Contrasting Serves, Futile Slams, and (More) IBM Shortcomings

In most of his matches, John Isner makes his opponents look short and their serves look weak.  What happens, then, when his opponent really is short, with one of the weakest serves in the game?

Third up on grandstand today, Isner takes on Filippo Volandri, the man who sets records Isner will never reach.  Three years ago, the Italian failed to hit a single ace for 19 straight matches.  Volandri may not be as short as some players on tour–the ATP site lists him at six feet–but it’s more common for him to fail to hit an ace in a match than it is for him to hit one.

In the last year, Isner has hit nearly 19% of his first serves for aces, good for best among tour regulars.  In the top 50, the other extreme is represented by Nikolay Davydenko, whose rate is just under 3%.  Volandri–despite playing many weaker opponents on the Challenger tour–sits at 0.8%.

The good news for Big John is that the 31-year-old Volandri is a nonentity on hard courts, having not played on the surface since losing in the first round of the Australian. The bad news? He’ll have to hit a lot of returns today.

As my forecast very delicately predicted, Fernando Verdasco didn’t live up to his seed, losing to the barely-unseeded Ivan Dodig yesterday in five sets.  That’s the fourth slam this year in which he’s lost in a five-setter.

Verdasco, with his flashy talent and underwhelming results, comes in for his share of fan mockery.  But this is one time he doesn’t deserve it.  Out of the several dozen players who enter all four slams each year, almost all will lose four matches.  While it may be frustrating to lose in five, losing in five, all else equal, says better things about your game than losing in three.

One of those five-set losses this year was to Andy Murray at Wimbledon; the other two previous contests were against Janko Tipsarevic and Kevin Anderson.  Perhaps Fernando should have finished off at least one of those matches, but none of his four slam losses this year are nearly as groan-inducing as, say, Ernests Gulbis‘s disaster yesterday against Andreas Haider-Maurer.  And his record is nothing compared to Marinko Matosevic‘s streak of 11 losses in 11 slam appearances.

Verdasco is the sixth man in the Open era to complete this distinctive slam feat, and he’s not in bad company. Last year, Isner did it–and added an exclamation point with a five-set loss in Davis Cup.  Before that, the most recent were Fernando Gonzalez in 2006 and Tim Henman in 2000.  Not bad company.

Anyway, if you’re drawn to this unusual feat, don’t miss Steve Johnson‘s first-round match with Tobias Kamke. It’s last on Court 13 today. Johnson is three-quarters of the way to the Fernando slam, losing all three of his matches at majors this year in five sets.  If he completes the set, it will be particularly impressive for at least one man: Kamke has won only two five-setters in his career.

As part of IBM’s ham-handed PR push leading up to another slam, the company gave analyst and coach Craig O’Shannessy some data.  He reported some results on both the ATP site and the New York Times Straight Sets blog.

This is a huge step up from the thinly-veiled advertisement I highlighted yesterday.  But it still, frustratingly, falls short.

One of the major points of Craig’s ATP piece is summarized at the beginning: “Most baseline points are a losing proposition,” and “Approaching the net is a goldmine.”  Later, he continues, “It seems amazing that players don’t venture forward more often to capitalize on the far higher winning percentage approaching offers over baseline play.”

Is this the data-driven, actionable advice I pleaded for last week? Not quite.

As I’m sure Craig would agree, opportunities to come to net aren’t always available, and they don’t arise in a vacuum.  Especially in today’s baseline-focused game, net points tend to occur when one player hits a particularly weak shot.  So if most net points end in victory for the player who approaches, is that because of the choice to come to net, or the weak shot that generated that opportunity?

Think about it probabilistically.  When Djokovic serves against Tsonga, let’s say he has a 75% chance of winning a first serve point.  If Tsonga hits a weak chip return in the middle of the court, allowing Novak to take several steps forward, we could figure that Djokovic’s chance of winning the point increases to 95%–perhaps higher.  When Novak puts away his second shot, he wins the point.  Formally speaking, his chance of winning jumps to 100%.

Now, in that example, what do you credit as the reason for Djokovic winning the point?  Landing a solid first serve, which gives him a 75% chance of winning instead of, say, 60%? A particularly good first serve, which forced the weak return?  Tsonga’s poor return? Or Novak’s “choice” to approach the net?

That final choice is laughable.  And this is the data he’s drawing from.  Aside from a few particularly aggressive players on tour, that’s the profile of a net point in 2013.

So, what’s the actionable advice here?  You probably shouldn’t approach the net without a reasonable opening, so … hit bigger serves to get more weak returns? Hit deep groundstrokes into corners? Take advantage of short balls?

These are the benefits we reap from “Big Data?”

IBM clearly wants to wow us with this stuff.  Yet the “findings” are so elementary as to be useless.  The solution is so simple: release the data, let fans and analysts innovate, and watch the quality of this work go through the roof.

11 thoughts on “Contrasting Serves, Futile Slams, and (More) IBM Shortcomings”

  1. Regarding O’Shannessy’s ATP post, I would have been more interested if he had somehow explored what many fans (like me) often wonder about in regards to approaching the net – namely, why some who clearly would benefit from a greater willingness to finish off a point at net when appropriate choose not to do so. A classic example to my mine was Robin Soderling: it seemed that it was only under the direction of Magnus Norman that Soderling took the risk and got the reward. Whereas there seem to be players who shun net play on principle, are never lucky enough to get as good a coach as Norman, and thus never learn what you speak of in terms of constructing a point that rewards a trip to the net. It might seem obvious but to a great many players it apparently isn’t something that enters their minds. Another example to me would be the mindless baseline bashing of Sharapova.

    I also caught O’Shannessy’s post on the NYT blog about more service breaks in the WTA than ATP, and in that case I was again disappointed – and again was looking for a more anecdotal inquiry, one that might explain what seems to me a curious discrepancy: I’ve long thought that the taller you are, the easier it is to get the serve in & thus the more powerful serve you can have, provided your timing and technique is good. We see that on the men’s side with Isner & other tall men having nasty serves, and David Ferrer on the other hand having to work like the dickens to protect his serve. But on the women’s side it doesn’t quite seem to add up, at least without the aid of stats to somehow penetrate the murk: I can think of at least two tall women who have crappy, unreliable serves – Sharapova and Azarenka – and two shorter women who have highly regarded serves – Serena and Sam Stosur. Of course both Serena and Sam are pretty bulked up, which helps in whacking a big serve, but I still wonder why we aren’t seeing any Isner equivalents on the WTA side.

    1. WTA’s Isner… Lucie Hradecká? At least some tiny part of the time? And she did have that match against Serena with all the tiebreaks…

      Amanmuradova has a pretty big serve and she’s 6’3″ or something. ‘course she’s terrible overall, but I think the type exists, it’s just that none of them have gotten anywhere yet. Most of the tall women with big muscles are playing volleyball or basketball, I guess.

      1. I can’t let this pass w/o just saying that I love Lucie. She used to come practice at the courts where I played in Queens, and man oh man can she crush the ball. Scary.

      2. Hradecká . . . didn’t know about her. Good call. I have to confess I don’t follow women’s tennis with anywhere near the attention I did back in the days of Graf et al. From a YouTube video I just peeked at (vs. Peng, Auckland 2012), Hradecká seems to have a very quirky style – two-handed on both sides with the occasional one-handed forehand? Interesting.

    2. Agreed. This is where the hawkeye data would be useful. We could identify short balls as opportunities to come in, see whether a player did or didn’t, and also note times the player came in when perhaps he shouldn’t have. Someone like Tsonga might approach the net the “correct” number of times, but only because his bad choices balance themselves out.

      I had planned on saying something about the gender difference piece as well, but went on too long about the first one. Regarding your point on the serves, it may be that in order to dominate with that shot (among either gender) you need to be hitting 130mph, and that just isn’t going to happen even for most of the tallest, strongest women. Or you need the trajectory that comes with being 6’4″+.

      I also didn’t like Craig’s argument that the WTA had it’s own perks because of the higher number of winners. I don’t dispute that women’s tennis has plenty of appeal, but that’s a horrible reason. A winner is made up of two parts: a good shot, and an opponent that can’t get to it. Who knows which one is a stronger influence on the number of WTA winners?

  2. Too right. Often I have been appalled by the propensity of British commentators to urge players forward. “Come to the net!” they cry, as if that in itself was a guarantee of success. Then we watch as the best players in the world – Nadal, Djokovic, Federer, Ferrer, Berdych, Del Potro, Tsonga, and others – stand rooted and watch passing shots wing by them. Or lobs rise gracefully over their heads.

    1. But likewise, the failure of lesser players to come to net when that is the best way to control and finish the point is equally glaring. It’s more like Jeff said – in today’s game, it’s still possible to come to net, but you’ve got to have developed an initiative of some sort to justify it. You can’t just come in at random, on the one hand, but on the other, if you never come in, that’s a sign your game is too one-dimensional and you’re likely losing some opportunities.

      1. Time is the limiting factor, off court that is.

        Watching a guy like Raonic at net is revealing. He obviously *can* volley, but his misses clearly indicate he doesn’t *own* a net game. I figure he’s making a choice every practice day to reinforce his necessary strengths and shore up his targetable weaknesses, rather than make the sizable investment required to truly own a net game.

        You have to invest a sizable chunk of practice time to master an optional skill you’ll never use if your mandatory skills suffer as a result (unless you *want* to be the next Stakhovsky.)

        1. Interesting point. I’m not sure Raonic is a good example, though, because regardless of whatever skills he’s choosing to practice, he seems to be regressing more than improving. As has been pointed out, his choices while serving under pressure can be bizarre at times. Once he was a “promising” player; now he seems more like “puzzling.”

    2. And this is why, at bare minimum, the sort of analysis Craig is doing should be player-specific, and identify a *range* of good net approach rates, and not a *threshold.*

      Maybe we’ll learn that Ferrer is much more successful when he approaches the net on 5-10% of points. More than 10%, he’s being reckless, moving forward at poorly chosen times, and getting passed. Less than 5%, he’s missing opportunities to put away points.

      Obviously that range will be different from player to player. And the amount of potential gain surely isn’t linear with a simple increase in net approaches. Ferrer picks up a lot of easy points going from 0% net approaches to 5%, but 10% to 15% may or may not benefit him at all.

Comments are closed.