone has as dangerous of a forehand as Zheng does, opponents make the return as much about the backhand as they can. No longer anything resembling a servebot, Qinwen threatens in more and more return games. Last year, she earned a break point in 43% of (charted) return games; this year, she is up to 49%.

Up and to the right

At the risk of repeating myself: Every trendline for the 22-year-old is headed in the right direction.

Zheng has never beaten Sabalenka, but after losing 6-1, 6-2 at the US Open, she pushed the Belarusian to three tough sets in Wuhan. She had lost five straight to Swiatek, then straight-setted her on the Parisian clay at the Olympics. Last fall, Qinwen salvaged just three games against Elena Rybakina in Beijing. This week she beat her. Zheng needed three sets to get past Jasmine Paolini last month; yesterday she allowed the Italian a measly 37 points.

The only player left in the Riyadh field that the Chinese woman hasn’t defeated is Coco Gauff. They’ve met just once, in Rome this season. Zheng won just 44% of points that day, landing an abysmal 41% of first serves. Should the pair meet to decide the season-ending championship, Gauff may still have enough of an edge to come out on top. But if we’ve learned anything from Qinwen’s four-month surge, it’s that she’s going to play a whole lot better this time.

* * *

Subscribe to the blog to receive each new post by email:

Does Mpetshi Perricard’s Backhand Even Matter?

*Giovanni Mpetshi Perricard in Basel, playing a longer rally than usual*.
*Credit: Skyscraper2010*

The story of last week’s tournament in Basel was the blistering service performance of Giovanni Mpetshi Perricard. The six-foot, eight-inch Frenchman racked up 109 aces in five matches, including more than one-third of his service points in Sunday’s final against Ben Shelton.

Mpetshi Perricard is a big server straight out of central casting. He can nail the corners at 150 miles per hour; on Sunday he hit one second serve at 146. He puts plenty of mustard on his groundstrokes as well. He often plays a high-risk brand of baseline tennis, recognizing that with a serve like his, he only needs to break once or twice–or just pick off a couple of return points in the tiebreak.

The Frenchman’s rapid rise through the ranks also fits his style. For a big server, wins can come in batches, when conditions–or, simply, tiebreak luck–are on his side. After an unexpected breakthrough on clay in Lyon, Mpetshi Perricard upset Sebastian Korda (in four tiebreaks!) and reached the second week at Wimbledon. Basel played faster than any tour event this year, and he took advantage. In between, he suffered through a 1-7 stretch in which he lost five straight tiebreaks and saw his double-fault rate balloon into double digits.

Much of Mpetshi Perricard’s future success will depend on his ability to handle these ups and downs. So far, he has struggled a bit to avoid the bad patches that spell doom for one-dimensional players. In his limited tour-level action, he has won more service points (70.2%) than anyone except Jannik Sinner. Yet five men hold more reliably than the Frenchman does, even after an unbroken week in Basel. The successes of Milos Raonic and John Isner–and even Shelton last year–come from playing better than usual under pressure, something Mpetshi Perricard has yet to consistently demonstrate.

I do love talking about servebots serving service aces. But while everybody raves about the GMP serve, I keep thinking about the backhand.

On the one hand

Mpetshi Perricard is now the fourth-highest ranked man with a one-handed backhand. His shot is nothing like the graceful, big-backswing, Federer- and Gasquet-inspired strokes of Grigor Dimitrov and Lorenzo Musetti. He often does little more than set up the racket to block the ball back. Strong as he is, the resulting flat shot can be much more than a mere defensive maneuver.

A few generations ago, it was standard to see big servers with one-handers. Think Richard Krajicek or Greg Rusedski; you might even put Pete Sampras in that category. More recently, Ivo Karlovic sported a one-handed backhand, though he mostly hit slices. Christopher Eubanks fits a broadly similar mold. Now, though, one-handers are dying breed, with just nine representatives in the top 100 of the ATP rankings.

Unlike Musetti or Stefanos Tsitsipas, Mpetshi Perricard isn’t likely to inspire the next generation of one-handed stars. No one is going to call this guy a throwback. On a good serving day, the Frenchman’s highlight reel features barely any groundstrokes at all.

What, then, do the numbers say? Is the Mpetshi Perricard backhand any good? Would he be better off with a two-hander like Raonic’s, Isner’s, or Reilly Opelka’s? Or, to return to the question I started with: For someone who specializes in ending rallies before they begin, does his backhand even matter?

Safely hidden

When the Frenchman’s game plan is working, his backhand is tucked away, out of sight. No backhands are necessary when the serve doesn’t come back, and when he controls the point, he prefers the forehand. Setting aside service returns, few players avoid their backhands as scrupulously as Mpetshi Perricard does.

The average ATPer hits 44% of their groundstrokes from the backhand side. Here are the most backhand-shy men with at least 15 matches in the Match Charting Project database, along with some other big servers of note:

Player                      BH/GS  
Ivo Karlovic                30.1%  
Jack Draper                 32.5%  
Ryan Harrison               35.2%  
Thiago Monteiro             35.5%  
Giovanni Mpetshi Perricard  35.5%  
Jaume Munar                 36.1%  
Vasek Pospisil              36.4%  
Alejandro Tabilo            37.1%  
Alexei Popyrin              37.3%  
Guido Pella                 37.4%  
Ben Shelton                 37.6%  
Maxime Cressy               37.9%  
…                                  
Christopher Eubanks         38.9%  
Matteo Berrettini           41.1%  
Milos Raonic                42.9%  
-- Average                  44.0%  
John Isner                  44.3%  
Reilly Opelka               45.6%  
Greg Rusedski               46.2%  
Nick Kyrgios                46.8%  
Pete Sampras                47.7%  
Richard Krajicek            48.3%  
Goran Ivanisevic            51.1%  
Mark Philippoussis          52.1%

Backhands per groundstroke is not the easiest stat to parse, because it is the product of so many different factors. Nearly everyone would like to keep their number low, so it’s partly a function of footwork and anticipation. (And sheer willingness to hit forehands from outlandish positions.) But it is also influenced by opponents, who will work more or less hard to find the backhand. Mpetshi Perricard’s place on this list, then, could be telling us various things. He hits his forehand when he can, and his movement is good enough to make it happen. Opponents might not be trying as hard as they could to force a backhand.

Yet another factor is how comfortable the player is with their slice. GMP hits his quite a bit, meaning that he unleashes the flat one-hander that much more rarely. The typical tour player hits their flat or top-spin backhand on 35% of groundstrokes. The Frenchman comes in at 25%, not as often as his most extreme peers, but in line with other big servers:

Player                      not-slice-BH/GS  
Ivo Karlovic                           6.1%  
Daniel Evans                          11.6%  
Milos Raonic                          20.2%  
Maxime Cressy                         20.4%  
Matteo Berrettini                     21.3%  
Grigor Dimitrov                       22.7%  
Corentin Moutet                       23.3%  
Christopher Eubanks                   23.6%  
Giovanni Mpetshi Perricard            25.2%  
Alexei Popyrin                        26.3%  
...
Bernard Tomic                         27.9%  
John Isner                            28.0%  
Ben Shelton                           28.5%  
Reilly Opelka                         31.5%  
-- Average                            34.7%

All of this is to say: Mpetshi Perricard hardly leans on the flat backhand. His serve keeps point short, and his preferences are for other shots. In the 138 points of the Basel final, he hit only 28 flat backhands, six of them on service returns.

Backhand impact

When the Frenchman is forced to hit a backhand (or chooses to–anything’s possible, I guess), the results aren’t great. When he goes for the flat backhand, he wins 43% of points, compared to a tour average of 49%. He takes more risks than his peers, but not overwhelmingly so: 9% of his one-handers end in a winner or forced error, while 12% are unforced errors. (Tour norms are 8% and 9%, respectively.)

These outcomes aren’t as extreme as his preferences. Of about 200 players with as many non-slice backhands in the MCP database, Mpetshi Perricard’s 43% comes in 21st from the bottom. Compared to other big servers, that win rate is positively respectable:

Player                      W/FE%   UFE%  inPointsWon%  
John Isner                   6.9%  12.8%         35.8%  
Milos Raonic                 7.3%  12.5%         40.4%  
Matteo Berrettini            4.8%  10.4%         42.5%  
Andy Roddick                 5.5%   7.8%         42.6%  
Felix Auger-Aliassime        6.0%   9.9%         43.4%  
Giovanni Mpetshi Perricard   8.9%  12.6%         43.5%  
Ben Shelton                  6.2%  11.1%         43.8%  
Nick Kyrgios                 7.9%  10.7%         43.9%  
Kevin Anderson               7.6%  11.0%         44.1%  
Hubert Hurkacz               6.2%  10.0%         45.6%  
-- Average                   7.3%   9.0%         48.6%  
Jack Draper                  6.5%   5.7%         49.1%

Against this group, the Frenchman’s winner (and forced error) rate really stands out. Given the outcomes when he doesn’t go for it, it’s possible he should be even more aggressive than he already is. Master tactician Milos Raonic took a similar tack, piling up as many unforced errors as GMP does, but without quite as many winners.

The picture is less rosy when we look at slice backhands. As noted, Mpetshi Perricard hits a lot of them–close to one-third of his groundstrokes from that wing. When he does, he wins 33% of points, compared to 42% for his peers. Only a handful of players have posted such low slice-backhand win rates, and they are mostly the names you would expect:

Player                      W/FE%   UFE%  inPointsWon%  
John Isner                   1.5%  11.1%         27.3%  
Christopher Eubanks          1.6%  10.7%         30.2%  
Kevin Anderson               3.2%   6.5%         31.4%  
Nicolas Jarry                4.0%  13.2%         32.7%  
Ivo Karlovic                 3.5%  12.0%         32.8%  
Giovanni Mpetshi Perricard   3.1%   3.9%         33.1%  
Ben Shelton                  2.0%   7.1%         33.3%  
...
Nick Kyrgios                 4.2%   6.1%         35.6%  
Felix Auger-Aliassime        3.6%   8.2%         38.1%  
Hubert Hurkacz               2.7%   3.9%         38.1%  
Milos Raonic                 4.7%   9.4%         40.3%  
Matteo Berrettini            2.8%   9.0%         41.5%  
-- Average                   3.3%   5.4%         41.6%

The Frenchman doesn’t miss much. Why just keep the ball in play, though, if you’re likely to lose the point anyway? By hitting so many slices, Mpetshi Perricard makes his flat-backhand numbers look better, but he probably doesn’t pick up any points by making the trade. Prolonging the point is a good strategy if you’re Casper Ruud–or, really, about 80% of the guys on tour. But if you play like GMP, it’s better to go big.

This is one way in which the one-hander may cost him. The two-handed backhand is particularly valuable in its ability to block overpowering shots without retreating to a fully defensive mode. While players with one-handers try to achieve the same thing with a slice, the stats tell us that it’s a poor imitation. The Frenchman’s in particular isn’t doing him any favors.

So, does it even matter?

Mpetshi Perricard doesn’t hit that many backhands, and he isn’t that much worse than average when he does. But, the margins in tennis are small, and the margins for big servers are smaller still. In 26 tour-level matches this year through the Basel final, GMP won exactly 50% of his points. (Not 50.1%, not 49.9%–50% on the dot.) Five players in the top 20 win 50.8% or less. That’s how close the Frenchman is to an even bigger breakthrough.

My backhand potency (BHP) stat quantifies the impact that each player’s (non-slice) backhands have on their broader results. The stat measures how often a shot ends the point in either direction, as well as what happens on the shot after that. Based on the matches we’ve charted this year, GMP’s BHP per 100 backhands stands at -4.3, one of the lower numbers on tour for players with at least 10 charted matches from the last 52 weeks:

Player                      BHP/100  
Nicolas Jarry                  -6.9  
Felix Auger Aliassime          -4.6  
Tallon Griekspoor              -4.6  
Flavio Cobolli                 -4.3  
Giovanni Mpetshi Perricard     -4.3  
Alexei Popyrin                 -4.2  
Dominic Thiem                  -4.1  
Botic Van De Zandschulp        -3.5  
Matteo Berrettini              -3.1  
Ben Shelton                    -2.4  
Stefanos Tsitsipas             -2.4

What does this mean for the bottom line? -4.3 BHP is equivalent to about -3 points per 100 backhands. Since he doesn’t hit many backhands, that’s about -1.1 per 100 points.

My best estimate, then, is that if we magically replaced the Frenchman’s backhand with a neutral one–say, that of Arthur Fils–he’d pick up 1.1 more points per 100. Instead of winning 50% of points at tour level, he’d win 51.1%. That isn’t good enough to crack the top 10, but it would probably get him into the top 20.

Quantifying the impact of slices is tougher, because the more conservative shot is less likely to end the point immediately, or even on the next shot. If we figure that Mpetshi Perricard’s slice is roughly the same distance below average as his flat backhand, that’s another 0.5 or 0.6 points per 100 he could gain by acquiring a tour-average shot. Daniil Medvedev has hung in the top five in the ATP rankings while winning 51.9% of points. Stringing all of these assumptions together, we can start to see how a capably-backhanded GMP could reach that level.

The bad news for the Frenchman is that climbing the ranks is hard. Mpetshi Perricard is the worst returner in the top 50, and it isn’t even close. He breaks in about 10% of his return games; no one else is below 14%. Earlier this year, I wrote about the similar challenges facing Ben Shelton: Historically, a lot of players have arrived on tour with big serves, huge potential, and tons of hype. Few of them have been able to shore up their weak points enough to crack the top ten, let alone achieve greater feats.

The good news: There is so much room for improvement. Even without polishing the strokes themselves, it’s possible that a more aggressive set of tactics could win him a few more points on return. In yesterday’s loss to Karen Khachanov, the Frenchman won the first set despite picking off just two of 37 return points. One-dimensional servebot or not, he can learn to do better than that.

* * *

Subscribe to the blog to receive each new post by email:

Matteo Berrettini and the Pursuit of Expected Value

*Matteo Berrettini hitting a forehand that will probably end the point*

A few weeks ago, Agustin Lebron made a broad claim:

Most strategic improvements in sports have been in the direction of increasing variance and living with the (better EV) results:
Baseball: more extra base hits, no more bunting.
Football: more passing game, going for it on 4th.
Golf: driver ball speed increases.
Bball: 3 pointer
Tennis: bigger serves/groundstrokes.
Snooker: cannoning the pack to extend breaks.
Chess: sub-optimal but niche exploitive lines.

“EV” means expected value or, roughly speaking, probability of success. Thanks to baseball’s sabermetric revolution and its influence on other sports, we better understand how players and teams win. Competitors, knowingly or not, are chasing EV.

Lebron’s claim is a bit more specific, that players and coaches across sports are playing riskier games because the ultimate payoff is greater. In baseball, a sacrifice bunt makes it more likely that a team scores one run, but less likely that the team piles on multiple runs. Hoopsters land two-point shots at a better rate than three-pointers, but the additional point makes the tradeoff worthwhile. Ice hockey coaches pull the goalie sooner than they used to, taking the chance that an extra skater will result in a game-tying goal, even if the decision could result in an easy score for the opposition.

Many more examples and counter-examples appear in the discussion around Lebron’s tweet and in the comments at Marginal Revolution.

It’s less clear that tennis ought to be grouped with these other sports. It may be true that players have uniformly bigger serves, or that they hit forehands harder than their predecessors. But is their pursuit of expected value causing them to take more risks?

Serve trends

Since 1991, when the ATP started recording stats like aces and double faults, aces have indeed gone up. That first year, about 5% of points ended with an untouched serve. The tour reached 7% by 2000, then cleared 8% in 2014. The rate has held fairly steady since then, sitting at 7.7% in 2024.

One way to hit more aces is to push closer to the edge. Aim for the lines, smack it as hard as you can, and accept that you’ll miss more, too. That would fit nicely with the increasing-variance hypothesis. But that’s not what has happened. As aces have gone up, the percent of first serves made has also risen:

The increasing-variance hypothesis holds for 1991-2000: Aces went up at the cost of fewer first serves in the box. Since then, though, players have kept hitting more aces (if only slightly), while landing even more first serves.

This is almost definitely thanks to better racket and string technology. You can swing harder than ever, with more spin than ever, and keep control of the ball. But that isn’t the whole story. For any given level of technology, players could take more risks, cracking still more aces at the expense of fewer first serves in. For nearly a quarter-century, that is not the decision pro men have made.

Second serves and double faults tell the same story:

Second serves offer an opportunity to take even more risks. If you go big and miss, you lose the point. But men have generally opted to take their chances with a ball in play. Double faults have cratered since the mid-90s and are currently at an all-time low. Yet players are winning about as many second serve points as ever.

In the last decade, we’ve seen a few players–Nick Kyrgios and Alexander Bublik come to mind–who do sometimes take their chances with a big second serve. Across 129 charted matches, Kyrgios hits aces on 4% of his second serves. Tour average is below 1%. Even Nick, though, doesn’t think it’s worth a major change in his risk profile. His career double fault rate is 4.2%: above average, but hardly an outlier among tour regulars.

The Aussie recognizes that the second serve evolved for a reason. Hitting a big second serve–deploying, in other words, two first serves–is a negative-EV play. It may be worth trotting out for variety’s sake, but not more than that.

Which brings us, finally, to Matteo Berrettini. The six-foot, five-inch Italian is the apotheosis of the big-serve, big-forehand, “plus-one” game. He’s the sort of player Lebron might have been thinking of when he bucketed tennis with the other sports on the list.

For all his power from the line, Berrettini is as conservative as they come. His career ace rate of 12.3% is outstanding, yet there is no apparent cost. He makes almost 64% of his first serves. He wins more second serve points than average, too, despite a miniscule double-fault rate of 2.4%. His game has gotten even safer as he reaches his late 20s. This year, he is hitting slightly more aces (12.8%) and landing far more first serves (68.6%) and committing fewer double faults (2.0%).

If the Italian is any indication, tennis is moving toward bigger serves and forehands. Yet when it comes to the serve, variance is headed in the opposite direction.

Rally aggression

What about groundstrokes? Nearly everyone these days talks about plus-one tennis. The serve–when it doesn’t end the point outright–generates opportunities to put the ball away. When those opportunities appear, don’t screw around! The strategy looks different in the hands of Berrettini than it does with, say, Jelena Ostapenko, but more than ever, players think in terms of recognizing and converting opportunities to end points.

Once the serve has landed, some players have indeed adopted a higher-variance approach that is probably unprecedented. Ostapenko, the freest swinger of all, ends nearly two-thirds of points on her own racket. Inevitably, she misses a huge fraction of those. Her Rally Aggression score of 182 (on a scale designed to run from -100 to 100) leads active players, and it massively outstrips anyone who started their career before about 2005.

Here, alas, we are hamstrung by data limitations. I discussed the men’s tour above because women’s ace and double-fault data only goes back to 2010 or so. The situation is even worse with groundstrokes. While the Match Charting Project now spans over 14,000 matches, relatively few of those predate 2010. Those “early” matches are heavily skewed toward a handful of top players.

We can still draw some comparisons. Lindsay Davenport and Maria Sharapova, often-erratic free swingers a generation or two before Ostapenko, grade out with Rally Aggression scores in the mid-40s. That’s below Iga Swiatek. Let that sink in for a moment. Today’s rock-solid, heavy-topspinning queen of clay plays as aggressively as two earlier-era emblems of high-risk slugging.

Again, we see the effect of better tech. When Ostapenko swings away, there is perhaps a 60% chance it lands in. If it does, it probably isn’t coming back. When Davenport (or to a greater extent, her own predecessors) took a big cut with a 60/40 chance of falling between the lines, it wasn’t quite as hard, and it didn’t have as much spin. It was that much less likely to end the point immediately, or in her favor at all. The chance of an error was always high; modern rackets and strings have upped the odds that the risk is worth taking.

Berrettini, though, once again illustrates that the risk isn’t necessary. The Italian’s Rally Aggression score is 24: above average but not by much. In part the number is low because he struggles to create opportunities on return (or when his serve fails to create chances), but in part he rates where he does because he doesn’t often miss. Roger Federer, for broadly similar reasons, is in the same range.

Modern tech allows players to hit as many winners as ever with less risk. Jannik Sinner, with his career Rally Aggression score of -24 and Carlos Alcaraz, at +8, point toward a lower-variance future, at least in the men’s game.

Ebbs and flows, serves and volleys

The biggest gap in the increasing-variance hypothesis is that it doesn’t explain the death of the serve-and-volley.

Few tactics in any sport are higher variance than old-school, rush-the-net-on-every-point serve-and-volleying. Think of Boris Becker at Wimbledon. He hit a bomb, and if it came back, he was often sprawled across the court simply trying to get a racket on the ball. Today’s net forays aren’t always so kinetic, but they remain high-risk. For every easy volley, there’s an untouchable passing-shot winner.

What’s more, the most dedicated form of serve-and-volleying, Jack Kramer’s “Big Game,” was explicitly an EV play, the brainchild of an actual engineer decades before anyone thought to put “sports” and “analytics” in the same sentence. Kramer and club-mate Cliff Roche worked out the angles and the probabilities, and the on-court results were so overwhelmingly positive that other Americans quickly followed suit. Thanks to a Davis Cup drubbing in 1946, Kramer’s game also changed the course of Australian tennis, inspiring Frank Sedgman and indirectly defining the style of innumerable hopefuls, including Rod Laver.

Serve-and-volleying, in the right hands, was the smart play for reasons that no longer persist. Returners couldn’t do much with a good serve. Court conditions made baseline tennis chancy: Much more tennis was played on grass, and almost none of that grass resembled the impeccable grounds at Wimbledon. Rushing the net was the only way to avoid losing on a bad bounce.

There’s a direct line running from Kramer, through Laver and Pete Sampras, to early-career Federer. Roger gave up serve-and-volleying only when Lleyton Hewitt showed how a sturdy, precise defense–made possible, again, by improved tech–could turn even a strong serve-and-volley attack into a negative-EV proposition.

The overarching theme here is that tennis pros will chase expected value, just as they have for a century. If they don’t, other players will come along with a better approach and displace them. The tactics that work in a given era are heavily driven by tech, and they may or may not move in the direction of higher variance.

The women’s game shows us the potential of high-risk tennis. So many top players go for broke that someone like Swiatek–an aggressive player by historical standards–looks conservative by comparison. Ostapenko-style slugging looks nothing like serve-and-volleying, but the philosophy is similar: Put the ball away before your opponent has the chance.

The men’s game, though, is becoming ever more precise. Sinner and Alcaraz don’t have low Rally Aggression scores because they play so passively. They just don’t miss very often. Berrettini is more aggressive, but only just. Few men hit serves harder or pepper the corners so persistently. Fewer still are so relentless in how they capitalize on a short ball. Yet he does that seemingly without cost. The Italian has plenty of limitations–injuries and a limited backhand, for starters–but they aren’t tactical. He and his colleagues have concluded that higher risks aren’t worth it, and they are probably right.

* * *

Subscribe to the blog to receive each new post by email:

Coco Gauff’s Big What-If

*Coco Gauff at the 2022 US Open. Credit: All-Pro Reels*

The best players are able to work around their weaknesses. Coco Gauff is so solid that she has overcome two: an unreliable forehand and a second serve that hands too many points to her opponents. On Wednesday in Wuhan, Gauff coughed up 5 double faults out of 19 second serves. Despite surrendering more than 10% of her serve points for the fifth consecutive match, she eased past Viktoriya Tomova. The Bulgarian managed just three games.

The forehand is a subject for another day. Lately, the serve has been a bigger concern, the one blot on an eight-match win streak (and counting) in China.

Start with season totals. Through last week’s Beijing final, Gauff has missed more than one in five of her second serves. The result: She has double-faulted 8.9% of her 2024 service points. No other woman in the WTA top 60 has double faulted so often.

The typical tour regular loses barely half so many points this way. Tour average is 5.1%. Fellow elites Iga Swiatek, Jessica Pegula, and Jasmine Paolini come in at 3% or lower; Emma Navarro just misses that mark at 3.1%. Even Aryna Sabalenka, with her recurring bouts of service shakiness and occasional risk-taking on the second serve, gives away only 4.5% of points.

Still, Coco rates as the fourth-best player in the world. She’ll be back to #3 on Monday, and she has a good chance of ending the season there. The rest of her game is so sturdy that she has piled up nearly 50 wins on the season despite committing 274 more double faults than Swiatek has.

This is uncharted territory. In the last 15 years–the extent of my serve stats for women’s tennis–only two players have hit double faults so often and still managed to finish in the top five. No one has cracked the top three:

DF Rate  Player             Year  Rank  
  10.4%  Aryna Sabalenka    2022     5  
   9.6%  Maria Sharapova    2011     4  
   8.9%  Coco Gauff         2024     ?  
   8.7%  Elena Dementieva   2009     5  
   8.4%  Maria Sharapova    2015     4  
   8.1%  Dinara Safina      2008     3  
   7.9%  Dinara Safina      2009     2  
   7.9%  Maria Sharapova    2014     2  
   7.9%  Karolina Pliskova  2021     4  
   7.6%  Victoria Azarenka  2013     2  
   7.6%  Aryna Sabalenka    2021     2  
   7.5%  Maria Sharapova    2013     4  
   7.3%  Maria Sharapova    2012     2  
   7.0%  Venus Williams     2010     5

The typical year-end number one double faults only 4.1% of the time. Victoria Azarenka’s 2012 season, at 6.8%, was the only such occasion over 6%. This isn’t exactly a law of physics, but if Gauff is to dislodge the two women atop her in the ranking table, she’ll probably need to make a substantial move in that direction.

What-ifs

It’s no easy task to fix a leaky serve. The good news for Coco is that it may be all she needs to do.

Back to the season totals. Gauff is basically tied with Swiatek as the best returner in the game. The American has won 48.4% of her return points this year, compared to Iga’s 48.5%. Gauff has played slightly weaker opposition, but in any case, it’s a minor gap. Both women stand well above the pack; no one else tops 47.5%. With no double faults working against her, Coco’s return game is worthy of a world number one.

By service points won–where the double faults come into play–Gauff ranks a more pedestrian 12th. That’s entirely because of the deliveries that miss. She wins more first-serve points than anyone except for Qinwen Zheng and Elena Rybakina. In an era without megastars, the combination of 1st or 2nd on return and 12th on serve might be good enough to lead the field, but with an all-rounder like Swiatek and a dominant slugger like Sabalenka to contend with, it doesn’t do the job.

Here, then, is the what-if. Wave a magic wand and proclaim that all of Gauff’s second serves find the box. The 9% of her service points that end in double faults turn into second serves in play: points that she wins at a 56% clip.

Do that, and her rate of serve points won–currently at 60.2%, good for 12th place–becomes 65.3%, better than anybody. A double-fault-free Coco Gauff would rack up more serve points than anyone on tour, while still winning almost as many return points as Iga does. A handful of key points might swing the year-end number one in either direction, but statistically, the American would be the best player in the world.

You might argue that even in the rosiest real-life scenario, Coco isn’t going to eliminate double faults entirely. Fair enough. Reduce her double fault rate to tour average, and she wins 62.5% of service points. Not as good as Swiatek, Sabalenka, or Rybakina (or, technically, Lulu Sun in her limited tour-level action), but ahead of everybody else.

Combine serve and return into total points won (TPW%), and we see how these wishful adjustments move Gauff clear of the field–or, at least, everyone except for Iga:

TPW%   Player                      
56.9%  Coco Gauff (no dfs)  
56.5%  Iga Swiatek                 
55.4%  Coco Gauff (avg dfs)  
54.3%  Coco Gauff (actual)  
54.3%  Aryna Sabalenka             
53.7%  Elena Rybakina              
53.1%  Karolina Muchova            
52.9%  Qinwen Zheng                
52.8%  Danielle Collins            
52.7%  Mirra Andreeva              
52.6%  Jessica Pegula              
52.3%  Victoria Azarenka           
52.3%  Maria Sakkari               
52.3%  Paula Badosa                
52.1%  Madison Keys                
52.0%  Jasmine Paolini

Actual-Coco is already near the top of the list. Take away all or half of her double faults, and at the very least she looks stronger than Sabalenka and Rybakina.

The specifics

This may seem a bit too abstract, especially since the total-points-won list has so many differences from the official ranking table. Greatness is not measured by points, but by titles, and some trophies count much more than others.

Remember that these points we’re changing took place in real–often close–matches. Reversing just a few of the double faults would have tipped the scales in Gauff’s direction. In the counterfactual, she probably didn’t lose 15 matches this year. She likely picked up more than two titles.

Take the most painful loss of the season: Coco’s fourth round defeat at the US Open. Against Emma Navarro, she committed a gut-wrenching 19 double faults. Despite that, she won 46.8% of total points. All else equal, had she landed those 19 second serves, Gauff would have almost exactly flipped the tally, winning 53.0% of points. Even with a tour-average double fault rate, she would have won 51.0% of points and–barring bad luck or a ill-timed choke–earned a victory.

Run the same exercise for the American’s other defeats this year, and we see just how strong her season could have been. If we reduce her double faults to a tour-average 5.1%, 4 of her 15 losses probably would have gone her way. Two more matches would have ended within a point of 50/50, safely in the range where a clutch (or lucky) break point or two can reverse the result.

Cut out double faults entirely, and Gauff wins at least 50.8% of points in six of the losses. She would have cleared 48% in four more, putting those in the range where luck could hand her the victory.

Even in the more conservative scenario, Gauff’s campaign looks quite different. Instead of losing to Anna Kalinskaya in the Dubai quarters, she would have faced off with Iga in the semi-finals. She wouldn’t have lost to Marta Kostyuk in Stuttgart: She’d have played Marketa Vondrousova for a place in the final. In Madrid, she would have handily beaten Madison Keys, earning a quarter-final date with Ons Jabeur. Flip the Navarro result in New York, and Coco could well have defended her US Open title.

Today’s action in Wuhan offered a glimpse of a sturdier future. Gauff cast aside Kostyuk with nary a double fault, advancing to the quarters in just 61 minutes. It was her quickest match since April–against an opponent who has bedeviled her in the past–and her first double-fault-free outing in 14 months.

The American has somehow established herself as a top-five player and grand slam champion despite handing her opponents more free points than any of her peers. A stingier Coco Gauff could soon be the best player in the world.

* * *

Subscribe to the blog to receive each new post by email:

Tomas Machac’s Defiant Angles

*Tomas Machac at the 2023 US Open. Credit: Hameltion*

2024 is quickly turning into the year of Tomas Machac. The 23-year-old Czech reached his first grand slam third round in Australia, straight-setting Frances Tiafoe for a first top-20 win. A quarter-final showing in Marseille and a defeat of Stan Wawrinka at Indian Wells earned him a place in the top 60.

Now, in Miami, he has dispatched top-tenner Andrey Rublev and outlasted Andy Murray for a place in the fourth round. The live rankings place him precariously in the top 50; tomorrow’s match against fellow second-week surprise Matteo Arnaldi give him a chance to make it official. While Jiri Lehecka, a year younger and considerably higher in the rankings, is the poster boy for the resurgence of Czech men’s tennis, Machac is right behind him.

The key to the Machac game is a compact, versatile backhand that seems capable of anything. Inside-out backhands are usually little more than a curiosity, a miracle of timing that many players don’t even bother to try. The Czech hits one in ten of his backhands that way. Against Rublev, he cracked five: one for a winner and two more that forced errors. He won all five.

The tactics that surround Machac’s backhand are a joy to watch. Since he doesn’t serve big, every point threatens to become a rally. But the Czech angles for court position like a much bigger hitter. He approached the net 35 times in yesterday’s Murray match alone. Counting the times he was forced to come forward as well, he played 48 points in the forecourt, winning 38 of them. Combined with a court-widening slice serve, the net play makes Machac just as much of a threat on the doubles court. With Zhang Zhizhen, he reached the semi-finals in Australia and won the title in Marseille. He and girlfriend Katerina Siniakova would make a dangerous mixed duo at the Paris Olympics.

The unknowns that could limit Machac’s ceiling are, well, everything else. His forehand is a bit hitchy and it is nowhere near as effective as his backhand. By my Forehand Potency metric (FHP), he earns barely any points off that wing, ranking among the likes of Adrian Mannarino and Mikael Ymer.

And then there’s the serve. While he is capable of firing bullets–one of his serves in Australia registered at 128 mph (208 kph)–he rarely goes that route. His first serves in Miami have hovered around 110 mph, so he sets up points with slices wide, especially in the deuce court. He manages a respectable ace total thanks to a well-disguised delivery and the surprise that comes from his occasional bombs down the T.

The Machac serve is not a liability, exactly, but it is not the standard first-strike weapon for a prospect in today’s men’s game. Let’s take a closer look.

Lean right

Aside from keeping an eye on the radar gun while watching Machac’s progress in Miami, I don’t have a lot of data to put his serve speed in context. The only available point-by-point serve speed data these days comes from Wimbledon and the US Open, where the Czech has played just two career main-draw matches.

At Wimbledon last year, Machac’s first serves clocked an average of 115 mph (184 kph), faster than about one-third of the field. The Wimbledon gun might have been a little hot, as most players scored better there than in New York, and by a wider margin than you’d expect from more serve-centered tactics. When the Czech played a match at the US Open in 2022, his average first serve speed was 107 mph (171 kph). Four-fifths of the field hit harder; most of the names in his part of the list are clay-courters. Presumably he has gotten stronger since then, so while 115 mph may be an overestimate, 107 mph is probably low.

These numbers confirm that the serve won’t hold him back too much. Some other men in the same neighborhood are Casper Ruud, Tommy Paul, and David Goffin. Neither Carlos Alcaraz nor Novak Djokovic averaged much faster than Machac on the Wimbledon gun last year, and they did just fine. The Czech has only a bit of ground to make up with the rest of his game, and Ruud offers one example that it can be done.

What makes Machac’s serve look so pedestrian is the frequency with which he spins wide serves in the deuce court. Against Murray yesterday, he hit 54% of his deuce-court firsts to the wide corner. Fewer than 40% went down the T, and most of the remainder were also to the forehand side. He was even more extreme in the ad court, spinning 61% of those first serves down the T to the opponent’s forehand.

60/40 sounds rather undramatic, like most tennis stats. But few men favor one direction so strongly, at least until they reach critical situations like break point, when they lean more heavily on their favorite angle. Machac tries to balance it out by aiming for the backhand with his second serves, though by a slightly narrower margin. That does the job: The gap between his first- and second-serve results is about the same as tour average.

In the deuce court, at least, the tactic is working. Against Murray yesterday, Machac won 18 of 22 (82%) when his first serve went wide, though he was nearly as successful down the T. Against Rublev, he won 13 of his 14 wide deuce-court first serves. Understandably, he didn’t hit many deuce-court serves anywhere else. When Murray broke back yesterday to keep the third set alive, it wasn’t the serve itself that let Machac down. Twice at deuce, the Czech missed first serves when he tried to go down the T. His wide second serves drew weak replies on both occasions, but he lost both points with unforced errors.

The dis-ad-vantage

Wide serves in the deuce court are a gamble. You let your opponent take a swing at a forehand–probably his preferred wing–but you pull him out of position. Clearly it can work. Few men rely more heavily on their forehand than Rublev does, yet Machac attacked that side at every opportunity.

Murray was cannier and kept things much closer than Rublev did. But even he was fighting a losing battle. Machac won 80% of total first-serve points in the deuce court yesterday, compared to 69% in the ad court. So far, the Czech’s opponents have been more like Murray than Rublev, but still, the serve-to-the-forehand gamble pays off.

While he likes to aim for the same wing in the ad court as well, Machac doesn’t get the same court-position advantage. Across ten matches logged so far by the Match Charting Project, he has won 78% first-serve points in the deuce court against 71% in the ad court.

The difference lies largely in what Machac can do with his plus-one shot. In the deuce court, he wins about half of first-serve points with his serve or plus-one. In the ad court, that number falls below 40%. 50% is excellent: Djokovic hardly does better than that, and even an imposing server like Ugo Humbert does worse. But 40% is dire. Only clay-courters win so few short first-serve points overall. There’s less room to put away the second shot when you’ve left the returner standing in the middle of the court.

There’s nothing inherently wrong with a split between deuce-court and ad-court results. If asked, most players would probably prefer to win more points in the ad court, since most break points start in that direction. But the effect of winning more break points is mostly cancelled out by earning fewer break chances in the first place. Anyway, Machac doesn’t have any particular problem saving break points. He survived 13 of 15 against Murray. At tour level since this time last year, he has saved 64.5% of break points faced while winning 65.5% of serve points overall. That’s a closer margin that most players can boast.

The deeper we dig, the more we find weaknesses and unusual preferences in Machac’s game. Paired with each one, it seems, is a way in which it could work to his advantage. So far, he has succeeded despite the oddities. His results against Rublev and Tiafoe suggest that stronger competition might not break the spell, though the demands of yesterday’s gutbuster with Murray makes me wonder if brainier competition will raise the bar.

As the men’s game gets ever more powerful, there is less room at the top for playing styles that break the mold. Machac has already hinted that he can counterbalance brute force with the right set of angles, especially if they create opportunities for him to deploy his top-tier backhand. Countryman Radek Stepanek cracked the top ten with his own brand of unorthodox unpredictability. Machac has a different set of quirks, but based on his rapid progress this year, he may be able to do the same.

* * *

Subscribe to the blog to receive each new post by email:

All Hail the Iga Swiatek … Serve?

*Iga Swiatek at the 2023 US Open. Credit: Hameltion*

There are a million things to praise about Iga Swiatek’s tennis these days. This puts commentators in a quandary, because her matches are often so short that there isn’t time to list them all. She is world-class at nearly every aspect of the game.

If there is an exception, it is her serve. While it is not a liability, it doesn’t appear to stand out as a weapon, and Swiatek continues to make technical tweaks to improve it. She doesn’t dominate first-serve points the way that Qinwen Zheng or Elena Rybakina does; she doesn’t pile up aces like Rybakina, Karolina Pliskova, or Aryna Sabalenka. The longer a point lasts, the more time she has to take control, so who needs a standout first strike?

A look at the bigger picture, though, tells us that Iga’s serve is just fine. She was broken just five times in six matches en route to her second Indian Wells title. In the last 52 weeks, she has held 81.6% of her service games, best on tour.

To quote myself when Alex Gruskin threw that stat at me a couple of weeks ago: Wait, WHAT?!

Here’s the top ten since Miami 2023:

Player              Hold%  
Iga Swiatek         81.6%  
Aryna Sabalenka     79.4%  
Elena Rybakina      78.5%  
Caroline Garcia     76.4%  
Madison Keys        76.4%  
Petra Kvitova       75.8%  
Katie Boulter       74.3%  
Qinwen Zheng        73.2%  
Liudmila Samsonova  73.1%  
Maria Sakkari       73.0%

I can already hear everybody sputtering out their “yeah but” explanations, and we’ll get to some of them in a moment. First, though, we need to acknowledge just how elite this is. Sabalenka held at 80.8% last year, her best campaign so far. In 2015, when Serena Williams went 53-6 and won three majors, she held 80.9% of service games. Ash Barty peaked at 80.1%. Pliskova has twice cracked 79% for a full season, but never 80%. Same for Kvitova.

WTA match stats are sparse before the mid-2010s, so I don’t have numbers for Navratilova, Graf, Davenport, Venus Williams, and the rest. (Navratilova won 75% of total games in 1983, so… wow.) Suffice it to say, hold percentages that start with an eight are the province of all-time serving greats. Iga has muscled her way into that group.

The all-rounder

The simplest explanation of Swiatek’s serve stats is that she wins a lot of all kinds of points. As long as she doesn’t double fault, she’s in a rally, and she doesn’t lose many rallies.

This is true, sort of. In the last 52 weeks, Iga has won almost half of her return points, good for a break percentage of 49.5%. That leads the tour as well, granting her a spot in the hyper-exclusive Top One Club.

The average player in the WTA top 50 has a hold percentage about 33 points higher than her break percentage. Iga’s difference of 32 points, then, is not far from the norm. Despite winning so many service points, she is an entirely different sort of player than Rybakina (43 point gap) or Caroline Garcia (55 [!] point gap). Swiatek tacks an average serve onto a game that is otherwise outlandish.

On the other hand, it’s easy to underrate average. Most players who are extremely good at one thing are lucky if the rest of their game can pull enough weight to keep them on tour. The biggest servers are often indifferent (at best) on return; the best baseline players are rarely blessed with world-beating serves. Here are the current top ten returners among the top 50 (plus #52 Sara Sorribes Tormo), shown with their hold percentages and the differences between their serve and return results:

Player               Break%  Hold%   Diff  
Iga Swiatek           49.5%  81.6%  32.1%  
Lesia Tsurenko        48.1%  56.4%   8.3%  
Sara Sorribes Tormo   47.5%  58.4%  10.9%  
Clara Burel           45.4%  61.3%  15.9%  
Coco Gauff            44.7%  71.2%  26.5%  
Daria Kasatkina       44.5%  62.4%  17.9%  
Marketa Vondrousova   44.0%  68.7%  24.7%  
Jessica Pegula        43.1%  72.0%  28.9%  
Elise Mertens         41.0%  65.2%  24.2%  
Katerina Siniakova    40.2%  61.3%  21.1%  
Ons Jabeur            40.1%  67.0%  26.9%

If we approximate “serve-specific skill” as the difference between hold and break percentage, we find that the best returners are–unsurprisingly–generally weaker servers. Everyone on this list is below average in serve-specific skill. Among this group of elite returners, though, Iga is the best server. Only a few women–familiar names like Pegula and Gauff–come close.

Here is the relationship in visual form:

Iga clearly occupies a world of her own.

What works

One thing Swiatek does well is that she can hit her serve hard. At least year’s US Open, 40 different women had at least 100 first serves that landed in the box and registered on the radar gun. The top of the list are the names you’d expect: Sabalenka, Qinwen Zheng, Samsonova, Gauff, and Keys. Next up were Elise Mertens and… Iga Swiatek. Iga’s average first serve was a rounding error away from Keys’s and just 2.5 km/h slower than Gauff’s.

Speed matters, obviously. All else equal, a faster serve means more aces, more short points, and more service points won. The spin that Swiatek generates may make her first serves more difficult than the radar gun indicates, as well. When five-foot, four-inch Yulia Putintseva challenged the Iga serve at Indian Wells, she often found herself making contact at or above head level. Putintseva, I suspect, would have preferred to take on a flatter hitter like Samsonova, even if it meant handling a few more miles per hour.

Raw speed might also be underrated. When I dug into some ATP numbers to tease out the effects of speed and precision last month, I found that speed seems to matter more than accuracy. Equivalent data isn’t available for the women’s game (and the men’s data itself was exceedingly sparse), but it seems reasonable to assume that the relationship would be similar.

The relative effects of speed and precision are particularly important to Swiatek, because she hits a lot of her serves down the middle of the box. (Technically, those serves could still be “precise,” in the sense that they land close to the service line, but they won’t be as unreturnable as the equivalent deep serve close to a sideline.) Match Charting Project data tells us that the average WTAer hits 21% of their first serves down the middle. Iga comes in at 32%. Returners start the point on the back foot, even if they don’t have to move their back foot very far.

Swiatek gets away with all those down-the-middle serves, partly because she is better than her peers in general, and partly because she sacrifices less effectiveness than average by choosing a more conservative target. Here are her first-serve winning percentages by direction:

Direction   Iga W%  Tour W%  
Deuce-Wide     68%      65%  
Deuce-Body     64%      57%  
Deuce-T        74%      67%  
Ad-T           69%      64%  
Ad-Body        63%      56%  
Ad-Wide        69%      64%

(I use “down the middle” and “body” interchangeably here, because that’s how Match Charting Project logs are coded. Within tennis, the term “body serve” often refers to a narrower category of balls aimed directly at the returner. Iga hits some of those, but an awful lot of her serves–even her first serves–are neither that sort of body serve nor a delivery aimed at a corner.)

The average player gives up eight percentage points when they go down the middle. Iga sacrifices only six. It also helps to be so good in general. A winning percentage of 63% or 64% will keep you in a service game; 56% or 57% will put it much more at risk.

82%, here we come

One benefit of scoring so many points with down-the-middle serves is that it allows Swiatek to save the angles for when it matters most. It’s tough to pinpoint exactly what the key moments are for Iga, since her matches are often so lopsided. Serving at 4-all in the first set against Maria Sakkari yesterday, she built a 30-15 advantage with three first serves to the body. She served wide on the next point and down the T at 40-15. Neither one came back. She didn’t lose another game the rest of the way.

My hypothesis, based on watching her recent matches, was that this was a recurring pattern, that Iga goes to the corners more on key points and thus holds serve even more often than her serve-point success would indicate. But this is wrong, at least facing break points over the last 52 weeks. Since Miami last year, she has won 64.6% of serve points, but only 60.6% against break point. Most women save break points less often than they win other serve points, because break points tend to be generated by stronger returners. But a margin of two percentage points is typical. Iga’s four-point gap is not.

In fact, Swiatek was dreadful facing break point last year. A few years ago I built a metric to measure each player’s success rate at break point, comparing their break points saved to the number of points they’d be expected to win based on the other serve points played in those matches. By Break Points Over Expected (BPOE) in 2023, Iga was dead last among tour regulars. She faced 311 break points, and if she had served as well on those points as she did in the rest of those matches, she would have saved 184 of them. Instead, she saved 165, a difference of -19. No other top player had a negative result worse than -5.

Fortunately for the Polish star, this is the kind of clutch (or anti-clutch) performance that tends not to persist. Either it’s bad luck, or the choking turns out to be temporary. And indeed, in 2024, Swiatek has turned things around. She has saved 76 of 109 break points faced instead of the expected number of 66. She probably won’t sustain that level of break point overperformance, but even a neutral score would further improve her tour-leading hold percentage. If she could prove out my hypothesis and win more break points than expected by saving her best serves for those moments, she would head further into untouchable territory.

No one will ever mistake the Swiatek serve for the cannon of Sabalenka or Rybakina. But Iga’s overall game means she wins more points than the heavier hitters. Her serve doesn’t have to be great, it just needs to stick around tour average. She has achieved that, and–pity her poor opponents–there is room for her to improve even more.

* * *

Subscribe to the blog to receive each new post by email:

Can Sebastian Baez Find Success on Hard Courts?

*Sebastian Baez in Cordoba last month. Credit: jmmuguerza*

Sebastian Baez is a marvel. In an era dominated by tall, all-court sluggers, the five-foot, seven-inch Argentinian has carved out a place on the circuit as a throwback clay-court specialist. Just a couple of months past his 23rd birthday, he has already won six tour-level titles and reached a new career-best ranking of 19th on the ATP computer.

The obvious comparison is Diego Schwartzman, another Argentinian on the small side who won titles and reached a French Open semi-final by grinding out victories and swinging above his weight. Schwartzman eventually cracked the top ten, but when he was Baez’s age, such a milestone looked extremely unlikely. When he turned 23, he stood outside the top 60, heading back to South America for another cycle through the continent’s late-season Challenger swing. He wouldn’t reach the top 20 for another two and half years.

If we assume Baez continues to improve throughout his mid-20s in the same way that Schwartzman did, a single-digit ranking seems achievable. He’s already 11th on tour in clay court Elo. Only a few players ahead of him on the official ranking table are younger.

The main stumbling block is the natural ceiling on dirtballers. There are far more ranking points available on hard courts than on clay, and one of the prime opportunities on the Argentinian’s favorite surface, in Madrid, plays fast because of its altitude. (Baez has competed there only once, losing in the second round last year to another sterling clay-courter, Stefanos Tsitsipas. Schwartzman never won more than two matches there, either.) For Schwartzman to gain a top-ten place, he needed more than a Roland Garros semi-final: He had recently reached indoor finals in Vienna and Cologne, and he was 14 months removed from defeating Taylor Fritz for a hard-court title in Los Cabos. Diego’s magic somehow worked on all surfaces. Even in a year when he posted outstanding results on clay, that was his only route to a single-digit ranking.

Baez owns one hard-court championship, from last year’s US Open warmup in Winston-Salem. But apart from that week, his story diverges from Schwartzman’s, with a career record on the surface of just 17-33. He has never won two completed matches at any other tour-level hard-court event. (His two third round appearances at majors were aided by retirements.) The Argentinian can be a star and a national hero without all-surface success, but surely he wants more. Can he achieve it?

Surface and speed

When I wrote about surface sensitivity a couple of weeks ago, Baez didn’t stand out as an extreme. Tsitsipas and Lorenzo Musetti were the men whose results were most dependent on slow courts. The numbers showed that Baez does better with a slower bounce, but the effect is less than half of what it is for Tsitsipas. The 23-year-old is more closely comparable in this department to Daniil Medvedev, who doesn’t like to play on clay–or eat it.

However, that analysis left out one major factor. I simply rated tournaments by the degree to which they helped servers end points quickly, regardless of surface. Indian Wells, on hard courts, came out as just barely speedier than Rome and slower than Madrid. Miami and the US Open were roughly equivalent to the Caja Magica as well.

Intuitively, there is more than one dimension to player preferences. Some men might just want more time to prepare, as could be the case with Tsitsipas and his one-handed backhand. But others–Baez among them–are much more comfortable on a certain type of surface, because of the type of bounce, the footing, or both. When we reduce surface type and speed to one variable, Baez and Medvedev come out equal. When we separate type from speed, they look very different.

This scatterplot shows 56 players on the two dimensions:

(The units are regression coefficients and essentially meaningless out of context. They do, however, show direction and magnitude of each player’s preferences.)

Among players with at least 100 tour-level matches since 2021, Baez ranks third in the degree of his preference for clay courts, behind Albert Ramos and roughly tied with Alexander Zverev. Once surface type is controlled for, he prefers faster courts. Santiago, where he won the title last week, is one of the quicker clay courts on the circuit, giving servers as many quick points as Wimbledon (really!). Rio de Janeiro, the site of his triumph the week before, is also faster than the average dirt, rating about the same as Indian Wells.

Medvedev is the exact opposite. Only Adrian Mannarino has a stronger demonstrated yen for hard courts. Once his choice of surface is secure, though, the Russian wants it as slow as possible. Only a few players (including Musetti and another slow-hard-specialist, Alex de Minaur) are so extreme.

Schwartzman–the model for a potential all-court Baez–prefers clay, and he likes all of his courts slow. His performance is even more speed-dependent than Medvedev’s, but his surface type preference isn’t nearly as strong as that of his younger countryman.

This is all rather abstract, and to some degree, it’s just a fancy way of saying that Baez struggles on hard courts. Let’s make things more concrete by looking at what happens when the Argentinian shifts to the tour’s more popular surface.

Translations

Hard-court tennis is more serve-dominated than the clay-cout variety. The typical tour regular wins, on average, 3% more service points on hard than on clay: 4% more first-serve points and 1% more second-serve points. They win 7% fewer return points. (That sounds like a paradox, since the serve and return numbers are different. The catch comes from specifying “tour regulars”–part-timers on hard courts have bigger serves than their equivalents at clay events.)

Here are the player-specific numbers for each man in the top 20 (except for Ben Shelton, who hasn’t played much on clay). The figures are ratios of each hard-court metric to the corresponding clay-court metric–serve points won, first-serve points won, and return points won–so the higher the number, the bigger the difference in favor of hard courts.

Player               SPW  1st SPW   RPW  
Ugo Humbert         1.10     1.08  0.95  
Novak Djokovic      1.09     1.11  0.91  
Alex de Minaur      1.08     1.08  0.99  
Daniil Medvedev     1.06     1.07  0.98  
Jannik Sinner       1.06     1.08  0.95  
Tommy Paul          1.06     1.08  1.08  
Grigor Dimitrov     1.06     1.04  0.95  
Holger Rune         1.05     1.06  0.92  
Taylor Fritz        1.05     1.09  0.93    
Alexander Zverev    1.05     1.03  0.89

Player               SPW  1st SPW   RPW  
Alexander Bublik    1.05     1.04  1.04  
Karen Khachanov     1.05     1.09  0.96  
Andrey Rublev       1.05     1.05  0.94  
Hubert Hurkacz      1.04     1.02  1.00  
Frances Tiafoe      1.03     1.05  0.95  
- ATP Regular -     1.03     1.04  0.93  
Carlos Alcaraz      1.02     1.02  0.92  
Stefanos Tsitsipas  1.02     1.03  0.89  
Casper Ruud         1.01     1.04  0.91  
Sebastian Baez      0.94     0.96  0.85

Baez is… different. Everyone in the top 20 wins more serve points on hard courts than on clay except for him. Only a few other men on tour have the same “backwards” split, and only Federico Coria is close to Baez in the degree of his weaker hard-court service performance. What costs the Argentinian even more is how his return numbers suffer away from clay. Almost everyone wins fewer return points on hard, but Baez takes the cake here too.

Perhaps needless to say, there’s no way that this can work. Baez wins about 62% of service points on clay, a respectable number and an impressive one for someone his size, but still below the average of a top-50 player on the surface. To win even fewer on hard suggests that he would struggle even at Challenger events. At Winston-Salem last August, Baez won 63.5% of his serve points and over 43% on return. That’s a combination that will win matches; he just hasn’t provided any evidence that he can pull it off once he crosses back out of North Carolina.

Growth rate

The one reason for optimism is that Baez is young, inexperienced on surfaces other than clay. Like Schwartzman, he grew up playing on dirt, and he rose through the rankings by winning South American Challengers, then picking up victories on the continent’s Golden Swing. Maybe there’s a necessary transition period?

Here are the same ratios as above, now by season for our two Argentinian heroes:

Player  Year   SPW  1st SPW   RPW  
Diego   2015  1.01     0.99  1.01  
Diego   2016  1.04     1.07  0.94  
Diego   2017  1.08     1.09  0.91  
Diego   2018  1.03     1.03  0.89  
Diego   2019  1.08     1.09  0.94  
Diego   2020  1.03     1.03  0.85  
Diego   2021  1.02     1.03  0.95  
Diego   2022  1.01     1.04  0.89  
Diego   2023  1.09     1.08  0.88  
                                   
Player  Year   SPW  1st SPW   RPW  
Baez    2022  0.91     0.94  0.82  
Baez    2023  0.96     0.99  0.88

Baez did close the gap between his hard-court and clay-court performances in his second year on tour. But he still shows a more marked surface preference than Schwartzman ever did. As soon as Diego arrived on tour, he was able to win more service points on hard courts–roughly the same ratio as the typical tour regular. Baez isn’t even close. Schwartzman had to retool his game to succeed on hard courts, and Baez will need to do so even more.

The 23-year-old truly is a throwback, an undersized grinder who spins in his serves, plays defense, and constructs points. It’s a joy to watch, and the package makes him one of the best players on tour for the 14 weeks or so each year when there are top-level clay events on offer. It doesn’t, however, work so well when there’s no dirt to kick out of his cleats. Fortunately Baez is young, and he has many years left to figure it out. He’ll need to.

* * *

Subscribe to the blog to receive each new post by email:

Alex de Minaur’s Adequate Inaccuracy

*Alex de Minaur at the 2024 Australian Open. (Getty Images: Julian Finney)*

Last week, Tennis Insights posted a graphic showing the average first serve speed and accuracy–distance from the nearest line–for the ATP top 20. There’s a ton of fascinating data packed into one image.

Hubert Hurkacz is fast and accurate, Novak Djokovic is nearly as precise, and Adrian Mannarino defies logic as always. The most noteworthy outlier here, especially just after his run to the Rotterdam final, was Alex de Minaur. The Australian gets plenty of pop on his first serve, hitting them faster than tour average, if slower than most of the other men in the top 20. This comes at a cost, though. As one of the shortest guys among the elite, he doesn’t hit the lines. He’s by far the least accurate server in this group:

Precision is great: It’s certainly working for Hurkacz and Djokovic. But everything is a tradeoff. Any pro player could hit more lines if there was no reason to serve hard. Or vice versa: If the goal was simply to light up the radar gun, these guys could add miles per hour by aiming at the middle of the box. Standing a modest six feet tall, De Minaur is even more constrained than his typical peer. No technical tweak is likely to move him into Hurkacz territory. He might make small improvements or swap some speed for more accuracy.

Small gains would be enough, too. De Minaur wins fewer first-serve points than the average top-50 player, but in the last 52 weeks, he has outpaced Carlos Alcaraz, Holger Rune, and Casper Ruud. He trails Alexander Zverev by about one percentage point. This isn’t Sebastian Baez (or even Mannarino) we’re talking about. Whatever the cost of de Minaur’s inaccuracy, he’s able to overcome it. It’s just a matter of what gains he could reap by making returners work a bit harder.

Here’s the question, then: How much does accuracy matter?

Speed first

For this group of players over the last 52 weeks, speed is by far the most important factor in first-serve success. Speed alone–ignoring accuracy or anything else–explains 72% of the variation in first-serve points won. Accuracy alone accounts for 43%. (The players who are good at one thing are often good at others, so most of those 72% and 43% overlap.) 43% might sound like a lot, but isn’t that far ahead of something as fundamental as height, which explains 33% of the variation.

Surprisingly, precision is even less critical when it comes to unreturnable first serves. Using unreturned serve counts from Match Charting Project data, accuracy explains just 30% of the variation in point-ending first serves, less than we could predict from height alone. (Speed alone explains 60% of the variation in unreturned serves.) I would have expected that accuracy would play a big part in aces and other unreturned serves, since a ball close to the line is that much harder to get a racket on. But while precision may increase the odds of any individual serve going untouched, average precision isn’t associated with untouchable serving.

The story is the same for any metric associated with first-serve success. Speed matters most. There’s immense overlap between the factors I’ve discussed: Taller players find it easier to hit the corners, and all else equal, they take less of a risk by hitting bigger. There is probably some value of height that isn’t captured by speed or accuracy, such as the ability to put more spin on the ball, but the main benefit shows up on the radar gun.

To tease out the impact of each variable, I ran a regression that predicts first-serve points won based on speed, accuracy, and height. The results should be taken with an enormous grain of salt, since we’re looking at just 20 players, some of them the game’s most outrageous outliers. Still, the findings are plausible:

Speed: Each additional mile per hour translates to an improvement of 0.43 percentage points in first-serve points won. (1 kph: +0.27 first-serve points won)
Accuracy: Decreasing distance from the line by 1 cm results in an improvement of 0.2 percentage points in first-serve points won.
Height: At least for these twenty players, the value of height is entirely captured by speed and accuracy. The margin of error for the height coefficient spans both positive and negative values. It is unlikely that height is a negative, though I suppose it’s possible, if speed and accuracy capture the height advantage on the serve itself, and height is a handicap on points that develop into rallies. Either way, the impact is minor, if it exists at all.

Approximately, then, one additional mile per hour is worth the same as two centimeters of accuracy. The height of the graph–110 to 130 mph–represents a variation of nearly 9 percentage points of first-serve points won. The width–70 to 52cm–represents a range of 3.6 percentage points. Broadly speaking, speed remains more important than accuracy, though a particular player might find it easier to improve precision than power.

Just one example of what the numbers are telling us: De Minaur has won 72.8% of his first-serve points over the last year, compared to the top-ten average of 75.3%. If this model were to hold true–a big if, as we’ll discuss shortly–that’s a gap he could close by improving precision by about 12 cm, to a tick better than tour average.

Drowning in caveats

For every question we answer, we’re rewarded with ten more questions.

I’m most interested in the choices that individual players could conceivably make, and the analysis so far offers only hints to that end. For this group of servers, we can say that a player who serves faster will win more points than his slower-serving peers. But we don’t know whether a specific player, if he was able to juice his serve by a mile or two per hour, would enjoy the same benefits. Hitting harder, or placing the ball more accuracy, is better, but by how much?

(I dug into the speed question way back in 2011 and found that one additional mile per hour–for the same player–was worth 0.2 percentage points. More recently, I found that for Serena Williams, an additional mile per hour was worth 0.5 percentage points. One of these days I’ll revisit the initial study with the benefit of many more years of data and perhaps a bit more wisdom.)

De Minaur was unusually precise in Rotterdam. Another Tennis Insights graphic indicates that his accuracy improved to about 55 cm for the week, an enormous gain of 15 cm from his usual rate, even more remarkable because his average speed was a bit quicker as well. (Playing indoors probably helped.) The model suggests that 15 cm is worth three percentage points. His boost of two miles per hour should have been worth nearly one more percentage point itself. Yet he won “only” 73.9% of his first-serve points–about one percentage point better than his non-Rotterdam average.

It’s just one week, so it isn’t worth fretting too much over the discrepancy. Still, it illustrates the value of the data we don’t have. (By “we,” I mean outsiders relying on public information. The data exists.) If we knew de Minaur’s accuracy and speed for every match, we could figure out their value to him specifically. Perhaps an uptick of one mile per hour is worth 0.43 percentage points only if you start serving like the players who serve faster–guys who are generally taller and can put more slice or kick on the ball. Those weapons aren’t available to the Aussie, so a marginal mile per hour may be less valuable. For him, accuracy might have a bigger payoff–relative to speed–than it does for other players. We just don’t know.

Sill, we’ve extracted a bit of understanding from the data. We’ve seen that accuracy translates into more first-serve points won, and we have a general idea of how many. De Minaur showed himself capable of hitting the lines as precisely as Andrey Rublev or Grigor Dimitrov, at least under a roof for one week. Just half that improvement, if he could sustain it–even if he didn’t get the full gains predicted by the model–would shore up a mediocre part of his game and lay the groundwork for a longer stay in the top ten.

* * *

Subscribe to the blog to receive each new post by email:

Surface Sensitivity and Ugo Humbert’s Serve

*Ugo Humbert in 2023. Credit: Hameltion*

Let’s start off with a couple of puzzles. I realize they aren’t the sort of things that keep most of you up at night, but they were odd enough to drive me to a flurry of coding, data analysis, and now blog writing.

On Wednesday, Ugo Humbert lost his first-round match in Rotterdam to Emil Ruusuvuori. It marked an unceremonious end to a hot streak for Humbert: He not only won the title in Marseille last week–launching himself into the Elo top ten–but he strung together 31 consecutive holds. 1,000 kilometers north, on a different indoor hard court, he got broken twice by a man ranked outside the top 50.

That’s the first puzzle: Why did the Frenchman lose? Again, it’s not that odd, as my Elo ratings gave Ruusuvuori a one-in-three shot to pull the upset. But it’s a match that Humbert should have won.

Head-scratcher number two: Why does Humbert always lose to Ruusuvuori? Wednesday’s decision marked their fifth meeting, and the Finn is undefeated. While the outcome is always close–Rotterdam was their fourth deciding set, and the other match went to two tiebreaks–the results are starting to get boring. Ruusuvuori is a solid player, and he is consistently able to blunt the Frenchman’s serve. But five in a row?

The answer to both mysteries is the same, and it’s more satisfying than I expected. Rotterdam is unusually slow for a hard court, especially indoors. Like most (or perhaps all) of the previous Humbert-Ruusuvuori venues, it plays slower than tour average. Just as important, Humbert’s game is unusually sensitive to surface speed. While that isn’t always true of big servers, he stands out as a fast-court specialist. We couldn’t have confidently predicted a Finnish upset, but we could have guessed that the Marseille champion would find this week’s tournament tougher going.

Rotterdam, it’s slow

The last time I published surface speed numbers, in late 2019, Rotterdam rated as the slowest indoor hard court on tour. Adjusting for the mix of players at the event, there were 10% fewer aces at the tournament than expected. It was a sharp decline from 2017 and 2018, when the venue sported more typically speedy indoor conditions.

Since then, the results have remained similar. Last year, the rate was 5% lower than expected, roughly tied with Stockholm as the slowest indoor surface on tour. Marseille, by contrast, gave players 12% more aces than usual.

There are limitations to using aces as a proxy for surface speed; I use aces because it’s the most relevant data that is widely available. Still, while you can quibble about the methodology or about a specific tournament’s place on the list, the overall rank order seems about right. Aces–adjusted for each event’s field–tell you much of the story.

With a growing mass of Match Charting Project data, we can do a little better. We have shot-by-shot logs for over one thousand matches since 2021. To compare conditions, I used my Serve Impact metric, which estimates how many points a player wins, directly or indirectly, because of his serve. It counts aces, other unreturned serves, and a fraction of the service points that take longer to decide. Depending on your motivation in measuring court speed, this isn’t perfect either: It doesn’t directly tell you anything about bounce height, for instance. But if you want to know what sort of players a tournament favors, Serve Impact gets you close.

By this more sophisticated metric, Rotterdam is… still slow. The venue takes away 4% of the points a player typically earns from his serve. Marseille and Montpellier each swing 7% in the other direction, Stockholm and Vienna provide a modest 3% boost, and Basel adds 8% to the server’s punch. With the exception of the short-lived tour stop in Gijon, Rotterdam has been the slowest indoor hard court of the 2020s. Even the clay in Lyon plays faster.

Here are the Serve Impact adjustments for the tournaments best represented in the dataset. Higher numbers mean faster conditions with more points decided based on the serve:

Tournament            ServeImpact  
Stuttgart                    1.29  
NextGen Finals               1.20  
Tour Finals                  1.16  
Wimbledon                    1.11  
Shanghai Masters             1.11  
Halle                        1.10  
Queen's Club                 1.08  
Basel                        1.08  
Washington                   1.08  
Dubai                        1.07  
                                   
Tournament            ServeImpact  
Antwerp                      1.05  
Gstaad                       1.05  
Australian Open              1.04  
Davis Cup Finals             1.04  
Cincinnati Masters           1.04  
Paris Masters                1.03  
Vienna                       1.03  
Miami Masters                1.02  
Madrid Masters               1.01  
US Open                      1.01  
                                   
Tournament            ServeImpact  
Canada Masters               1.00  
Rotterdam                    0.96  
Indian Wells Masters         0.95  
Rome Masters                 0.92  
Acapulco                     0.87  
Barcelona                    0.87  
Roland Garros                0.83  
Monte Carlo Masters          0.83

Average Serve Impact is around 34%, so the 4% hit in Rotterdam knocks that down to about 32.6%. Humbert has an above-average serve, so the slow-court penalty is greater still. He isn’t going to win any awards for rallying prowess, especially against someone as sturdy as Ruusuvuori, so the points that he doesn’t secure with his serve will disproportionately go against him.

The first three meetings in the Humbert-Ruusuvuori head-to-head were on clay, at Roland Garros, Madrid, and Rome. The fourth came on grass, at ‘s-Hertogenbosch. It rates a bit faster from 2021-23 than Halle or Queen’s Club by the Serve Impact metric, though it rated as the slowest grass court on tour last year by my older ace-rate algorithm. Maybe it was less server-friendly in 2023, just in time for Humbert to be flummoxed once again.

Surface sensitivity

We tend to take for granted that players are suited to conditions in predictable ways. Big servers like fast surfaces, right? Broadly speaking, yes, but it’s not a hard-and-fast rule. Bounce height makes a difference, footwork matters, and some players are just more comfortable on some surfaces than others.

Armed with surface speed ratings, this is something we can test. If a player is particularly sensitive to conditions, each tournament’s Serve Impact rating should have a predictable influence on his match outcomes. I tried that for all tour regulars, controlling for player strength by using overall Elo ratings at the time of each match.

The resulting numbers are an abstraction on top of an abstraction, so they’re a bit difficult to get your head around. I’ve tried to simplify matters by rendering them in terms of Elo points. A player who is very sensitive to surface and does better on hard courts is, effectively, a better player in faster conditions. The ‘Sensitivity’ numbers given here are the benefit–denominated in Elo points–of each single percentage point that a surface is faster than average. For players who like it slow, negative numbers express the same idea, the Elo-point advantage of a one-percentage-point slowdown.

Here is the list of all players with at least 100 tour-level matches since 2021, plus Rafael Nadal:

Player                       Sensitivity  
Tallon Griekspoor                   11.1  
Ugo Humbert                          9.5  
Richard Gasquet                      9.1  
Novak Djokovic                       8.7  
Adrian Mannarino                     7.9  
Sebastian Korda                      4.9  
Jordan Thompson                      4.3  
Matteo Berrettini                    4.0  
Aslan Karatsev                       3.6  
Tommy Paul                           3.4  
Marcos Giron                         2.9  
Marton Fucsovics                     2.9  
Marin Cilic                          2.6  
Felix Auger-Aliassime                2.1  
Hubert Hurkacz                       1.7  
                                          
Player                       Sensitivity  
Frances Tiafoe                       1.6  
Carlos Alcaraz                       1.4  
Emil Ruusuvuori                      1.3  
Brandon Nakashima                    1.3  
Cristian Garin                       0.6  
Alexander Zverev                     0.5  
Alexander Bublik                     0.5  
Ilya Ivashka                         0.0  
Arthur Rinderknech                  -0.1  
Taylor Fritz                        -0.3  
Jan Lennard Struff                  -0.3  
Lorenzo Sonego                      -0.4  
Mackenzie Mcdonald                  -0.5  
Andy Murray                         -0.9  
Grigor Dimitrov                     -1.1  
                                          
Player                       Sensitivity  
Roberto Bautista Agut               -1.2  
Alex de Minaur                      -1.2  
Karen Khachanov                     -1.4  
Jannik Sinner                       -1.4  
Yoshihito Nishioka                  -1.5  
Miomir Kecmanovic                   -1.9  
Andrey Rublev                       -2.2  
Daniel Evans                        -2.2  
Cameron Norrie                      -2.5  
Holger Rune                         -2.9  
Roberto Carballes Baena             -3.0  
Botic van de Zandschulp             -3.1  
Daniil Medvedev                     -3.4  
Denis Shapovalov                    -3.5  
Sebastian Baez                      -3.7  
                                          
Player                       Sensitivity  
Laslo Djere                         -4.1  
Dusan Lajovic                       -4.1  
Pablo Carreno Busta                 -4.4  
Jaume Munar                         -4.6  
Fabio Fognini                       -4.8  
Nikoloz Basilashvili                -4.9  
Casper Ruud                         -5.0  
Diego Schwartzman                   -5.4  
Francisco Cerundolo                 -5.9  
Alexei Popyrin                      -6.4  
Albert Ramos                        -6.8  
Rafael Nadal                        -9.9  
Alejandro Davidovich Fokina        -10.1  
Stefanos Tsitsipas                 -10.2  
Lorenzo Musetti                    -11.2

There’s Ugo! He’s not quite as surface sensitive as Tallon Griekspoor, but a couple of points is within the margin of error. A sensitivity rating of 9.5 means that Humbert is about 100 Elo points worse in Rotterdam than he is Marseille, as long as I’ve accurately estimated the server-friendliness of the respective playing conditions. Ruusuvuori may also like it faster, but only marginally so; he’s effectively neutral.

Keen-eyed readers may have noted that I earlier referred to “overall” Elo. I’m not using surface-specific Elo ratings here, because I don’t want to adjust for surface twice. Surface-specific ratings already capture some of this: Humbert’s hElo (for hard courts) is 120 points higher than his cElo (for clay courts), which tallies reasonably well with these more fine-grained distinctions. What hElo and cElo can’t tell us, though, is how much his (or anyone else’s) performance will vary on the same surface, depending on the conditions at each specific venue.

It’s easy to get lost in the weeds of Elo-based forecasting calculations, but it’s important to remember they are just tools to help measure a real-world phenomenon. Not every big server is equally at sea on clay; some dirtballers are less dependent on slow conditions than others. Small differences in surface speed are, for most matchups, a minor consideration. But for some players, conditions matter a lot. Ugo Humbert likes his surfaces fast, as much as almost anyone else on tour. In Rotterdam, the conditions did not cooperate.

* * *

Subscribe to the blog to receive each new post by email: