Category: Return stats

Daniil Medvedev’s Instinct For Survival

Daniil Medvedev at the 2023 Italian Open

Clay-court tennis is known for its slow bounces, defensive court positions, and long rallies. Still, a whole lot of points are determined by the bang-bang, plus-one tactics that define the modern game.

The first week of this year’s European clay season was a wake-up call. The champions in Estoril, Houston, and Marrakech were Hubert Hurkacz, Ben Shelton, and Matteo Berrettini, hardly a trio of counterpunching grinders. In Estoril, 70% of Hubi’s serve points ended in four shots or less–and he won 83% of them. In Houston, three-quarters of Shelton’s ended so quickly, and runner-up Frances Tiafoe’s serve points were even shorter. Berrettini finished 77% of his serve points in four shots or less, winning 76% of them. In other words, the Italian won nearly 60% of his serve points with his serve and plus-one alone.

Tournaments since then have settled into something closer to the stereotype. Marton Fucsovics outlasted Mariano Navone in a Bucharest slugfest. The Munich final was decided between two big hitters, Taylor Fritz and Jan-Lennard Struff, but Struff secured the victory with far fewer short serve points than Berrettini and company.

Yet quick points have an outsized effect on clay-court outcomes. When Stefanos Tsitsipas beat Casper Ruud in Monte Carlo, he finished nearly 70% of his serve points in four shots or less–a Hurkaczian performance befitting a server of his caliber. A week later in Barcelona, the relevant number fell to 63%, not much better than tour average on the surface. Stef found himself exposed, fighting out more rallies against one of the game’s best baseliners. He was broken three times and lost in straights.

Fans tend to look at rally-length stats and focus on winning percentage. How did Jannik Sinner fare on points between 0 and 4 shots? Did Carlos Alcaraz win more than half of 10-plus-shot rallies? While these sliced-up winning percentages matter, you can often tell more about a match–including the likely victor–by looking at the frequency of point types. When Berrettini finishes so many of his serve points quickly, his game is working as intended, and he’s probably winning. If he’s spending more time in long rallies, his opponent has more chances to dictate play, hinting at the opposite outcome.

On clay, then, the battle is to survive, to drag the server into a rally. Nobody on tour does that better than Daniil Medvedev.

Octopus on dirt

In the typical men’s clay-court match, 61% of points end in four strokes or less. That’s based on Match Charting Project data since 2015, spanning over 200,000 clay-court points. Here’s how the returner fares in each type of rally:

              Frequency  Win %  
Short (0-4)       61.2%  33.1%  
Medium (5-9)      27.5%  35.3%  
Long (10+)        11.2%  55.8%

The longer the rally, the better the returner’s chances, even if the process is gradual. Five- and six-shot rallies still lean in the server’s direction, though not as much as shorter ones. Ten-plusses are effectively neutral. They look slightly returner-friendly because rallies of exactly ten shots are won by the returner, and that’s the most frequent length in the ten-plus bucket. (If we drew the line at nine or eleven, we’d have the opposite problem.)

Now check out Medvedev:

              Frequency  Win %  
Short (0-4)       52.5%  35.5%  
Medium (5-9)      30.3%  36.1%  
Long (10+)        17.2%  52.3%

The short- and medium-point winning percentages are a bit better, but the real story is in the frequency column. The average match has about 80 serve points for each player. In that time, Medvedev erases about seven short points and adds about five long ones.

In this sense, being a good returner isn’t about cracking return winners or wrong-footing the server. The goal is simply to stay alive. Get the return back, preferably placed well-enough to take away a high-percentage plus-one winner. In last year’s Rome final, Medvedev dragged Holger Rune into long service points almost exactly in line with his career averages: 54% short points, 31% mediums. Rune did just fine through those first nine shots. But when Medvedev reached the ten-shot mark–10 times in 67 Rune service points–he snatched away all but one. Two of those long points gave Medvedev a break for the first set; another 22-shot gutbuster secured the break when Rune failed to serve out the second set.

The Russian’s defense is even more impressive when we compare him to men with better clay-court pedigrees. Here are the top 20 players (minimum 500 charted clay-court return points since 2015) ranked by frequency of short return points:

Player                       Frequency  Win %  
Daniil Medvedev                  52.5%  35.5%  
Diego Schwartzman                53.5%  36.5%  
Rafael Nadal                     54.7%  39.5%  
Alex de Minaur                   54.9%  31.2%  
David Ferrer                     55.1%  34.6%  
Marton Fucsovics                 55.6%  41.8%  
Andy Murray                      55.6%  40.0%  
Novak Djokovic                   55.7%  36.4%  
Gael Monfils                     55.8%  34.7%  
Francisco Cerundolo              55.8%  38.6%  
Stefanos Tsitsipas               56.1%  31.8%  
Jannik Sinner                    56.9%  36.0%  
Jaume Munar                      57.0%  36.5%  
Hubert Hurkacz                   57.1%  30.3%  
Alexander Zverev                 57.5%  34.8%  
Alejandro Davidovich Fokina      58.3%  33.7%  
Sebastian Baez                   58.3%  36.3%  
Gilles Simon                     58.6%  35.9%  
Dominic Thiem                    58.6%  32.5%  
Guido Pella                      58.8%  35.1%

The entire list is packed in a range of about six percentage points, so the full point between Medvedev and Diego Schwartzman–not to mention the two-plus points between him and Rafael Nadal–illustrates just how much of an outlier he is. A low frequency isn’t necessarily better: I’d take Rafa’s combination of frequency and winning percentage over Medvedev, just as I’m sure you would have before reading the first word of this article. But while the Russian doesn’t pick off as many short return points as Nadal, Andy Murray, or Fucsovics(?), his conservatism is hardly a liability. He wins nearly as many as Schwartzman, Sinner, or Novak Djokovic. All this despite a game style tailored to neutralizing the rally further down the line.

The ten-point truth

Medvedev’s long-rally domination of Rune can be misleading. As we’ve seen, he wins about half of clay-court return points that reach ten strokes. Most players do. The benefit of generating long rallies isn’t to sweep the lot: Nobody comes close to accomplishing that, as we will see. The goal is to neutralize rallies. The average server wins 64% of clay-court points, so anything the returner can do to increase the number of 50/50 points is a good deal.

There may be a knock-on effect, as well. Wear out the server, and he might not have as much energy for the next delivery. He might also take more risks in an attempt to end the next points quickly.

The best baseliners don’t need a knock-on effect. Medvedev excels at creating long points, but other men are much better at securing those rallies for themselves. Here are the top 20 among players with at least 100 charted long return points:

Player                   Long Points  Win %  
Kei Nishikori                    141  69.5%  
David Ferrer                     129  67.4%  
Nicolas Jarry                    126  65.1%  
Rafael Nadal                     863  62.9%  
Gilles Simon                     121  62.0%  
Philipp Kohlschreiber            133  60.9%  
Aljaz Bedene                     167  60.5%  
Richard Gasquet                  116  60.3%  
Andrey Rublev                    271  60.1%  
Roberto Carballes Baena          158  60.1%  
Carlos Alcaraz                   381  60.1%  
Botic van de Zandschulp          210  60.0%  
Robin Haase                      142  59.2%  
Sebastian Baez                   284  59.2%  
Borna Coric                      164  59.1%  
Pablo Carreno Busta              203  58.6%  
Lorenzo Musetti                  144  58.3%  
Novak Djokovic                  1099  58.0%  
Alexander Zverev                 793  57.9%  
Juan Martin del Potro            168  57.7%

(Nicolas Jarry?!)

That’s a very different list than what we saw above. The skills required to stretch out a rally are not quite the same as those needed to finish them off. The ideal, then, is a player who balances the two. Kei Nishikori’s win percentage is excellent, but Medvedev is nearly twice as likely to push any given return point to the ten-shot mark. Jarry plays ten-shot rallies on return less than one-third as often as the Russian does.

The key is to think in marginal terms. Longer points work in the returner’s favor, so we can think of every long point as a medium point that the returner successfully extended. The average player increases his chance of winning a rally by 21 percentage points (from ~35% to ~56%) by nudging it from “medium” to “long.” Call that the “marginal value” of a long rally. When we multiply a player’s marginal long-rally value with his frequency of generating long rallies, we get the total payoff of this defensive skill. The average player reaches ten shots about 11% of the time, so their payoff is 21% * 11% = 2.3%. It’s not a meaningful number on its own, but it provides a reference point for individual stats. If a returner’s payoff is higher, they get more benefit than average from their ability to generate long rallies.

Here’s the top 20 (plus a few other players of note), as measured by this combination of long-rally frequency and success rate:

Player                   Long Pts   Freq  MargValue  Payoff  
David Ferrer                  129  14.2%      30.5%    5.3%  
Roberto Carballes Baena       158  14.1%      29.0%    4.4%  
Gilles Simon                  121  14.7%      35.5%    3.9%  
Aljaz Bedene                  167  12.6%      31.2%    3.7%  
Lorenzo Sonego                131  11.7%      23.0%    3.6%  
Pablo Carreno Busta           203  13.0%      31.7%    3.5%  
Robin Haase                   142  11.5%      28.7%    3.5%  
Rafael Nadal                  863  14.4%      39.8%    3.3%  
Novak Djokovic               1099  16.2%      37.5%    3.3%  
Stefanos Tsitsipas            558  13.3%      32.4%    3.2%  
Juan Martin del Potro         168  12.6%      32.1%    3.2%  
Marton Fucsovics              137  16.0%      31.7%    3.1%  
Diego Schwartzman             751  17.3%      37.7%    3.1%  
Alexander Zverev              793  13.0%      36.0%    2.8%  
Daniil Medvedev               394  17.2%      36.1%    2.8%  
Richard Gasquet               116  10.5%      34.0%    2.8%  
Kei Nishikori                 141   9.6%      41.0%    2.7%  
Dominic Thiem                 931  13.2%      34.9%    2.7%  
Holger Rune                   185  11.1%      33.3%    2.7%  
Cameron Norrie                103  10.0%      29.5%    2.7%  
                                                             
Player                   Long Pts   Freq  MargValue  Payoff  
Jannik Sinner                 352  12.9%      38.5%    2.4%  
…                                                            
Andy Murray                   299  12.5%      33.1%    2.3%  
AVERAGE                                                2.3%  
…                                                            
Casper Ruud                   560   9.3%      35.9%    1.7%  
…                                                            
Carlos Alcaraz                381   8.4%      41.7%    1.5%  
Nicolas Jarry                 126   5.6%      38.4%    1.5%  
…                                                            
Stan Wawrinka                 193   8.4%      36.2%    1.4%

We have additional evidence, then, that David Ferrer is the 79th best player of the last century. These numbers might even understate his long-rally prowess, since I’ve limited this analysis to 2015-present. The timeframe probably hurts Nadal as well. Also, there aren’t many long points, so the small sample makes the top of the list somewhat misleading: I’m certainly not ready to take Lorenzo Sonego’s long-rally skills over most of the guys below him on the list.

Caveats aside, we have a plausible estimate of how much value each player reaps from his ability to drag servers into long rallies. Ruud and (especially) Alcaraz are very good past the ten-shot mark, but they don’t get there very often. Medvedev remains our king of negating short service points and creating long ones, but many of his peers are better at working a marathon rally to their own advantage.

No matter how we order the list, the key takeaway is that frequency is as important as win percentage. Returners rarely have a chance to finish points early, so extending the rally is almost always a positive step. Do that a lot, and you don’t have to convert a particularly high rate of those long points. Medvedev doesn’t, and he has become one of the tour’s best players on his least favorite surface. Annoyingly often, he breaks serve simply by putting one more ball in play.

* * *

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

 

Danielle Collins, Destroyer of Second Serves

Danielle Collins at the 2023 Citi Open. Credit: Hameltion

Yesterday in Charleston, Paula Badosa hit 21 second serves. She missed five of them, and it’s easy to see why. With Danielle Collins on the other side of the net, Badosa salvaged just six second-serve points, a 29% success rate. Go big with your second serve, and you’ll rack up double faults. Play it safe, and Collins will destroy you.

The American has never been kind to second serves, and her current hot streak is no exception. With the defeat of Badosa to open her Charleston campaign, Collins has won eight in a row, including upsets of Elena Rybakina and fellow giant-killer Ekaterina Alexandrova. Here are the second-serve return stats for her title run in Miami:

Round  Opponent     W% vs 2nd  
Final  Rybakina           45%  
SF     Alexandrova        67%  
QF     Garcia             50%  
R16    Cirstea            62%  
R32    Avanesyan          82%  
R64    Potapova           86%  
R128   Pera               71%  
       AVERAGE            64%

Defending champion Ons Jabeur, waiting in the Charleston second round, has every reason to be nervous. The last time she faced Collins, in Miami two years ago, the American snatched 69% of Jabeur’s second-serve points.

These numbers are outrageous. The average top-50 player on the WTA tour wins about 56% of second-serve return points. Over the last 52 weeks, Collins is one of only four women to post a mark of 59% or better. The others–Iga Swiatek, Daria Kasatkina, and Lesia Tsurenko–take a different approach, defending with consistency and strategy. No one among the legions of lower-percentage sluggers handles second serves better than Collins does. Jelena Ostapenko is close, winning 58.8% of second-serve return points, though against slightly weaker opposition. Rybakina and Aryna Sabalenka come in under 56%, and Alexandrova–perhaps the freest swinger of all–sits at 54.5%.

Badosa, then, has nothing to be ashamed of. On a typical day, Collins will maul your second serve. At her current level, you might as well be a ball machine set to easy.

Second to none

Collins’s second-serve return skill is rather specific. Many of the game’s best returners–Iga and Kasatkina, Coco Gauff, Jessica Pegula–are better than average against both first and second serves. They excel at handling a variety of serves, and they rack up points when they put balls in play and let rallies develop.

Danielle is different. She’s fourth-best among the top 50 in second-serve points won, but 41st when the same group is ranked by first-serve points won. Match Charting Project data tells us that she is among the worst on tour at putting first serves back in play. Her aggression against attackable first serves prevents her results from being too dire. But she struggles to get the point going when she can’t take a big swing.

Here’s a visualization of each player’s success rates against first and second serves. The relationship between the two is fairly close–much closer than equivalent results on serve–so the trend line from the lower left to upper right is evident. Women located toward the upper left corner, like Collins, are those who are better at returning seconds than firsts. Those toward the lower right, such as Karolina Muchova, are better (relative to average) at handling firsts than seconds.

The plot makes it clear how Collins stacks up against her peers. She cleans up second serves as well as otherwise superior returners, like Pegula and Gauff. Against first serves, she plays more like other big hitters, such as Alexandrova and Rybakina. On return, she is basically the same as Ostapenko, even if her overall approach isn’t as mind-bogglingly aggressive as the Latvian’s.

The first-draft game plan against Danielle, then, is to make some first serves. For the love of all that is holy, make some first serves.

Desperate measures

In Saturday’s Miami final, Rybakina did not do that. The fourth-ranked woman in the world owns what might be the best first serve in the game; the problem is that she doesn’t land many of them. When I wrote about her back in February, she was showing signs of greater consistency. Since then, however, she has reverted to her usual rate of making fewer than 60% of her first serves. Over six matches in Miami, Rybakina cracked 53 aces but found the service box only 58% of the time.

Against Collins, her first-serve rate fell to a measly 53%. That, more than anything else, determined the outcome of an awfully close match. The American earned seven break points, and Rybakina landed her first serve on just two of them. One or two more unreturned firsts at those critical moments, and the story of the final would have been quite different.

Given Collins’s assault on second serves, it is worth asking: Are opponents going about this the wrong way? If the American is relatively weak against harder serves, why not accept more double faults and hit two first serves against her?

Fans have speculated about a “double-first” strategy for years. Back in 2014, Carl Bialik examined its potential, and I followed up a year later. The general conclusion is that two first serves is not a good idea, though for a few women–Victoria Azarenka and Sara Errani among them, in Carl’s analysis at the time–it could have improved their results.

(I say could because we don’t know the knock-on effects of such a radical approach. Carl and I both assumed that if a player hit two first serves, all of their serves would continue to be as effective as before. That might not hold true if returners saw less variety coming from the other side of the net.)

In my follow-up, I found that many individual matches offered opportunities for a double-first attack. It was next to impossible to predict them ahead of time, so it still didn’t make for much of a strategy. But it left open the possibility that there was something to be exploited by skipping second serves altogether.

Collins almost presents such an opportunity. The following scatterplot shows each player in the WTA top 50 and how the double-first strategy would fare against them.

Returners to the right of the line–that is, everybody–are those who would do better against two first serves than against the status quo. Swiatek is an outlier here: Servers would fare almost exactly as well against her if they hit two first serves. Collins is next: Opponents would sacrifice only 0.5 percentage points of the serve win rate if they never hit a second serve.

In Miami, though, Collins’s second-serve return was even more fearsome than usual. Again, it would be difficult to predict specific matches when a double-first strategy pays off, but some of her opponents probably would have accepted more double faults if it meant watching fewer return winners come rocketing back. Here is the same graph, with bubbles added for each of Collins’s Miami matches and another for the average of her Miami opponents:

Sorana Cirstea is tucked in there behind Alexandrova; their service results against Collins were almost identical. Again, points to the left of the line indicate situations where the double-first strategy would have won more points than the way things actually went. It wouldn’t have been wise for Rybakina or Garcia, but Danielle’s other five opponents would have benefited.

The true solution to the Collins conundrum lies in between: some second serves, but more risk-taking all around. I outlined some of those tradeoffs when I wrote about Qinwen Zheng in January. Simply praying for more first serves doesn’t do the trick. With Danielle set to retire at the end of the season, the rest of the tour doesn’t have much time to figure it out.

* * *

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

 

The Clutch Defense of Emma Navarro

Emma Navarro at the 2023 US Open. Credit: Hameltion

The dizzying rise of Emma Navarro continues. She finished 2023 at a career-high 32nd in the rankings, rose to 23rd before Indian Wells, and now, on the back of yesterday’s upset of Aryna Sabalenka, she could crack the top 20 on Monday.

Not long ago, many fans thought of Navarro as a vulture, riding a bunch of small-tournament victories to an inflated ranking. Now, with back-to-back wins over Elina Svitolina and Sabalenka on one of the sport’s biggest stages–and not on clay, her favorite surface–the doubters are quieting down. The American already ranks 19th on the Elo table, another list she’ll continue to climb when this week’s results go in the books.

Yesterday’s triumph was less straightforward than it looks at first glance. The scoreline–6-3, 3-6, 6-2–hides just how close it was. Navarro won just 83 points to Sabalenka’s 80. The second-seeded Belarussian lost the match despite winning return points at a slightly higher clip than her opponent. Sabalenka’s ratio of winners to unforced errors was 38:28, the type of attack that has won her innumerable matches, and one that looks better than Navarro’s 21:16.

The underdog appeared to be the clever, resourceful player on court, making improbable returns and outlasting her more aggressive foe on the long points. Yet the numbers don’t bear out much of that, either. 20 points lasted at least seven strokes, and each player won 10. Sabalenka won five of the longest eight. Navarro’s returning won the day, as we shall see in a moment, but it was not particularly impressive against a far-from-peak Sabalenka. In the last year, opponents have gotten 70% of Aryna’s serves back in play. Navarro managed 67%.

Despite all that, Navarro walked off court with a smile on her face. What worked?

Timing is everything

The top-level answer is that Navarro converted break points, and Sabalenka didn’t. The underdog seized four of her five chances. In each game that she generated a break point, she secured the break. Sabalenka, on the other hand, earned more opportunities but took advantage of just two. She squandered a chance to put the first set back on serve at 5-3, and she could have erased Navarro’s break advantage at 3-1 in the decider. In neither of those games did the American slip again.

Break points, like points in tiebreaks, tend to be more complicated than average. Servers are a bit more careful to put balls in the court–and thus more conservative–and returners are hyper-focused. A high-pressure point is less likely to end with an unreturned serve; long rallies are more common. Navarro–with some help from her opponent–took this to an extreme.

The American, despite putting slightly fewer serves back in play than Sabalenka’s average opponent, kept the point going on each one of her five break points. She also returned every serve at 15-30, three of four at 30-30, and both at deuce. Here’s how Sabalenka’s rate of unreturned serves looks when separated by whether she was in trouble–defined as whether Navarro had already won two points in the game:

Situation       Points  Unret  Unret%  
Not In Trouble      51     16   31.4%  
In Trouble          22      4   18.2%

Even that distinction understates things. Two of Sabalenka’s unreturned serves in the “in trouble” category came at 40-30. When the second seed was really on the ropes, Navarro got the ball back on 15 of 17 tries. Pressure points are less likely to end quickly, but not by such an enormous margin.

Whether Sabalenka became uncharacteristically shaky under pressure, Navarro morphed into a return savant, or it was pure dumb luck, those few points determined the outcome of the match. In extended rallies, as we’ve seen, the American was not the overwhelmingly superior player, but that’s not the point. Sabalenka dominates most of her opponents by winning more cheap points than they do. If she wins just half of the rest of the points–on her serve and her opponent’s–she comes out on top. Take away most of the cheap points, and her edge is gone. Navarro won 24 of 49 return points–roughly half–when she put the ball back in play. Because she was so resourceful at key moments, she held Sabalenka to just 54% of serve points at 30-all or later.

Lessons

There are so many ways this match could have ended differently. Sabalenka could’ve served a little better under pressure, or Navarro could have returned a little worse. The whole scenario was made more likely by the conditions, slow-playing Indian Wells courts and balls, combined with wind that distracted the favorite more than the underdog.

Another culprit on the Belarussian side was Aryna’s plus-one. Eleven of her unforced errors came on the first shot after her serve, many of them wild and inexcusable, one of them two points away from defeat. It is to Navarro’s credit that she got so many serves back, but a more typical Sabalenka performance would have put away more of the desperate returns.

This is all a description of what happened last night, not speculation about a trend, or any kind of prediction. Sabalenka usually hits about as many unreturnable serves in pressure situations as she does at other times. In the limited data we have so far on Navarro, there’s no evidence that she is much better returning at key moments. Clutch performance in tennis is only rarely persistent: It’s easy to identify matches or tournaments when a player was particularly good or bad when it mattered most, more or less impossible to forecast it. If we hit rewind and replayed the match from the start, Navarro might still pull the upset, but it wouldn’t develop the same way.

What the match does give us is a little more evidence that Navarro is here to stay. She drew even with the second-best player in the world, staying calm enough throughout the proceedings to deliver her best tennis when the stakes were highest. She might not win a rematch with Sabalenka, but her position in the top 20–whether or not the WTA makes it official next week–is no fluke.

* * *

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

 

The Downward Slide of Stefanos Tsitsipas’s Backhand Return

Stefanos Tsitsipas ahead of the 2023 US Open. Credit: Hameltion

Yesterday in Indian Wells, Jiri Lehecka knocked out Stefanos Tsitsipas with a masterclass of precision power hitting. The Czech tallied 27 winners to Tsitsipas’s 8, and that’s only after a belated burst of energy from the Greek in the second set. When I wrote about Lehecka in January, I chided him for an “excess of self-restraint,” hitting too many balls down the middle to take full advantage of his baseline weapons. He avoided that trap yesterday, and Tsitsipas paid the price.

Still, Lehecka didn’t seize upon every exploitable edge. The Tsitsipas backhand, pretty as it is, is a bit cumbersome, often leaving him slow to react. The time he needs to unleash the one-hander is a key reason why Stef is one of the most surface-sensitive players on tour, preferring courts that give him an extra split-second to prepare. Indian Wells plays slow for a hard-court event, but it’s hardly the same as the Monte-Carlo dirt. Combined with Lehecka’s power, a hard court has the potential to expose the Greek’s weaker side.

Lehecka generally stuck to the routine, sending both forehands and backhands cross-court, rarely doing anything in particular to force his opponent to hit a backhand. Still, the Czech made one concession to his opponent’s tendencies: He hit more serves than usual to the backhand. Lehecka’s favorite serve is the deuce-court slider wide, and he scored a few points blasting balls that Tsitsipas’s forehand couldn’t reach. But more often, he forced Stef to hurry the backhand, or chip a return.

The results were devastating. The Czech typically hits about half of his serves to each wing. 51% of his first serves and 52% of his seconds go to the backhand. Yesterday, he went that way 61% and 58% of the time, respectively. When Lehecka hit a first serve to the backhand, Tsitsipas got it back less than 60% of the time… and he won only 12% of those points. When a second serve went to the backhand, Stef got a more respectable 71% back, but still won just 14%. It was a short match, so we’re not talking about that much action: In nine return games, Tsitsipas won just four points when he had to make a backhand return.

The word is out–not that it was ever really a secret. Two weeks ago in the Acapulco quarter-finals, Alex de Minaur hit 90% of his serves to Tsitsipas’s backhand, beating the Greek for the first time at tour-level in eleven tries. While de Minaur’s persistence was unusual, it made a useful point. Stef–not long ago the third-ranked player in the world–has fallen out of the top ten, and one of the causes is something that every opponent can attack until he patches it up.

Serve this way

Yesterday’s match notwithstanding, Tsitsipas remains an elite server. He held 88.8% of his service games last year, the best mark of his career. His return numbers, though, are sinking. In 2021, he broke one-quarter of the time; last year, that number fell to 19.3%. By that metric, he’s in the bottom third of the ATP top 50.

It’s hard to pinpoint just one element of his return game that has gone astray, because everything is trending downwards. According to Match Charting Project data, he gets 66% of serves back in play–a below-average if acceptable figure–but wins just 42% of those points, one of the lowest marks on tour. Both rates are worse than his career averages of 68% and 44%, respectively.

The decline can’t be entirely blamed on a tour-wide tactical consensus, though Stef’s opponents aren’t helping. Here is a ten-match rolling average of the percentage of first serves hit to the Tsitsipas backhand on hard courts since late 2018:

(Clay courts add another wrinkle to the question, because everyone tends to get more serves to the backhand side on dirt. Four of Tsitsipas’s five most extreme matches by this metric were on clay.)

As always, it can be tough to gain an intuition with an unfamiliar tennis metric. 60% may not sound like a lot, but most servers cluster around the tour average of 52.5%. The servers who most frequently target the backhand side are clay-courters: Albert Ramos tops the list at at 63.5%, with Fernando Verdasco right behind him. At the other extreme, around 45%, are some right-handers, often those who can use height and/or wrist action to open the court with lots of slice. Daniil Medvedev and Andy Murray are two of the best-known proponents of this approach, and Medvedev is partly responsible for some of the troughs in Tsitsipas’s trend line above.

More to the point, the tour is taking aim at Stef’s backhand as much as it does anyone’s. Only Ivo Karlovic was faced with more backhand returns on hard courts. Denis Shapovalov, another one-hander, is in the same range. Again, the message about Tsitsipas’s backhand isn’t new, but it’s no accident that servers are picking on it at the same time that his return numbers take a nosedive.

Crash

Anyone who doesn’t attack the Greek’s backhand return these days is leaving points on the table. Here is another ten-match rolling average, this one showing Tsitsipas’s rate of return points won when his opponent lands a first serve to his backhand:

If you’ll forgive me some technical jargon, that’s… not good.

As we’ve seen, Tsitsipas won just 12% of those points against Lehecka. He won only 14% against Taylor Fritz in Australia and 16% in Los Cabos against Casper Ruud. Daniil Medvedev held him to 11% last fall in Vienna. And at the Tour Finals, Jannik Sinner hit 21 first serves in that direction. Tsitsipas won none of them.

Across 125 career charted hard-court matches, Stef has won 23.1% of return points on first serves to his backhand and 24.7% to his forehand. Since the start of 2023, those numbers have fallen to 20.6% and 23.4%. Every important return stat is trending downward, and the backhand numbers are declining fastest of all.

The only question remaining for Tsitsipas’s opponents is this: How much is too much? De Minaur set a new standard by going to the backhand with 90% of his serves, both first and seconds. That’s not unheard of on clay courts (Lorenzo Musetti has come close in two previous meetings with Stef), but it’s very unusual on a hard court. Only a handful of Tsitsipas’s opponents have topped even 70% in one direction.

Against most players, such a balance is probably appropriate. Too much to one side, and you lose the element of surprise. But because so much serving is split 50/50 (or 53/47), we don’t have much data to test the hypothesis. If there’s a 30% chance the server will go one way, will a returner really have an edge in the more likely direction? Against de Minaur, Tsitsipas figured things out quickly enough and inched over toward the backhand side. But not for long: The Aussie cracked one of his few forehand-side serves for an ace, far out of Stef’s reach. Tsitsipas is a good test case for servers looking to experiment: His success rate when he gets the return back in play is near the bottom of the table, so he’s unlikely to turn a match around just because he guesses right a few times.

I have no idea whether, at this point in the Greek’s career, his backhand return is something that can be fixed. In the short term, it will be easier for opponents to expose it than it will be for him to find a solution. Tsitsipas’s return numbers, already dire, could get worse before they get better.

* * *

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

 

Yuan Yue Will Return Your Serve Now

Yuan Yue at Wimbledon last year. Credit: si.robi

Here’s an impressive stat for you: Last week in Austin, champion Yuan Yue won more than half of the return points she played. In fact, had she picked up just one more point against Wang Yafan’s serve in the quarter-finals, she would have won at least 50% of return points in each of the five matches she played.

This isn’t earth-shaking stuff: There are about a dozen tournaments every year where the champion wins more than the 51% of return points than Yuan did in Texas. Iga Swiatek won 56% at the French; Aryna Sabalenka cleared 52% in Australia. Lauren Davis won 53% in Hobart last year. The average single-match loser on the WTA tour loses about half of their return points, so it’s not far-fetched that a titlist would rack up these numbers for five or six days running.

Still, this is Yuan Yue we’re talking about. Not only was the 25-year-old Chinese woman a longshot to win the title–it was her first on tour–she has hardly established a reputation as a steady returner. A dangerous one, perhaps: In last year’s Seoul final against Jessica Pegula, Yuan turned one out of six of the American’s deliveries into a return winner or forced error. But the overall results weren’t so impressive, as she won fewer than 40% of return points in the match. For all of those big swings, there were lots of swings and misses. Out of five matches in an otherwise encouraging week in Seoul, she won more than 46% of return points only once.

The game Yuan brought to Austin was something different. Like San Diego champ Katie Boulter, she took fewer risks than usual, trusting that she could win points a shot or two later. Facing Pegula, and in another losing effort to Emma Navarro in the Hobart semi-finals, Yuan’s average return point lasted four strokes. Against Wang Yafan on Friday, return points took six. Yuan hit just three return winners in that match; in the final against Wang Xiyu, she didn’t hit any. Presumably she’s ok with that.

The magnitude of Yuan’s achievement isn’t quite the same as Boulter’s: The Brit beat five opponents ranked in the top 40, and Yuan didn’t face anyone in the top 60. Yet the week marks a major step forward. The 25-year-old cracked the official top 50, and Elo now rates her as the second-best Chinese woman on tour. If she continues to put returns in play the way she did in Austin, she could climb even higher.

One more ball

Returns in play are good, but they come at a cost. Do you aim to stay in as many points as possible, accepting that a lot of your returns will be weak, or do you swing big, piling up errors in exchange for a handful of return winners and better odds when your returns find the court?

While the pros and cons are different for every player, no one escapes the tradeoff. Even across players, there is a persistent negative correlation between returns in play and in-play returns won. If you make more, you win fewer of them. The following plot shows those two numbers for every woman with at least five matches in the Match Charting Project dataset from the last 52 weeks:

The best place to be is the upper-right corner, with a lot of returns in play and a lot of those points won. Except… that sector is mostly empty. The women who get the most serves back–Kasatkina, Avanesyan, Sorribes Tormo–win those points at an average rate. Even that success rate is boosted a bit by the slower courts where those players tend to succeed. By contrast, the players who win the most in-play return points–Swiatek, Ostapenko, Yastremska–achieve that by missing a lot, or by being an all-time great. Even Swiatek doesn’t put an above-average number of returns in play.

With our new sense of what these numbers mean, let’s take another look at Yuan’s step forward. The limited data at hand includes five of her matches from before this week, which we can compare to the quarter-final and final from Austin. Here’s the same graph, but with points added for Yuan’s sample of previous matches (Yuan-Prev) and for the two in Austin (Yuan-ATX):

Um, yeah. Wow. There are caveats, of course: It’s just two matches, and neither Wang is a particularly stellar server. (On the other hand, Yuan’s previous opponents were middling servers as well.) If this shift is even a little bit sustainable, Yuan will no longer be just a fringe figure on tour.

Runaway momentum

While we’re extrapolating from too-small sample sizes, I’ll give you another one, one that doesn’t paint such a rosy picture for last week’s champion.

Break points go to the returner more often than return points in general, because more break points are generated against weaker servers (or by stronger returners, or both). The women currently ranked in the top 50 win 44.4% of their return points, and they convert 46.3% of break points.

Yuan, in 33 tour-level matches since this time last year, has won 44.3% of her return points, but only 43.2% of her break point chances. A gap of three percentage points (between 43% and the expected 46%) is statistical shorthand for too many missed opportunities. I checked those numbers only because the Austin champ, in both the quarters and the final, showed signs of letting momentum get away from her. Against Wang Yafan, she got broken right after securing the first set, then struggled to regain the advantage. In the final, she served for the title at 5-2 in the second set, dropping serve twice before finishing the job in a tiebreak.

As I say, these are small samples. We tend to ascribe too much importance to hot and cold streaks–they would arise even if every point were decided by a roll of the dice. (I suffered through an epic Chutes and Ladders slump yesterday, probably because of the clutch play of my four-year-old opponent.) Still, there’s some evidence that Yuan struggles under pressure, even if she overcame it several times last week.

In this context, there’s a bit of negative spin we can put on all those returns in play. I’ve written before that momentum (and clutch, and streakiness, all that stuff) is tough to measure in tennis because the structure of the sport is anti-streak. If you hit a good serve in the deuce court, you have to hit one in the ad court. Four aces in a row? Congrats, you get to do something else now–you might even have to sit down for a couple of minutes. And that’s to say nothing of your opponent’s ability to give you shots other than the ones you’re hitting well.

But against her compatriots last week, Yuan inadvertently created conditions in which streaks could take root. All those returns in play–combined with a fair number of longer points that developed on her own serve–reduced the separation between serve and return. From 5-2 in the second set of the final, Yuan’s backhand went awry, and there was little she could do to avoid it. Her own serve wasn’t imposing enough to end points quickly, and she was out of the habit of taking big cuts on return. It was easy to get into a rut.

The momentum eventually shifted, of course. Yuan won 12 of the last 15 points of her quarter-final, and in the final, she won 10 of 12 points from 5-6 in the second set to reach 6-1 in the tiebreak. She pried herself out of one pattern and immediately found a different one, one that still largely avoided short points but ended in her favor. Yuan’s conservative returning paid off on paper, but the unending string of long(ish) points may have made it harder for her to regain control on the few occasions that she lost it.

Yuan’s fellow champion last week, Katie Boulter, already stumbled at her next obstacle, losing a straight-setter yesterday in Indian Wells to Camila Giorgi. Yuan’s first test in the desert is Varvara Gracheva, a middling server who could prove susceptible to the Chinese woman’s improved game. Next up would be a tantalizing second-rounder with China’s number one, Qinwen Zheng. Zheng’s intimidating–if erratic–serve could tell us a lot more about just what Yuan is now capable of.

* * *

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

 

Katie Boulter’s Game of Patience

Katie Boulter in 2021. Credit: Chris Czermak

You can be forgiven if you didn’t have Katie Boulter on your radar going into the 2024 season. Her career-best ranking in the top 60 was a bit misleading, stemming largely from a debut title on the grass at Nottingham, where she picked up the championship without facing a single top-130 player. Nothing she can do about that, of course, but when it came to things she could control, the results were not always so encouraging. She wrapped up 2023, at 27 years of age, with a career total of just 25 hard-court victories.

Since then, it’s been a whole new Katie. She opened her campaign with an upset of 5th-ranked Jessica Pegula at the United Cup, and counting qualifying, she has won at least one match at six straight events. To solidify her new status, she won the title in San Diego last week, knocking out four seeded players in the process. Nothing cheap here: All five of her opponents were ranked in the top 36, and the weakest among them–Lesia Tsurenko–had won each of their three previous meetings.

After years in the ITF wilderness, relying on British grass-season wild cards for tour-level appearances, Boulter has earned a place in the WTA top 30. Thanks to the giant-killing in Southern California, her Elo rating backs it up, placing her 28th on tour, 26th on hard courts. The Brit has always had a first serve and forehand that made it possible to dream big. It has just been a matter of harnessing the big weapons and filling out her game with complementary parts that wouldn’t leave her exposed.

The results from San Diego suggest that she has finally done so. Commentators like to talk about the importance of playing “within” oneself, waiting for opportunities instead of swinging for winners when none are on offer. Boulter hasn’t always done that, and she has the unforced error totals to prove it. She found a better balance against last week’s barrage of top-40 foes, playing a bit more conservatively and cutting down mistakes without sacrificing too many winners. Now, she’s on everyone’s radar.

Katie bolt the door

Boulter ascended through the ranks with what might best be described as low-percentage tennis. In her 100+ matches on tour, she has landed just 57% of her first serves, racking up more double faults than aces. 13% of her groundstrokes are unforced errors, well above tour average. Her typical return point lasts just barely three strokes.

In twelve matches logged by the Match Charting Project before last week’s run in San Diego, a mixed bag ranging from ITFs in 2017 to her final-round victory over Jodie Burrage in Nottingham last year, her Rally Aggression Score worked out to 84, putting her in the same range as free swingers such as Caroline Garcia and Camila Giorgi. Against Burrage, her Rally Aggression Score was 267, a remarkable figure for a stat that I initially scaled so that it would max out around 100. Just about every possible shot–plus a few impossible ones–was a winner or unforced error. The average point in that match lasted just 2.9 shots.

It’s possible to win with numbers like these–just ask Aryna Sabalenka, Petra Kvitova, or Jelena Ostapenko. But the margins are slim. It’s particularly tough to thread the needle while making as few first serves as Boulter does. Surprisingly, she struggled on that front in San Diego, making just 54% of her first deliveries for the week, including a mere 51% against Emma Navarro in the semi-final. When only one-quarter of total points begin with your best shot, you can’t risk missing so much with the rest of your arsenal.

The Brit’s most effective step forward, especially against Kostyuk in Sunday’s final, was to keep herself in more return points. She has traditionally swung big on return, ending 11% of second-serve points with a return winner or forced error, and even finishing 6% of first-serve points that way. She didn’t sustain those numbers in the final, but the tradeoff was worth it. In those previous 12 charted matches, she put just 60% of first-serve returns in play, compared to a tour average closer to 75%. Facing Navarro, she kept the point alive on 65% of first serve returns; against Kostyuk, she put the ball back in play 79% of the time.

The result, more or less by definition, was longer return points. While the Burrage match was extreme, it is not uncommon for Boulter to average in the neighborhood of three strokes per point on return–still in Ostapenko territory. The San Diego semi was a wholly different animal, with her average return point lasting 4.5 shots. In the final, it was 4.1 shots. Such a radical shift would be a bad sign for many big hitters: Ostapenko and her ilk end points early because that’s the way they want to play. Yet Boulter thrived on her newfound patience.

The 27-year-old won 47% of her return points in San Diego, five percentage points better than her tour-level average and about the same as a typical performance from Pegula or Coco Gauff. On Sunday against Kostyuk, Boulter played 81 return points and won 40 of them.

Winners, slightly delayed

Even though she did not smack as many return winners as usual over the weekend, Boulter still won plenty of points from the baseline. She just took fewer risks in the process. Excluding aces and double faults, here’s a top-level comparison of the frequency with which she hit winners (plus forced errors) and unforced errors in her last two matches, compared with her previous charted matches and WTA average:

Match(es)    W+FE  UFE  
vs Kostyuk    14%   9%  
vs Navarro    19%   8%  
Previous      16%  13%  
WTA Average   12%  10%

The Navarro match was an unalloyed triumph, as you might expect from the 6-3, 6-1 scoreline. But even the Kostyuk numbers point to a major step forward. It’s not easy to cut down unforced errors from one side of tour average to the other, especially against someone like the Ukrainian, who feeds opponents an ever-shifting mix of speeds and depths. In the past, the Brit might have lost her rhythm and gone on an error spree. Instead, she was even steadier. On 28 key return points–game point, deuce, or break point–Boulter committed just one unforced error.

Facing women without the weapons to quickly end points themselves, Boulter discovered that a winner on the fourth or fifth shot of the point is just as effective, and less risky, than a big swing on the second or third shot. In her previous charted matches, she won about 49% of points that lasted four to six shots. Against Kostyuk, she won 76%, dictating play the way she has always preferred to do, just one or two shots later.

Presumably, it won’t always be quite this easy. Boulter won’t keep her unforced error rate in single digits forevermore; more aggressive opponents will tempt her into playing her old game. (Though the Brit did straight-set Donna Vekic in the quarters.) Still, she appears to have discovered new capabilities, seven years after her first appearance at Wimbledon. This year, she’ll likely return to the All-England Club as a seeded player, with all the expectations and pressure that entails. A decade into her pro tennis career, Boulter has earned herself a lot more attention.

* * *

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

 

How To Play One-Set Shootouts Like Daniil Medvedev

Daniil Medvedev in 2023, practicing… something. Credit: Hameltion

In yesterday’s Australian Open quarter-final match against Hubert Hurkacz, Daniil Medvedev came through with his second five-set win of the tournament. In the decider, Hurkacz’s level dropped, Medvedev kept his ground game tight, and the Russian converted the one break point on offer. Four hours of tennis, compressed into a few crucial moments, and Medvedev has a place in the semi-finals.

Not long ago, Medvedev gained a reputation as a disappointment in deciding sets. He lost 11 of 15 three- and five-setters in 2022, and yesterday’s match was the first time in nine tries–going back to Melbourne two years ago–that he had beaten a top-ten player in a climactic set.

But such trends are easy to exaggerate. For one, three of those eight consecutive losses were clustered at the 2022 Tour Finals, where the Russian managed, remarkably, to drop third-set tiebreaks in all of his round-robin matches. Not the best way to ensure a restful offseason, but hardly an indictment of his ability to hang around late into matches with the best players in the game.

Further, except for the 2022 season, Medvedev has developed a knack for cleaning up close matches with everybody else:

Year   Decider W-L  Decider W%  
2024           2-0      100.0%  
2023          14-6       70.0%  
2022          4-11       26.7%  
2021          14-5       73.7%  
2020           9-4       69.2%  
2019         10-11       47.6%  
2018          16-9       64.0%  
2017          13-6       68.4%  
2016          23-9       71.9% 
---- 
Total       105-61       63.3%

2016 shouldn’t really count, since it’s a mix of ITFs, Challengers, and early forays onto the main tour, but given the results, I figured it was worth including. Wherever you draw the line, it’s hardly the case that Medvedev struggles in such matches. Recently, I looked into what a player’s third-set record “should” be, given their skill level, and a mark above 60% is better than expected for nearly anyone.

You might argue that the Russian shouldn’t have racked up so many deciders. He was expected to finish off Emil Ruusuvuori much more quickly than he did in the second round in Australia, and even on clay, he should never have gotten dragged to a fifth set at Roland Garros by Thiago Seyboth Wild, much less lost it. But everyone takes the scenic route sometimes. 14 of Medvedev’s deciding sets last year came against the top 50, 10 of them against the top 20.

The final set shift

When a match is reduced to a one-set shootout, it becomes a bit less serve-centric. This is a persistent finding in all high-pressure situations, from tiebreaks to break points to fifth sets. Servers get a bit more cautious, returners heighten their focus, and quick points are harder to come by.

The effects are small but real. In the 1,200-plus men’s deciding sets since 2017 logged by the Match Charting Project, servers win 1.1% fewer points in the final set that they did in the first two or four. They land fractionally more of their first serves, but only by increasing their margins: The percent of unreturned serves falls by more than 5%. The average rally increases from 4.1 strokes to 4.3.

There are two fundamental ways to benefit from those changes. First, you can buck the trend, continuing to serve big while your opponent succumbs to the natural tendency toward caution. That’s part of the reason that John Isner and Roger Federer were two of the very few players to win more tiebreaks than expected over long periods of time. It’s not easy, especially if fatigue is setting in. But if you can keep serving the way you did for two or four sets, you have a minor edge in the decider.

Second, you can be the type of player who excels in deciding-set-style tennis. If you had to pick between Medvedev and Hurkacz in a contest where more serves would come back and points would last longer, the choice is simple, right? It’s no guarantee, to be sure: The shift is a minor one, and it may not show up in any given match. Yesterday, more points were decided in four shots or less in the fifth set than in the first four. But on average, the trend moves in the other direction, right into the Russian’s wheelhouse.

Evidence shows that Medvedev follows these prescriptions, maintaining his attack on serve while taking advantage of more cautious opponents. Other top players, to varying degrees, do the same.

Let’s start with the basics. For each stat, I calculated every player’s performance in deciding sets, and in all previous sets. The numbers I’m about to show you are the ratio between those numbers, a measure of how much their tactics change when the final set begins. Positive numbers mean they do more of it in the decider, negative means they do less. We’ll look at the four Australian Open semi-finalists, plus Carlos Alcaraz (because of course) and Hurkacz (because of his deciding-set notoriety). Keep in mind that Novak Djokovic’s figures are limited to matches since 2017.

Here are the rate of serve points won, and the rate of first serves in:

Player             SPW%  1stIn%  
Carlos Alcaraz     3.9%    4.4%  
Jannik Sinner      2.6%   -1.2%  
Novak Djokovic     1.5%   -1.1%  
Hubert Hurkacz     0.8%   -1.9%  
-- Average --     -1.1%    0.7%  
Daniil Medvedev   -1.2%   -1.7%  
Alexander Zverev  -4.5%    3.2%

Medvedev is in line with tour average when it comes to winning service points: He doesn’t hold on to as many in deciding sets. Average isn’t bad in this case, though it looks mediocre in this company. A more encouraging sign, at least in terms of the tactical approach, is the change in first serves in. The Russian, in line with Djokovic, Hurkacz, and Jannik Sinner, seems to take a few more chances in the shootout. Alcaraz defies gravity, serving more conservatively yet winning more points, and Zverev looks out of place, a caricature of prudence.

Now let’s look at the percentage of serves that don’t come back (Unret%), as well as the percent of service points won in three shots or less (SPW% <=3):

Player            Unret%  SPW% <=3  
Novak Djokovic     10.9%      5.4%  
Carlos Alcaraz      0.2%      1.0%  
Daniil Medvedev    -0.6%     -2.0%  
Hubert Hurkacz     -1.1%      0.2%  
-- Average --      -5.7%     -3.6%  
Jannik Sinner      -7.4%      0.3%  
Alexander Zverev  -13.4%    -11.2%

The first rule of writing about men's tennis: Whatever the topic, you'll eventually end up showering praise on Djokovic. In recent years, he has learned how to get more out of his serve, and he turns that knob even further in deciding sets. Most players struggle to simply stay above water in the final set; Djokovic starts serving bigger.

Medvedev's rate of unreturned serves is the sort of positive sign it takes a connoisseur to appreciate: "-0.6%" doesn't turn up on many Hall of Fame plaques. But when the typical player serves so much more carefully, the Russian's consistency works to his advantage. His three-shots-or-less win rate does not stand out as much, but it is still less of a step backward than the typical tour player takes.

Once again, deciding-set Alexander Zverev is an unusual beast.

Opportunistic returning

If the challenge on serve is to keep attacking in the final set, the task on return is to take advantage of an opponent who probably isn't doing that. Ideally, that might mean more aggression on the return, but a 1% or 5% weaker first serve is still only so playable. Instead, players should make sure not to squander the chances they're given: Make more returns, then tighten up the ground game for the inevitable rallies.

Here are three stats to illustrate deciding-set return tendencies, again expressed as ratios between how each player performs in the final set, compared to previous sets:

Player            Ret InPlay%  UFE/Pt    FH%  
Alexander Zverev         6.7%    1.1%   1.0%  
Daniil Medvedev          3.9%   -3.2%  -1.2%  
Novak Djokovic           3.0%  -10.5%   1.5%  
Hubert Hurkacz           2.9%   -1.7%   0.3%  
Carlos Alcaraz           2.7%  -10.4%  -1.9%  
-- Average --            2.5%   -2.4%  -0.3%  
Jannik Sinner           -1.2%    0.1%   0.4%

Zverev, as we might have guessed, gets a lot of deciding-set returns in play. He's exceedingly conservative by every other measure we've seen, so why not here? Behind him, heading the non-pusher category, is Medvedev, who gets nearly 4% more returns in play in the final set that he did up to that point.

Unlike Zverev, the Russian also stays in control throughout the rally. He doesn't suddenly discover the otherworldly control of Djokovic and Alcaraz, who somehow reduce their unforced error rates by 10% in the deciding set, but he leads the rest of the pack, cutting down his mistakes by more than the tour average.

The third metric shown here--forehands as a percentage of all groundstrokes--is simply a curiosity. There's no right or wrong way to choose strokes, at least not at the level of the whole tour. As we saw last week, Medvedev and Zverev go for backhands on the plus-one shot more than anyone else, because they are in the unusual position that it might really be their stronger option. If a player improves his ground game in the fifth set--and this is nothing more than a hypothesis--it might show up in the numbers as more shots from his preferred wing. None of these men show a dramatic shift in shot selection, but I can't help but notice that Medvedev hits a few more backhands in the final set than he did in the two of four sets it took to get there.

If Medvedev reaches a fifth set in tomorrow's semi-final against Zverev, he won't need this level of savvy to know what's going on. The German's tactics, whether by design or instinct, are abundantly clear. Zverev can turn a shootout into a war of attrition, with two fifth-set tiebreaks already in Melbourne and an astonishing record of 22 deciding sets won in his last 26 attempts. While it will doubtless be a grind, the Russian might just be able to use his opponent's passivity against him. Faced with the tiny margins of a grand slam fifth set, every edge is worth exploiting.

* * *

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

 

How Coco Gauff Escaped a Trap of Her Own Making

Also today: Jannik Sinner’s near-unbreakability

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

Coco Gauff is not a pusher, but she can do an awfully good impression of one. In yesterday’s Australian Open quarter-final against Marta Kostyuk, the American coughed up 50 unforced errors against just 17 winners. The average rally lasted 4.6 strokes, a modest number that was rescued from marathon territory only by Gauff’s many unreturned serves.

Coco’s forehand, the usual culprit when things get messy, was on full display yesterday. While the stroke has shown signs of improvement–only 9% of them contributed to the unforced error tally, below both tour average and Gauff’s own standard–it remains loopy, and it gets ever-more cautious under pressure. Kostyuk was willing to go after the high-bouncing mid-court groundstrokes, often putting Gauff on the run. Fortunately for the American, her defense rarely deserts her. She eked out a three-hour, 7-6, 6-7, 6-2 victory for a place in the Melbourne semi-finals.

My impression watching the match was that Gauff put an unreasonable number of returns–especially forehand returns–in the middle of the court, not too deep, and that Kostyuk was punishing them. I was partly right: The Ukrainian forced Coco to hit forehand after forehand against the serve, more than two-thirds of her service returns all told. Gauff did indeed send more of those balls down the middle, closer to the service line than the baseline. And Kostyuk attacked… but to little avail.

Let’s get into the numbers. The Match Charting Project divides the court into thirds, both in terms of direction (forehand side, backhand side, and middle) and depth (shallow [in the service boxes], deep [closer to the service line than the baseline], and very deep). All else equal, shots deep and/or to the sides of the court are better, though of course they are riskier. Some returns will inevitably end up down the middle and shallow; the goal is only to avoid it when possible.

Here is how Gauff’s performance yesterday compared to tour average and her own typical rate of service returns that went down the middle and didn’t land close to the baseline:

RETURNS          Middle/Not Very Deep  
Tour Average                    34.0%  
Coco Average                    40.5%  
Coco vs Kostyuk                 43.7%

Indifferent return placement is nothing new for the American, and she left even more hittable plus-ones for Kostyuk than usual. It wasn’t as bad as last year’s US Open final against Aryna Sabalenka, when Gauff put more than half of her returns in the less effective zones, but Kostyuk is no Sabalenka when it comes to imposing her will with the serve.

Return placement matters. On average, tour players win 46% of points when they land a down-the-middle, not very deep return. When they put the ball anywhere else–closer to the baseline or a sideline–they win 56%. Gauff is a little better behind the weak returns, but for her career, the gap is still present: 47% versus 55%.

Except… that isn’t what happened yesterday!

RETURN OUTCOMES  Mid/NVP W%  Better W%  
Tour Average          46.2%      56.3%  
Coco Average          46.9%      54.8%  
Coco vs Kostyuk       60.0%      55.2%

When Gauff placed a return near a line, her results yesterday were typical. But Kostyuk was unable to capitalize on the rest. Among 88 matches logged by the Match Charting Project, Gauff has won 60% of those middle/not-very-deep returns only a dozen times, usually in blowouts.

Judging from the American’s performance on return, she could have made quick work of yesterday’s contest, too. The sticking point came on her own side of the ball, where her non-committal forehands didn’t work out as well.

Minus-ones

On the WTA tour, when the return lands in play, the server has nearly lost her advantage. A good first serve can give her a lingering edge, or a well-placed return can tilt the balance in the other direction, but overall, the point begins again as a neutral proposition. Servers win 52% of those points.

Gauff, on average, does a little better, converting her serve 53% of the time. There are signs she’s improving, as well. In the US Open final against Sabalenka, she won 55%, and in the Auckland final this month versus Elina Svitolina, she picked up 59%. Apart from lopsided matches, the high-50s are the best anyone can do on an ongoing basis: Iga Swiatek’s average is 57%, and Sabalenka’s is 55%.

Coco won 39% against Kostyuk.

Gauff’s lack of confidence in her forehand showed up in multiple ways. First, she didn’t use it as much as a plus-one weapon. She usually hits 57% of her plus-one shots from the forehand side, in line with tour average. Yesterday, that rate was just 51%, something that had more to do with her own choices than any return magic that Kostyuk conjured up.

Then, she didn’t do much with those forehands. The following table shows plus-one forehand rates (3F%), the percentage of plus-one forehands hit down the middle (FH Mid%), and the server’s winning percentage (FH Mid W%) behind those down-the-middle forehands:

PLUS-ONES          3F%  FH Mid%  FH Mid W%  
Tour Average     56.6%    29.9%      45.9%  
Coco Average     57.2%    35.0%      47.0%  
Coco vs Kostyuk  50.7%    39.5%      40.0%

Gauff magnified her own tendency to go back down the middle with her second-shot forehand. It didn’t work, as she won just 40% of those points, compared to her typical rate of 47%.

Even beyond the plus-one, Coco just kept pushing the forehand. She went down the middle with 46% of her forehands, compared with her usual 37% and the tour average of 28%. She won barely one-third of the points when she did so, partly because of the nine unforced errors she racked up playing an already conservative shot. Two of those missed down-the-middle forehands came on back-to-back points when she could hardly afford them, taking her from 15-all to 15-40 when trying to close out the match at 5-3 in the second set.

In the end, as we’ve seen, Gauff’s defense saved her. She won more than half of Kostyuk’s serve points despite lackluster returning. Had she served just a little better–she missed six straight first serves in that 5-3 game–she would have finished the job an hour sooner. Had she attacked a bit more effectively with her second shots, even the off-day from the line wouldn’t have amounted to much.

To state the obvious: She’ll have to play better to beat Sabalenka in tomorrow’s semi-finals. One thing, at least, will work in Coco’s favor: She’ll have many fewer choices to make. The defending champion will dictate play and give her less time to think than Kostyuk did. Gauff withstood the Belarusian barrage in New York, winning the US Open title despite a couple of detours against less aggressive players in the early rounds. The American can’t play tomorrow like she did yesterday, but thankfully, Sabalenka won’t let her.

* * *

Jannik Sinner’s near-unbreakability

Jannik Sinner has lost his serve just twice en route to the Australian Open semi-finals. He has faced 28 break points and saved 26 of them.

Since 1991, when the ATP started keeping the relevant stats, he is the 26th player to reach the final four at a major with so few breaks of his own serve:

Tournament  Semi-finalist       BP Faced  Broken  
2013 USO    Rafael Nadal               6       0  
2018 Wimb   John Isner                 7       0  
2015 Wimb   Roger Federer              3       1  
1994 Wimb   Pete Sampras               9       1  
2015 AO     Novak Djokovic            11       1  
2014 Wimb   Roger Federer             12       1  
1997 Wimb   Pete Sampras              12       1  
2010 USO    Rafael Nadal              14       1  
2012 RG     Rafael Nadal              17       1  
2004 Wimb   Roger Federer             17       1  

Tournament  Semi-finalist       BP Faced  Broken  
2014 Wimb   Milos Raonic               9       2  
2011 RG     Novak Djokovic*            9       2  
2007 USO    Roger Federer              9       2  
2006 Wimb   Roger Federer              9       2  
2006 Wimb   Rafael Nadal               9       2  
2015 USO    Roger Federer             11       2  
2014 AO     Roger Federer             11       2  
1997 USO    Greg Rusedski             11       2  
1993 AO     Pete Sampras**            12       2  
2013 Wimb   JM del Potro              13       2  
2019 AO     Rafael Nadal              15       2  
2008 Wimb   Roger Federer             15       2  
2005 AO     Andy Roddick              15       2  
1998 Wimb   Pete Sampras              17       2  
2000 AO     Yevgeny Kafelnikov        22       2  
2024 AO     Jannik Sinner             28       2

* Djokovic won one round by W/O and another by retirement
** I don't have stats for Sampras's QF, but the final score suggests that he wasn't broken

Pretty good company! As the table makes clear, though, Sinner’s 28 break points faced is not so elite. In fact, the average major semi-finalist faces exactly 28 break points in his first five matches.

The Italian’s accomplishment, then, is saving so many. 26 of 28 is a 93% clip, and that is more rarefied air:

Tounament  Player      Faced  Saved   Save%  
2013 USO   Nadal           6      6  100.0%  
2018 Wimb  Isner           7      7  100.0%  
2012 RG    Nadal          17     16   94.1%  
2004 Wimb  Federer        17     16   94.1%  
2010 USO   Nadal          14     13   92.9%  
2024 AO    Sinner         28     26   92.9%  
2014 Wimb  Federer        12     11   91.7%  
1997 Wimb  Sampras        12     11   91.7%  
2015 AO    Djokovic       11     10   90.9%  
2000 AO    Kafelnikov     22     20   90.9%

Things will get tougher on Friday, when Sinner faces all-time-great returner Novak Djokovic for a place in the final. Then again, Djokovic failed to convert his first 15 break points against Taylor Fritz yesterday–maybe he was just preparing for the matchup with Sinner.

* * *

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

 

Can Lorenzo Sonego Hunt Down Enough Forehands?

Lorenzo Sonego at Monte Carlo in 2022. Credit: si.robi

This year, Australian Open broadcasts threw a screwball into their traditional post-match statsheet. In the addition to the usual numbers–winners, unforced errors, break points won, and so forth–the graphic shows something called “Hunting 3rd Shot Forehand.” I must have missed a memo. This is the first I’ve heard of such a thing.

A puff piece for the Tennis Australia data group offers something of a definition. The new stat measures “the times the server forehands their first post-serve hit, indicating their desire to dictate the point.” Um, ok. In other words, when the service return comes back, how often does the server hit a forehand with his next shot?

The intention behind the metric is straightforward. You hear a lot these days about the “plus-one”–the server’s second shot. While the serve is the most important stroke in tennis, the plus-one shot is the next-most crucial opportunity to attack. Both because it arises often, and because it offers a chance to define the direction of the rally, even if it’s not yet possible to put the ball away.

It is easier to dictate play with a forehand than a backhand; the potential trajectories of the stroke give a player more options. Beyond that, most men have better forehands than backhands. (The stat appears on broadcasts for both men’s and women’s matches, but today I’m going to talk about the men’s game.) If the goal is to command the rally with the plus-one shot, it’s better to hit a forehand than a backhand. A higher “Hunting 3rd Shot Forehand” number, then, is better.

The post-match graphic, with new stat second from the bottom

Before we go further: I simply can’t use this name. It’s long and confusing. (Is the player hunting for the forehand? For a winner? For a silly rabbit?) I’m going to call it “3rd Shot FH%” or “3F%” for short.

(And yes, I promise to get to Sonego eventually.)

The stat is not as straightforward as the intention behind it. The implication of 3F%, I think, is something like, “How hard did the player try to hit plus-one forehands?” A possible further implication is, “How well did the returner prevent his opponent from hitting plus-one forehands?” The second question prompts yet another: “How well did the server keep the returner from sending balls to his backhand?”

It may be possible to separate some of those questions, but there’s a lot more spadework to do before we get there.

What is normal?

(You might doubt whether I am well-situated to answer. Still, we soldier on.)

Your TV screen shows you some “Hunting 3rd Shot Forehand” numbers. Are they good?

The 3F% metric can be calculated from Match Charting Project data, so we have thousands of data points to draw upon. Based on men’s matches since 2014, the average 3F% is 64.7%. The middle third of player-matches falls between 59.3% and 70.9%. Take a little liberty with rounding, and we can say that “normal” is the range from 60% to 70%. Less than 60%, and you’re doing something wrong–or you’d rather hit your backhand, or your opponent had a day. More than 70%, and you were really getting things done in the plus-one department.

Some players consistently land at the far ends of the distribution. Here are career numbers for the top ten active players by this metric, along with 15 more names of interest:

Player                3F%  
Lorenzo Sonego      78.6%  
Rafael Nadal        77.7%  
Joao Sousa          77.6%  
Denis Shapovalov    77.1%  
Albert Ramos        76.0%  
Jeremy Chardy       75.7%  
Milos Raonic        74.6%  
Stefanos Tsitsipas  74.4%  
Casper Ruud         73.9%  
Grigor Dimitrov     73.9%  
* * * * * * * * * * * * *                    
Holger Rune         69.7%  
Dominic Thiem       69.3%  
Hubert Hurkacz      67.5%  
Carlos Alcaraz      67.2%  
Andrey Rublev       64.7%  
Jannik Sinner       62.6%  
Alex de Minaur      61.2%  
Stan Wawrinka       60.9%  
Andy Murray         60.0%  
Taylor Fritz        58.3%  
Diego Schwartzman   56.9%  
Novak Djokovic      56.2%  
Frances Tiafoe      55.8%  
Alexander Zverev    51.1%  
Daniil Medvedev     50.0%

There’s Lorenzo!

The top of the list gives you an idea of what sorts of game styles result in lots of plus-one forehands. Big serves help. Left-handedness works in your favor, perhaps since everyone trains so hard to return to a right-hander’s backhand side. Some clay-courters do well, as they are less likely to think of the serve as a point-ending shot on its own, focusing instead on how it can set up the point.

It also helps to try to hit plus-one forehands. Neither Zverev nor Medvedev seem to think in those terms, so their low 3F% ratings don’t reflect any lack of execution.

Does this even matter?

Some valuable on-screen real estate–and an enormous amount of coaching time–would be wasted if 3F% didn’t correlate with points won. Fortunately for the conventional wisdom, it does: A plus-one forehand is more likely to lead to a point for the server than a plus-one backhand is.

57.5% of plus-one forehands eventually turn into a point won, compared to 50.9% of plus-one backhands. That’s a ratio of 1.13, a number that will be more useful as a reference point in a moment.

The value of a plus-one forehand depends on the player. Matteo Berrettini wins 58.5% of plus-one forehand points but only 44.6% of plus-one backhands. That’s a ratio of 1.31, one of the highest of any active player. For him, 3F% certainly matters: All else equal, more plus-one forehand points leads to better results overall.

(A word of caution, though: The marginal plus-one forehand point–that is, the next return that he might have ran around to hit a forehand, but didn’t–might not have improved his results for the better. Presumably Matteo knows his own capabilities, and he hits forehands only on those points where they improve his odds of winning. The marginal plus-one forehand, for a player like him, is a fairly desperate foray into the doubles alley.)

For others, the plus-one choice barely registers. Zverev wins 56.7% of plus-one forehands and 52.8% of backhands, a ratio of 1.07. Every other top-tenner has a wider split, but there are more extreme examples. Adrian Mannarino wins more points behind his plus-one backhand than forehand, 55.4% to 52.3%.

Sonego, our 3F% champion, gains nearly as much from his plus-one forehands as Berrettini does. He wins 57.1% of plus-one forehand points, against 44.4% for backhands. It isn’t easy to find his backhand, but it’s worth the attempt.

What about first and second serves?

As far as I know, broadcasts don’t separate the “Hunting” metric into first and second serves. But they should! Early in the rally, the effect of the serve retains plenty of influence.

The following table shows some of the tour-wide averages I’ve discussed so far–3F%, plus-one forehand points won, and plus-one backhand points won–broken down into first-serve and second-serve points:

AVERAGES     3F%  FH W%  BH W%   
1st serve  71.1%  60.3%  53.9%  
2nd serve  55.2%  52.2%  48.1%
Total      64.7%  57.5%  50.9%

Bigger serves generate weaker (and less targeted) returns, which invite more forehands. Behind second serves, ATPers only manage to hit forehands on 55% of their plus-one shots. On the other hand, the gap in points won isn’t as wide.

A fascinating outlier is Andrey Rublev. He finds the forehand on over 80% of his first-serve points, one of the highest numbers on tour. Behind the second serve, though, he hits plus-one forehands only 43% of the time–one of the lowest! It’s no secret that his second serve is a liability, but such a gap still comes as a surprise.

Sonego is a more typical case: lots more plus-one forehands on the first serve than the second (83% to 71%), and a wide gap in the results between forehands and backhands regardless of which serve it supports.

Converting from the backhand corner

With a few exceptions like Mannarino, most players want to hit as many plus-one forehands as they plausibly can. If the return goes to their forehand corner, obviously they’ll hit a forehand. If the return comes back up the middle, it’s either a no-doubt forehand or an easy decision to take a couple of steps around the ball and avoid the backhand.

The real decisions happen when the return goes to the backhand corner. Now we’ve moved into true Match Charting Project territory. I don’t know if the Australian Open has the data to drill down this far; either way, it probably won’t show up on your TV screen. In this corner of the internet, though, we’ll dive in.

About one-quarter of in-play returns go to the server’s backhand corner. Ernests Gulbis set the standard for plus-one backhanding, running around just 2% of those balls. On average, players go for the forehand 26.6% of the time. Even Zverev and Medvedev go that route sometimes: 9% for the German and 8% for the Russian.

Here again, Sonego sets the standard. He runs around 49% of those returns, winning 53% of the resulting plus-one forehands versus 47% of the backhands he can’t avoid. No other active player creates so many forehand opportunities. Of retired players in the charting dataset, only Carlos Moya and Leonardo Mayer were more extreme.

Here’s the same tour-averages table as above, now limited to points with returns to the backhand corner:

BH CORNER    3F%  FH W%  BH W%  
1st serve  34.9%  61.3%  53.0%  
2nd serve  16.9%  52.3%  47.3% 
Total      26.6%  58.7%  50.0%

It’s possible that many players–though probably not Sonego–are leaving some points on the table here. I’m surprised to see that the gap in win percentages between plus-one forehands and backhand is bigger for backhand-corner returns than returns in general. Backhand-corner returns are somewhat similar to each other–certainly more similar than returns in general. Thus I would expect that players would find an equilibrium in which they ran around enough shots that their forehand and backhand winning percentages end up closer together. Perhaps some ATPers overestimate the quality of their backhands, or maybe they don’t want to look foolish taking a chance in the doubles alley. Or they might just know what they’re doing, and the guy typing on his laptop should shut up about it.

Hunting Alcaraz

Sonego beat Dan Evans in Melbourne yesterday, earning him a date on Thursday with second-seeded Carlos Alcaraz. While there’s more to the match than Sonego’s hunt to maximize his 3F%, the battle for the Italian’s plus-one court position will play a big part.

Alcaraz is a bit better than the typical tour player at landing his returns in the server’s backhand corner, something he does 30.8% of the time, compared to the norm of 27.0%. But it doesn’t make him particularly effective at avoiding his opponents’ plus-one forehand. They find the preferred shot 64.5% of the time, almost exactly tour average. The story is the same when we look at first and second serves separately: Carlitos neither prevents nor encourages plus-one forehands.

There are, naturally, returners who consistently limit plus-one options; others don’t have the skills to avert a barrage of forehands. Jenson Brooksby allows opponents plus-one forehands on just 57.7% of his returns; Andy Murray and (surprisingly?) Rublev keep opponents’ numbers down around 59%. At the other extreme, Cam Norrie allows servers to hit plus-one forehands almost three-quarters of the time. He’s one of many lefties who struggle by this metric: Since serve returns are disproportionately backhands themselves, left-handers must often go down the line to put a return in a right-hander’s backhand corner. Norrie finds that corner only one-fifth of the time.

Well-targeted returns are good; forcing servers to hit plus-one backhands pays dividends. Alcaraz, though, is proof that you can make your money on the fourth shot and beyond. Opponents hit plenty of plus-one forehands against him, yet no matter what they choose for the second shot, they struggle to win the point. First serves, second serves, plus-one forehands, plus-one backhands … Carlitos beats tour average by multiple percentage points in every category. This table shows the difference between how players fare against Alcaraz and the average level, in percentage points:

VS CARLOS  rel FH W%  rel BH W%    
1st serve      -6.9%      -2.5%  
2nd serve      -2.2%      -5.0% 
Total          -5.7%      -3.5%

In other words, a plus-one forehand is 5.7 percentage points less likely to turn out well against Alcaraz than it is against an average ATP player. That’s a hefty margin for something that accounts for nearly half of the typical player’s service points.

It’s fun to know that Sonego occupies the unique position that he does on tour, and it’s entertaining to see some of the far-fetched places from which he’ll smack an inside-out forehand. It might even be useful to see the Italian’s “Hunting” stat at the conclusion of tomorrow’s match.

Alas, “setting up the point” and “winning the point” are two different things. Sonego might hunt down enough forehands against Alcaraz to manage the first, but the second is a considerably bigger ask.

* * *

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

 

The Manufactured Attack of Caroline Garcia

Caroline Garcia in 2019. Credit: Peter Menzel

Last night, Caroline Garcia scored what many fans saw as an upset, straight-setting two-time Australian Open champion Naomi Osaka. While Garcia was seeded 16th and Osaka is just beginning a comeback, no one ever knows quite what to expect when the Frenchwoman takes the court. The former champ, for her part, has always been at her best on big stages.

The result was almost pedestrian. Garcia turned in a performance that exemplified the tennis of her late 20s: Serving big, returning pugnaciously, taking risks, and–on the rare occasions that Osaka left her an opening–net rushing. Osaka served well, but the 16th seed out-aced her, 13 to 11. More than three-quarters of points were decided in three shots or less, and Garcia stole a few more of those from her opponent than Osaka did from her. In a contest defined by small margins–one break of serve and a tiebreak–that was all it took.

The strange thing is, Caro didn’t use to play like this. She plays shorter points than any other tour regular, an average of 2.9 shots per point in charted matches from the last 52 weeks. It isn’t just about her powerful first serve: Her return points end even sooner than her serve points do. Back in 2018, when she first reached her career-best ranking of 4th on the WTA computer, she was averaging over four shots per point, a rally length that would put her in the range of Jessica Pegula and Maria Sakkari: in other words, a very different sort of player.

Here is the evolution of Garcia’s rally length, shown as a rolling 10-match average, for the 84 matches in the charting dataset:

Last night’s rally length was a blink-and-you’ll-miss-it 2.5 shots, the second-lowest figure I have on record for Garcia. Only a match against Donna Vekic last year comes in slightly lower, though last week’s match in Adelaide against Jelena Ostapenko may have been even more extreme. Osaka’s big game helped keep the number down, but it takes two to so comprehensively avoid the long-rally tango.

Garcia’s first serve has always been a weapon. But her tactical approach behind it has fluctuated wildly. The career trend of her Aggression Score in rallies illustrates how she has careened from one extreme to another. Aggression Score is scaled so that the most passive players rate around -100 and the most aggressive around 100, though Ostapenko and others have pushed the maximum figures further into triple digits. Here is how Garcia’s score has changed over time, again as rolling ten-match averages:

I don’t think there any other player in tennis–man or woman, past or present–who has followed a path like this. As she established herself as an elite on tour, even as she rose into the top five, she became more and more conservative. For reference, players who posted scores around zero in 2023 were Sakkari and Martina Trevisan, hardly styles that will remind you of Garcia’s. Eventually she reversed course, not only regaining her former style but surpassing it, ranking among Liudmila Samsonova and Aryna Sabalenka as one of the most aggressive players on tour, a rung below the class-of-her-own Ostapenko.

Is it working?

The oddest thing about the multiple phases of Garcia’s career is that she has reached the No. 4 ranking with two different styles. In each of her first three charted matches after achieving the peak ranking in 2018, she posted negative rally aggression scores. In two matches against Sabalenka, she averaged 3.9 and 3.7 shots per point; against Karolina Pliskova in the Tianjin final, the typical point lasted 4.3 strokes. When she returned to the No. 4 ranking at the end of 2022, after years in the wilderness, she was frequently posting triple-digit aggression scores and average rally lengths below 3.

The main effect of Garcia’s current style is that it makes the most of her serve. From 2015 to 2017, she won just over 66% of her first-serve points, a mark that is good but sub-elite. She fell all the way to 62% in 2021 before the big shift; since then, she has won more than 70% of her first-serve points. She ranked fourth in that stat heading into the Australian Open, and she converted nearly 90% of her first serves against Osaka. Her success behind the second serve hasn’t shown the same improvement, but the overall picture is a good one: She won more total serve points in 2023 than ever before.

The return game is a different story. This is where even a casual viewer can’t miss Caro’s new tactics: She’s not afraid to stand well inside the baseline to return serve, and yesterday she net-rushed one Osaka serve, SABR-style. Measured by court position, if not by winners and error stats, Garcia is even more aggressive than Ostapenko.

At her best, the Frenchwoman posted acceptable return numbers, if not great ones. Her best single-season mark, winning 42.7% of her return points in 2017, put her in the bottom third of top-50 players. As she has upped the intensity of her attack, this key number has headed south:

In the last 52 weeks, she has won just 38.3% of return points, worst among the top 50 by two full percentage points. Among the top 20, no one else is below 42%. She can get away with it because her own serve is so rarely broken, but such ineffectual return results will make it difficult to mount another assault on the top five. Breaking serve so rarely dooms her to a career of three-setters and narrow decisions. Those sorts of results can sometimes be encouraging–as in her pair of recent three-set losses to Iga Swiatek–but have a knack for halting winning streaks, too.

It doesn’t have to be this way. Players don’t sign contracts agreeing to deploy the same tactics on both sides of the ball. Garcia won return games far more often in her less aggressive days, breaking 33% of the time in 2017 compared to a dreadful 23% last year.

Some of Caro’s 2017 skills are still in evidence. She is solid enough in long rallies that she doesn’t need to so actively avoid them: In the last year, she has won a respectable 48% of points that lasted seven or more strokes, and if you remove the two Swiatek matches, she breaks even. While the Osaka match was primarily determined by short points, Garcia won 17 of 29 (59%) that went to a fourth shot.

Without any major changes, Garcia will remain the sort of player who aggravates fans and opponents alike, a dangerous lurker capable of delivering upsets, inexplicable marathons, and lame early exits in equal measure. Like any hyper-aggressive player, Caro’s results can be seemingly random, with all the frustration that entails. Unlike Ostapenko, Sabalenka, and the many ball-bashers on tour, though, Garcia has chosen to play this way, rebuilding her game into something that the 2018 version of herself would hardly recognize. If she can somehow join her late-career serve to her earlier return-game tactics, the randomness will disappear, and Caro may make yet another appearance in the top five.

* *

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