Yesterday, Richard Gasquet won a fourth-round match at a Grand Slam.
If that doesn’t surprise you, you haven’t been paying much attention to Gasquet for, say, the last eight years. The Frenchman with the stunning backhand has advanced to the fourth round at a Slam 17 times now, making him only the 35th man in the Open era to do so. The problem is what happens next.
Entering yesterday’s match, Gasquet was 1-15 in round-of-16 matches at majors, his one victory coming at 2007 Wimbledon over Jo-Wilfried Tsonga. Since then, he’s lost his last eleven tries, including one to Tsonga and two to David Ferrer, his quarterfinal opponent this week. No player has lost more than 15 fourth-round Slam matches; only Wayne Ferreira reached the same plateau, and Lleyton Hewitt will match it if he loses today.
One thing that has held him back is an inability to beat higher-ranked players, as Carl Bialik noted earlier this year. At slams, he has played 28 matches against players with superior ATP rankings, and won only four of them. Against lower-ranked players, he is 62-11. Since Gasquet’s ranking has rarely reached the top eight, that mark hasn’t helped him the fourth round, where players outside of the top eight generally meet a higher-ranked opponent.
Now that Gasquet has broken through with his second Grand Slam quarterfinal appearance, history suggests he’ll go no further. He has beaten Ferrer only one time in nine tries, and that was five years ago. And Ferrer’s ranking puts him firmly in the category of guys Gasquet doesn’t beat at majors.
There’s one reason for hope, though. Despite all the disappointment in the fourth round, he has never lost a Grand Slam quarterfinal.
Daniela Hantuchova‘s appearance in the quarterfinals of this year’s US Open is surprising for a different reason. When the tournament began, her spot was pegged for Petra Kvitova, before an ailing Kvitova was upset by Alison Riske. For all my talk recently about easy bracket on the men’s side, no one in either single’s draw has faced such lowly-ranked competition.
Hantuchova’s four opponents thus far include two qualifiers and two wild cards. Among them, only Riske is ranked inside the top 100, and she’s #81. By contrast, Hantuchova’s presumptive quarterfinal opponent, Victoria Azarenka, will have faced #13 and #28.
Of over 850 women’s Slam quarterfinalists since 1987, only six have reached the quarters without playing someone in the top 80. The luckiest path was that of Claudia Kohde Kilsch, who reached the 1989 Wimbledon quarterfinals by beating #126, #246, #247, and #131. Then her luck ran out: Steffi Graf ended her run in the quarters. Steffi herself is one of the six, having won her first four rounds at 1993 Wimbledon without playing anyone ranked better than #87.
These lucky draws have become less common in recent years. Of the six, only one has occurred since Steffi’s run in 1993. Nadia Petrova reached the quarterfinals at the 2006 Australian Open without having to beat anyone ranked better than #100.
After four easy matches, there’s little pattern to how these players fare in the quarters. As we might expect, the success rate in their fifth matches has much more to do with their quarterfinal opponents than the women they faced to get there.
And perhaps you’ve heard: Tommy Robredo defeated Roger Federer in straight sets.
It was the first time in twelve meetings that Robredo beat Fed. It’s the Spaniard’s first quarterfinal appearance in New York, despite seven previous fourth-round showings (including one against Roger, in 2009). Even Gasquet hasn’t been that bad, losing in the US Open round of 16 a mere four times. And Robredo pulled off the upset while winning fewer return points than his opponent did–something that happens in only one of 15 US Open men’s matches.
When oddities like this occur–Gasquet’s match is another, as he won only 48.5% of total points–it is almost always because the winner played much better on high-leverage points. In many matches, those important moments are at the back end of tiebreaks, when two points can make or break a set. In Federer’s loss, the finger-pointing is directed at break points. Roger barely converted any of them. It’s been a problem for Fed for years, particularly in his last several Slam losses.
It’s difficult to know how to evaluate poor break point performances. In one sense, it’s obvious: If Fed was going to win the match, he needed to win more. A failure to convert break points is a good explanation for any loss.
But what does it say about Fed’s current level, or about what we can expect from him going forward? Is he suddenly weak on break points? When I ran the numbers a couple of years ago, he was winning slightly fewer return points in the ad court, but the difference isn’t nearly extreme enough to explain a 2-for-16 performance on break points.
What’s particularly frustrating about squandering so many break points is that he earned them with good play on other return points. And, of course, there’s no difference between a typical ad-court point and a break point except for the pressure.
So, if Federer is still generating all those break-point opportunities, is he simply suffering through a run of bad luck? Has he lost his clutch superpowers? Have other players ceased to fear him in big moments? Judging from the growing number of surprising defeats in Roger’s record, it certainly seems to be something more than bad luck.
Finally, a couple of notes.
Don’t miss this win probability graph of the Raonic-Gasquet match. Mike says it’s “almost too interesting.”
In the New York Times Straight Sets blog (known for its coverage of the United States Open), Clayton Chin gives a brief overview of a forecasting method. He emphasizes his reliance on the Monte Carlo method–a technique that utilizes thousands or even millions of simulations–which isn’t necessary here.
If you estimate each player’s serve and return points won, it’s straightforward to calculate each player’s chances of winning a game, set, or match. Generally speaking, Monte Carlo techniques are useful when such closed-form solutions aren’t available.
The most important part of Chin’s approach is one he doesn’t shed any light on. If Serena is holding serve at a certain rate and breaking serve at a certain rate over the course of the year, how do you generate hold and break rates for an individual match? It can be done, and many have tried, but that’s much more challenging that simulating outcomes at the match or tournament level. Without that glimpse under the hood, it’s tough to know how much weight to give his results.