Category: Research

The Weak, Weak Newport Field

The ATP 250-level tournament in Newport this week is empty of the game’s best players.  The top seed is John Isner, ranked 46, and the 8th seed is Tobias Kamke, who is barely within the top 100.  This is no surprise.  Newport has one of the weakest ATP fields every year, situated as it is the week after Wimbledon, simultaneous with Davis Cup.

In a little study I did last year, I discovered that at least in 2009, Newport did have the weakest field of any ATP 250 event.  If you click the link, you’ll find a variety of metrics, but I think we can focus on just one: the median rank of main draw players.  By using median instead of average, the numbers aren’t skewed by a lowly-ranked wild card or qualifier.

In 2009, the players in the Newport draw had a median ranking of 125–that is, half the players in the main draw of an ATP event were ranked above 125.  Grand slams usually manage about 110 players below the 125 mark, but Newport only got 16–and most of those were closer to 125 than to 1.  Last year, the median fell to 129.5.  It may be a small consolation that Johannesburg’s field was equally weak.

A glance at this year’s draw can tell you that not much has changed.  Thanks to many late withdrawals, the cut fell to 218, which is considerably higher than the cut at some challengers.  For all that, the field quality has improved somewhat, to a median rank of 111.  That leaves Jo’burg in the dust; the South African event had a median rank of 118.5.

The non-challenger challengers

A few tour-level events–Newport, Jo’burg, and perhaps San Jose–obscure the line between the tour and challenger levels.  In the eyes of the ranking system, they are very different–Newport is worth 250 points to the winner, while no challenger is worth more than 125.  But for all intents and purposes, Newport and Jo’burg are challengers.

Last year, the May event in Bordeaux attracted a field with a median rank of 128–just above last year’s Newport and Jo’burg numbers.  This March, the odd 24-man field at Le Gosier had a median rank of 123.  Already in 2011, six challengers with 32-man fields had median ranks below 150, putting them in the same ballpark as the lowest rungs of the tour.

All of this is another strike against the ranking system, which treats Newport as if it were equivalent to, say, Sydney, where the last direct acceptance this year (#53 Benjamin Becker) was higher-ranked than Newport’s second seed (#60 Grigor Dimitrov).  Bad news for properly ordering second-tier pros, but good news for Isner, who can take advantage of this week’s cupcake draw to bounce back to as high as #36.

Bernard Tomic and the ATP Top 100: In Perspective

With his quarterfinal showing at Wimbledon, Bernard Tomic will break into the ATP top 100 for the first time on Monday.  He’ll do so with style, jumping from #158 to approximately #70.  (He will be considerably higher in my rankings–before the tournament, I had him just inside the top 50.)

As I’ve written before, a player’s chances of reaching the top of the men’s game have a lot to do with how early he cracks the top 100.  If you’re going to be a top-tenner, odds are you’re flashing some measure of those skills as a teenager.  In fact, to quote myself:

In the last 30 years, only one #1-ranked player (Pat Rafter) hadn’t reached the top 100 as a teenager, and he made it into the top 100 when he was 20.  Almost every eventual top-10 player had broken into the top 100 by age 21.

In that sense, Tomic is well ahead of the curve.  He doesn’t turn 19 until October, making him five months younger than Ryan Harrison, another teenager soon to break into the top 100.  Reaching #70 at such a young age isn’t a guarantee of future success, but it strongly points in that direction.  Again from my earlier post: 11% of players who cracked the top 100 at age 18 went on to become #1, and more than half (61%) eventually reached the top ten.

Tomic’s “comps”

Let’s take a narrower look and examine the 20 players who broke into the top 100 at ages closest to Tomic’s current age of 18.7 years.  It’s an impressive list, including Andy Roddick and Ivan Lendl, along with another 11 top-tenners.  Of these players the only “busts” were Andreas Vinciguerra (peak ranking: 33), Richard Fromberg (peak: 24), and Evgeny Korolev, who may yet improve on his peak ranking of 46.

In this group of 20 players, the average peak ranking is 11, and the median peak ranking is 8.  The average number of weeks in the top 100 is 362 (roughly eight years) and the median number of weeks is 410 (more than nine years).  Even 410 slightly understates a reasonable projection, since a few of these players (Roddick, Gael Monfils, Tommy Robredo, and Mikhail Youzhny) are guaranteed to add to their totals.

What may be most impressive about Tomic’s ranking at such a young age is that he has accomplished it the hard way.  He’s gotten plenty of wild cards–including at the Australian Open, where he reached the third round–but he qualified at Wimbledon, and a substantial chunk of his ranking points come from the challenger level, where he has reached four semifinals in 2011 alone.  His only “cheap” points are from Indian Wells, where he was wildcarded in, then beat Rohan Bopanna in the first round.

Now, Tomic’s ranking ensures that wild cards won’t be an issue, except at a few Masters 1000 tournaments.  If history is any guide, he’ll be a regular feature in the top echelon of the tour for most of this decade.

Fun With French Open Rally Length

At the ATP level, the ability to hang around in long rallies seems to be a key to success, especially on clay.  Most of the top players are such good defenders that a one-dimensional serve/forehand game just doesn’t cut it.

One stat you’ll occasionally see on television broadcasts is the number of points that reach a certain length, along with how each player is performing on those points.  The cutoff I’ve seen most frequently is 8 shots, and that seems like a reasonable enough line to draw.

Armed with point-by-point data from (most of) the men’s singles matches at the French Open, we can take a closer look.  The following table shows three numbers for each of the 16 players who reached the 4th round:

  • Average rally length–that’s the total number of shots per point for every point that the player contested.
  • Percentage of points that reached eight or more shots.
  • Percentage of eight-or-more-shot rallies that the player won.
PLAYER              Shots/Pt     8+  8+Wins  
Juan Ignacio Chela       5.3  25.7%   48.0%  
Gilles Simon             5.3  25.7%   59.8%  
Andy Murray              5.1  22.8%   50.5%  
Viktor Troicki           4.7  19.2%   48.9%  
Rafael Nadal             4.6  18.3%   56.5%  
Robin Soderling          4.6  19.5%   55.1%  
David Ferrer             4.5  16.8%   70.7%  
Alejandro Falla          4.5  19.7%   47.9%  
Gael Monfils             4.3  17.3%   44.8%  
Albert Montanes          4.3  15.2%   46.1%  
Fabio Fognini            4.3  15.5%   59.5%  
Novak Djokovic           4.1  16.0%   63.6%  
Richard Gasquet          4.0  13.9%   57.0%  
Roger Federer            3.9  14.0%   49.7%  
Ivan Ljubicic            3.7  11.8%   49.4%  
Stanislas Wawrinka       3.6  11.1%   46.2%

Unsurprisingly, the first two stats correlate quite closely.  The more eight-shot rallies you play, the higher your per-point average will be.  What may be more of a surprise is that the number of eight-shot rallies you play doesn’t appear to have much effect on your success in eight-shot rallies.  Andy Murray may be an instructive example here: He’s good at keeping himself in long points, but not always so good at doing what he needs to do to win them.

These numbers are far from authoritative–none of these stats comprise more than seven matches, and many comprise only four.  With so little data, a single opponent can skew the numbers.  For instance, Nadal was closer to the top of the rally-length leaderboard in the Australian Open, but a disproportionate number of his points came against John Isner, who is normally at the extreme other end.  Matches against Ljubicic and Federer also kept Nadal’s average down.

The same warning should be made about Ferrer’s impressive 70.7% winning percentage on long points.  I don’t doubt that he’s usually quite good in such rallies, but his four matches included two against players who are the exact opposite: Jarkko Nieminen and Sergiy Stakhovsky.

As more data of this sort becomes available, it will be interesting to see what trends emerge.

The French Open’s New Balls

Italian translation at settesei.it

Much has been made over the new balls at Roland Garros this year–players have complained that they are lighter, heavier, that they bounce differently.  As far as bounce and spin is concerned, there isn’t much we can glean from the available data.  But we can take a broad look at server dominance to get a sense of how the French is playing this year.

Several months ago, I looked at most of the ATP-level matches from 2010, and determined that the server wins points on different surfaces at the following rates:

  • Clay: 61.5%
  • Hard: 63.7%
  • Grass: 65.9%

The gaps between those numbers may not look very big, but they represent a major indicator in the differences between surfaces.  If the gap between clay and hard is 2.2%, then 2.2% must be a pretty big deal!

I’ve also determined the following regarding ace rates–again, using 2010 data.  “Ace rate” is simply the percentage of serves that are aces:

  • Clay: 5.5%
  • Hard: 8.5%
  • Grass: 10.5%

Now that’s a big difference.

Roland Garros

What about the French?  Taking the 2010 event as a whole, players won 62.4% of service points, and served aces 6.6% of the time.  Thus, I think it’s reasonable to conclude that the courts at RG last year played faster than most other clay-court ATP events.  Weather and–you guessed it–other equipment, such as balls, can also make a difference.

This year, 112 of the 127 men’s matches have already been played, including over 17,000 points, so I think it’s safe to start drawing conclusions.  This year, servers are only winning 62.0% of points–roughly halfway between the clay-court average and the results from last year’s “fast” RG.  More dramatically, players are only scoring aces on 5.6% of points, well below last year’s figure at the French.

I can’t shed any light on the specific quirks shown by the new balls, but for whatever reason, the French is playing more like an average clay-court event than it did last year.

Winners and Losers in the French Open Draw

Yesterday I offered a full breakdown of the French Open draw, with each player’s chances of advancing to each of several rounds.  In any draw of this kind, there are winners and losers, thanks to the luck of the … well, draw.

In a 128-player field seeded in the manner that Grand Slams are seeded, there are nearly 100,000 permutations of the draw.  The vast majority don’t matter — for instance, if you swapped Thiemo de Bakker and John Isner so that de Bakker played Nadal in the first round and Isner played Djokovic in the first round (instead of vice versa), no one’s chances of winning the title would change much.

But, of course, many of the possible permutations would matter a whole lot.  Just ask de Bakker or Isner!  Imagine how much better it would be for Isner to have a first round draw against, say, Yen-Hsun Lu, followed by a probable second-rounder against Sergiy Stakhovsky.  In fact, that’s Kei Nishikori’s draw, and in that sense, Nishikori was very lucky that the chips fell where they did.

Stepping back

In my previous draw simulations–like the one I published yesterday–I took the actual bracket as a given.  To generate the probabilities you see in yesterday’s chart, I had a computer program “play” the tournament 100,000 times, each time pitting Isner against Nadal in the first round, then the winner of that match against the winner of Giraldo/Andujar, and so on.

There’s a different way we could approach this.  Instead of starting the simulation from the point at which the draw is set, we could start from the point at which the field was set and seeded.  At that point, Isner would know that he is not seeded–and thus, that he would probably face a seed in the first or second round–but not which higher-ranked player he would face.

So, instead of 100,000 simulations of the actual French Open bracket, we can do 100,000 simulations of the draw itself, followed by simulating each ensuing bracket.  Sometimes, Isner draws Nadal, sometimes he draws Hanescu, and so on.

Measuring draw implications

A good way to gauge a player’s overall chances at a tournament is his predicted prize money.  Most players don’t have a significant chance of winning most tournaments (especially slams), so to compare Giraldo’s 0.01% chance of winning the title with Cuevas’s 0.02% chance doesn’t tell us much.  But if we consider the possibility that each player reaches each round, we can estimate that Giraldo will take home E24,600, while Cuevas will collect E29,500.  These numbers represent an average of the first-round prize money, second-round prize money, and so on, weighted by the probability that the player will reach each of those stages.

With this metric, we can compare the implications of the actual draw with the implications of the randomized draw, in which, for instance, Nadal could play any one of the 96 unseeded players in the first round.

Let’s compare the two outcomes in an extreme case.  As we’ve seen, the draw was not kind to John Isner.  My algorithm gives him a 12% chance of reaching the second round, and less than a 1% chance of reaching the semis.  Crunch the numbers, and you have predicted prize money of E22,700.  When you randomize the draw and he no longer has to beat Nadal in the first round, his chances of reaching the second round leap to 60%, and he has a 2% shot at a semifinal berth.  Predicted prize money: E40,100.

As it turns out, Isner is our biggest loser.  His predicted prize money fell more than 40% between the beginning and end of the draw ceremony.  What’s remarkable is that the next four players on the list all come from the same 1/16th of the draw–you guessed it, Djokovic’s section.

The draw effect on Thiemo de Bakker is similar to that on Isner–it doesn’t get any worse than drawing Djokovic in the first round.  Next on the list are Ernests Gulbis, Ivo Karlovic, and Juan Martin del Potro.  Karlovic and Gulbis not only have the misfortune of drawing Delpo in the first two rounds, but if by some chance they get past the Argentine, then they face Djokovic!  Each of those players lost more than 30% of their predicted prize money through the vagaries of the draw.

Del Potro is an interesting case.  As is, his predicted prize money is E184,600.  Before the draw was set, he could expect E266,000.  The biggest difference, of course, is his chance of reaching the round of 16.  In real life, he’ll need to beat Djokovic to get there, and he has a 30% chance of getting that far.  Before the bracket was drawn, the expectation was that he’d need only to defeat someone in the top 16 (or possibly, a player who had upset someone in the top 16).  He had a 63% chance of doing so.

Winners

Naturally, if there are so many players whose predicted prize money decreased, some players must have benefited from the way the draw played out.

One of the biggest winners was Andy Murray.  As we’ve seen, plenty of dangerous players are concentrated in Djokovic’s quarter; in fact, Djokovic, Nadal, and Federer were all hurt by the draw.  But Murray’s draw boosted his predicted prize money from E191,700 to E240,800.  He’ll face a qualifier or lucky loser in each of the first two rounds, then no one more challenging than Milos Raonic in the third.  Next would be Dolgopolov or Troicki–no walkovers, but compare that to Nadal’s possible fourth-rounder of Verdasco, and you see how the breaks went in Andy’s favor.

Murray’s quarter is the softest of the four, and other men benefit even more.  In fact, the two players whose chances the draw boosted the most are Nicholas Almagro and Jurgen Melzer, who will likely play in the fourth round for a matchup with Murray in the quarters.  Almagro, Melzer, and Juan Ignacio Chela (also in this section) all saw their predicted prize money jump by more than 40%.  For example, Almagro went from E76,100 to E112,400–and more than doubled his chances of winning the title from 0.6% to 1.3%–by landing where he did in the bracket.

Regardless of any player’s specific placement, the best man will probably win.  But the draw certainly has a say in how tricky the route to the title will be.

Breaking In and Breaking Through

Yesterday we looked at players who broke into the top 100 when they were teenagers.  As expected, those guys generally went on to great success–17 of the last 25 eventually reached the top 10, and at least two more may still do so.

We can gain a broader perspective by analyzing more than just teenagers.  If a 19-year-old entering the top 100 is likely to become a top-10 player, what chances do 22-year-olds or 26-year-olds have?  By examining a few decades of the ATP ranking system, we can begin to answer these questions.

I used a sample of 590 players–everyone who entered the top 100 between 1980 and 2005.  (It’s possible that a few recent players will continue to improve, but the vast majority of players get close to their peak within five years, so 2005 seems like  a reasonable cutoff date.)  A bit less than half of those 590 broke into the top 100 between the ages of 20 and 22, about a third were older, and the remainder were teenagers.

As you can see in the table below, there is a clear correlation between breaking into the top 100 at an early age and reaching the higher echelons of the pro game.  In the last 30 years, only one #1-ranked player (Pat Rafter) hadn’t reached the top 100 as a teenager, and he made it into the top 100 when he was 20.  Almost every eventual top-10 player had broken into the top 100 by age 21.

Age  Players  Top50  Top20  Top10  Top5  Top1
16         4   100%   100%   100%   75%   50%
17        16   100%    88%    69%   56%   38%
18        38    87%    76%    61%   34%   11%
19        61    89%    48%    41%   20%    8%
20        88    86%    48%    25%   13%    1%
21        99    63%    22%    12%    7%    0%
22        83    47%     8%     5%    2%    0%
23        61    44%    16%     3%    0%    0%
24        62    31%     3%     0%    0%    0%
25        32    25%     0%     0%    0%    0%
26        10    60%    10%     0%    0%    0%
27        16    31%     0%     0%    0%    0%
28        10    20%     0%     0%    0%    0%
29+       10     0%     0%     0%    0%    0%

It’s not entirely clear that these trends are consistent from decade to decade–yesterday, I noted that fewer teenagers had reached the top 100 in the last ten years or so.  It’s possible that as the quality of the game improves and a larger amount of training is necessary to prepare for the pro tour, there will be fewer prodigies like Nadal, who broke in at age 16, and Richard Gasquet, who arrived as a 17-year-old.

But even if the ages shift by a year or two, the overall conclusions should hold.  The older you are when you arrive in the top 100, the less likely it is that you will advance considerably further.

One obvious application of this data is to make predictions regarding players as they enter the top 100.  The last two men to break in are Benoit Paire (age 22) and Matthias Bachinger (age 24).  Paire is still young enough to have an outside shot at the top 10; Bachinger will have a hard time doing much better than #50.  Another recent newbie is Go Soeda, a 26-year-old.  To find someone who made a top-20 success out of so late a breakthrough, you have to go back to Steve Denton in the mid-80s.

Another way to use this information is to find top prospects among current players.  Among active tour pros, the four men who broke in at the youngest ages are Nadal, Gasquet, Juan Martin del Potro, and Novak Djokovic.  The next two might surprise you: Kei Nishikori and Donald Young.  Nishikori has only now recovered from battles with injury–perhaps he will start to make good on his promise.  Young may be a unique case–were it not for his many, many wildcards, he would not have reached the top 100 so early.

Another surprise is the active player with the 10th-youngest age-of-reaching-100: Evgeny Korolev.  The Russian has also struggled with injury, but he did crack the top 50 last year.

The more oft-mentioned “prospects” are a little further down the list.  Grigor Dimitrov broke in at 19.7 years of age, while Milos Raonic appeared just after his 20th birthday–a few days older than the first appearance of Mischa Zverev.  Alexander Dolgopolov is further down than you might expect, having broken in at age 21.3, while Ryan Sweeting didn’t get there until 23.5.

Of course, “age of first appearance in the top 100” is just one metric, and it doesn’t tell the whole story.  Perhaps players who spend several years in college account for that blip in the table at age 23–John Isner, for instance, didn’t reach the top 100 until he was nearly 23, and he has already hit a peak ranking of #18.  The metric might also underrate the chances of those who suffer prolonged injury at an early age–perhaps if Nishikori had lost his two years to injury one season sooner, he would have only recently reached the top 100 with the same skills and potential.

Warts and all, this angle is a good reminder of why we should keep a close eye on youngsters in the futures and challenger tours–the latest, greatest 23-year-old is almost guaranteed not to be the future of the sport.

Teenagers in the Top 100

If Ryan Harrison qualifies for the French Open and reaches the second round, he’ll probably break into the top 100. I wouldn’t bet on that degree of success at Roland Garros, but the relevant point is that the young American is close–if he falters in Paris, a couple of deep runs at challenger events will do the trick.

Harrison just turned 19, and he is the youngest player in the top 150. When Grigor Dimitrov turns 20 next Monday, Harrison will the be the top-ranked 19-year-old in the world. There is a widespread sense that reaching the top 100 is one measure of “making it,” and an equally popular notion that if a player hits that benchmark at such a young age, he is probably destined for success.

Indeed, hitting the top 100 as a teenager is rare, and it’s getting even less common.  Of the 940 players who have spent time in the top 100 in the history of the ATP ranking computer, fewer than 150 (16%) broke in when they were teenagers.  Since the beginning of 2001, 209 players have broken in, including only 25 teenagers (12%).

As you might expect, those 25 have generally gone on to very successful careers.  20 have reached the top 20, and 17 have climbed into the top 10.  It’s even better than that, since in time, Dimitrov and Kei Nishikori seem likely to make those numbers 22 and 19 out of 25.

If Harrison breaks into the top 100 by the end of July, he’ll become the 20th youngest player to do so since the beginning of 2001.  If we want to get technical and limit the span to exactly 10 years, he’ll become the 16th youngest player since mid-2001.  (Early 2001 was a good time for teenagers, with Jose Acasuso, Andy Roddick, Mikhail Youhzny, and Tommy Robredo all reaching the top 100 in the span of three months.)

Incidentally, Bernard Tomic has a chance to make an even more impressive mark, as he is five and a half months younger than Harrison.  However, he’s 50 spots and 130 points lower on the ranking computer, so his appearance in the top 100 as a teenager seems far less assured.

After the jump, see the full list of teenagers who reached the top 100 since 2001.

Continue reading Teenagers in the Top 100

The Odds of Breaking Back

Perhaps the most unquestioned piece of conventional wisdom in tennis is this: after breaking serve, a player is particularly vulnerable to being broken himself.  It certainly seems to be true–to take just one example, in the Isner/Karlovic match last week, there were only two breaks of serve, and they were consecutive.

As with most bits of conventional wisdom, it’s not clear exactly what people mean by it.  When Djokovic crushes someone 6-0 6-1, do we really think his serve is more vulnerable after each of his five or six breaks than it is after the one game his opponent holds?  When a player does break back, is he then more vulnerable in his next service game?

Today, I’ll try to address the more basic versions of the cliche.  The results are a bit surprising.

The dataset

I’m working with all of the 2011 Australian Open matches from courts where Hawkeye was in place.  That’s about 80 of the men’s singles matches, and roughly the same number of women’s matches.  I’ve run the numbers on both genders but will keep them separate, for reasons that will become clear.

These matches give us over 2,700 men’s games across about 300 sets, and nearly 2,000 women’s games over a bit more than 200 sets.

Breaking back: Men

At this year’s Aussie Open, 24% of all men’s games were service breaks.  If we take the conventional wisdom literally, we would hypothesize that in the game following a service break, another break would occur more than 24% of the time.

But it doesn’t.  In the game following a service break, the server is broken only 19.5% of the time.  (I’m excluding service breaks that end a set or take a set to a tiebreak.)  In other words, in the aggregate, a player is more likely to hold serve after breaking serve than he is after his opponent holds.

Of course, as I suggested by mentioning Djokovic a moment ago, there’s a huge selection bias here.  A player who breaks serve is (all else equal) likely to be a better player than one who doesn’t.  The best players in the most lopsided matches are breaking serve frequently, and because they are the better player, it makes sense that they are more likely (again, all else equal) to hold their own serve.

Without looking at individual matchups, it’s not immediately clear how to address this problem.  For one thing, I’m not convinced it’s a problem.  When Federer broke Kohlschreiber today, a commentator may have said, “Roger is particularly vulnerable here, let’s see if he can consolidate the break.”  One could easily respond: “Roger just showed us he’s in tremendous form; the very fact he just broke serve is an indication that he’s less vulnerable than usual on serve right now.”  And so it proved: Roger broke four times; Kohlschreiber never broke back.

What might be more instructive is to look at situations where the player who broke serve is considered to be roughly equal or inferior to his opponent.  Had Kohlschreiber broken serve early in the match, even given the assumption that he must be playing well in order to do so, the conventional wisdom would suggest that Federer is more likely to break back.  Perhaps that’s true.  It’s not something I can answer today–quantifying the matchups is beyond the scope of this afternoon project.  It’s also problematic in that it would also shrink the size of our already-small dataset.

In any event, it is clear that we can’t take this bit of conventional wisdom at face value.  It may be true in certain scenarios–some players may crumple under the pressure of consolidating a break, and others may rise to the occasion after losing serve.  But it is wrong to say that, in general, players are more vulnerable on serve after a break.

Breaking back: Women

As you might expect, breaks of serve are more prevalent in the women’s game, as are breaks-following-breaks.

At the 2011 AO, women broke serve 36.5% of the time.  In games following breaks of serve, they broke 36.0%.  In contrast to the men’s results, this suggests that in the women’s game, a service break doesn’t tell us as much about the strength of the player who has accomplished the break–or, if it does, that a server is more vulnerable after breaking serve.  Anecdotally, it certainly seems that differences in mental strength play a larger role in WTA matches, so I would expect that the break-back rate would be higher.

As I’ve said, this is far from the final word.  As usual, the conventional wisdom masks many subtleties that only further analysis can unearth.

Monte Carlo Projections

Clay court rankings and projections are tough.  Most of the top 50 ATP players only compete on clay for a couple months of the year, so at the beginning of the clay-court swing, we’re using surface-specific results from almost a year ago; at the end, we’re depending heavily on each player’s recent results.

Of course, the very top of the list is easy.  Rafael Nadal hasn’t lost a clay court match since the 2009 French Open.  It gets messy soon after that, since Nadal left the rest of the field fighting for crumbs.  Roger Federer is the clear #2 in this field, with Andy Murray a distant third.  (And maybe he should be even more distant.)

These projections are clay-specific, as you can probably tell by some of the percentages.  My clay rankings, however, are heavily regressed back to overall rankings so, for instance, Milos Raonic gets plenty of credit for his recent success on hard courts.  (And today he justified that credit.)

The tournament organizers made it tough for me to do pure projections, since four main draw matches were complete by the time qualifiers were placed.  Thus, the numbers are below are done as if I didn’t know anything about the outcome of today’s main draw matches.  I used rankings generated last Monday, so I might be selling Potito Starace (and, to a lesser extent, Victor Hanescu) a little short by excluding results from Casablanca.

Enjoy!

 

Player                   R32   R16    QF    SF     F     W
(1)Rafael Nadal         100% 92.7% 79.9% 66.8% 52.4% 36.6%
(q)Julien Benneteau      54%  4.4%  1.7%  0.6%  0.2%  0.0%
Jarkko Nieminen          46%  2.9%  1.0%  0.3%  0.1%  0.0%
(q)Vincent Millot        26%  5.6%  0.3%  0.1%  0.0%  0.0%
Guillermo Garcia-Lopez   74% 31.8%  5.5%  2.1%  0.7%  0.2%
Denis Istomin            27% 12.8%  1.3%  0.4%  0.1%  0.0%
(13)Richard Gasquet      73% 49.8% 10.3%  5.3%  2.4%  0.9% 

Player                   R32   R16    QF    SF     F     W
(12)Jo-Wilfried Tsonga   59% 35.2% 17.0%  4.0%  1.7%  0.6%
Juan Monaco              41% 21.6%  8.9%  1.8%  0.7%  0.2%
Ivan Ljubicic            55% 25.5% 10.5%  2.2%  0.9%  0.2%
Jeremy Chardy            45% 17.7%  6.3%  1.1%  0.3%  0.1%
(q)Olivier Rochus        46%  8.7%  2.6%  0.3%  0.1%  0.0%
Juan Ignacio Chela       54% 12.4%  4.2%  0.6%  0.1%  0.0%
(5)(WC)Tomas Berdych    100% 78.9% 50.5% 14.7%  7.8%  3.2% 

Player                   R32   R16    QF    SF     F     W
(3)(WC)Andy Murray      100% 70.6% 54.3% 38.4% 16.5%  8.7%
(WC)Radek Stepanek       30%  5.9%  2.5%  0.9%  0.2%  0.0%
Marcos Baghdatis         70% 23.5% 14.5%  7.8%  2.3%  0.8%
Albert Montanes          64% 30.9%  8.9%  4.0%  0.8%  0.2%
Xavier Malisse           36% 11.1%  2.0%  0.6%  0.1%  0.0%
Thomaz Bellucci          52% 30.9% 10.1%  4.7%  1.1%  0.3%
(16)Gilles Simon         48% 27.0%  7.7%  3.4%  0.7%  0.2% 

Player                   R32   R16    QF    SF     F     W
(10)Mikhail Youzhny      52% 38.1% 18.3%  7.0%  1.7%  0.5%
Florian Mayer            48% 33.3% 13.8%  4.8%  1.1%  0.3%
(q)Frederico Gil         41% 10.2%  2.1%  0.3%  0.0%  0.0%
Sergiy Stakhovsky        59% 18.4%  5.0%  1.1%  0.1%  0.0%
Daniel Gimeno-Traver     40%  7.9%  2.9%  0.5%  0.1%  0.0%
Santiago Giraldo         60% 15.6%  6.6%  1.8%  0.2%  0.1%
(8)Gael Monfils         100% 76.5% 51.3% 24.7%  7.7%  3.0% 

Player                   R32   R16    QF    SF     F     W
(6)Fernando Verdasco    100% 71.3% 54.5% 31.0% 13.6%  5.2%
Tommy Robredo            73% 24.6% 15.1%  5.7%  1.6%  0.4%
Ivan Dodig               27%  4.0%  1.5%  0.3%  0.0%  0.0%
Kevin Anderson           44% 16.8%  3.4%  0.7%  0.1%  0.0%
Fabio Fognini            56% 24.1%  6.0%  1.5%  0.3%  0.0%
(WC)Jean-Rene Lisnard    14%  3.4%  0.3%  0.0%  0.0%  0.0%
(11)Viktor Troicki       86% 55.6% 19.2%  6.5%  1.7%  0.4% 

Player                   R32   R16    QF    SF     F     W
Ernests Gulbis           67% 44.8% 22.0% 12.2%  5.0%  1.8%
(14)Alexandr Dolgopolov  33% 16.9%  5.5%  2.2%  0.6%  0.1%
Milos Raonic             59% 24.4%  8.8%  3.9%  1.1%  0.3%
Michael Llodra           41% 13.9%  4.1%  1.4%  0.3%  0.1%
Janko Tipsarevic         55% 13.5%  5.3%  1.9%  0.4%  0.1%
Feliciano Lopez          45% 10.7%  3.7%  1.2%  0.3%  0.0%
(4)David Ferrer         100% 75.8% 50.7% 31.4% 14.3%  6.0% 

Player                   R32   R16    QF    SF     F     W
(7)Jurgen Melzer        100% 54.7% 29.3%  9.3%  4.2%  1.3%
Nikolay Davydenko        72% 37.5% 20.3%  6.8%  3.2%  1.1%
Robin Haase              28%  7.8%  2.6%  0.5%  0.1%  0.0%
(q)Maximo Gonzalez       55% 12.3%  2.8%  0.3%  0.1%  0.0%
Victor Hanescu           45%  8.6%  1.8%  0.2%  0.0%  0.0%
Marcel Granollers        16%  8.5%  2.1%  0.3%  0.1%  0.0%
(9)Nicolas Almagro       84% 70.5% 41.1% 15.2%  7.7%  2.8% 

Player                   R32   R16    QF    SF     F     W
(15)Marin Cilic          86% 68.2% 19.9% 10.4%  4.8%  1.5%
(q)Filippo Volandri      14%  5.7%  0.6%  0.1%  0.0%  0.0%
(q)Pere Riba             43%  9.7%  1.0%  0.2%  0.0%  0.0%
Potito Starace           57% 16.4%  1.7%  0.3%  0.1%  0.0%
Philipp Kohlschreiber    61% 11.6%  6.4%  3.0%  1.2%  0.3%
Andrey Golubev           39%  5.2%  2.4%  0.9%  0.3%  0.1%
(2)Roger Federer        100% 83.2% 68.1% 52.6% 39.0% 22.1%

Net-rushing, or The Stats We Don’t Have

In yesterday’s morning recap, I made the following comment about Nadal’s baseline game:

The one baffling thing is Nadal’s reluctance to come to net.   He was often standing right on the baseline, even hitting groundstrokes from a step inside the baseline.  Yet he almost never came forward unless forced.  Even with an imperfect net game, even against the passing-shot machine that is Djokovic, I think he would’ve been more successful taking advantage of some of those offensive positions.

In the comments, Tom Welsh laid out the flip side of the argument concisely:

During the Nadal-Djokovic match yesterday I noticed several occasions when each of those brilliant players came in to the net and was left looking like a hopeless beginner – either by a passing shot, or a sizzling ground stroke to the short ribs, or by a perfect lob landing just a couple of feet inside the baseline. I’m not tennis player, but it seems to me that no one can afford to come in these days unless the opponent is stretched to the breaking point. Even then, it’s taking a big risk.

That’s the argument in a nutshell.  Even more briefly:

  • PRO: Players should be more aggressive and come forward more often.
  • CON: In the modern-day game, approaching the net is usually too risky.

Which is it?

Pick your poison

The first thing that needs to be understood is that, against an elite tennis player, anything is a risk.  Short of a decisive smash, any shot you hit is likely to come back, and there’s a non-zero chance that what comes back is going to be a winner.  Choosing to come forward isn’t a decision between risk and no risk, it’s a matter of degree.

The main difference is that, if you come forward and fail, you’ll look like a fool, and your opponent will look brilliant, in the ways Tom described.  If you stay back and fail, it’s somehow more understandable–in a 15-stroke rally between top players, somebody has to lose.  Of course, you lose the point either way.

One of the problems of arguing this point with anecdotal evidence is that I think we, as both fans and players, remember the brilliant passing shots and jaw-dropping lobs.  If you rush the net and your opponent misses what would’ve been a sensational running forehand, you remember the amazing shot-that-almost-was.  Human brains don’t default to probabilistic calculations, while brilliant moments catch and keep our attention.

Commentators, steeped in strategy of the 70’s and 80’s, will always want too much net-rushing.  Most players will tend to stay back too much.  If we can ever establish the proper opportunities to come forward, the “correct” answer will turn out to be somewhere in between.

Where the stats fail

Answering the question analytically will be very difficult, and given the information currently out there, it’s flat-out impossible.

In the meantime, let’s think through what it would take to answer the question.  Starting with what I take to be the question itself:

Given his skillset, his opponent’s skillset, and each player’s position on the court, when should a player come forward?

That’s a lot of stuff we can’t quantify.  Even if we posit a couple of generic pro players, it’s still an unanswerable question.

Particularly useless are existing net stats.  Occasionally during a match, a broadcast will show us that so-and-so has won 5 out of 8 points at net.  The commentators reliably chime in, usually suggesting that the player has better net skills that we give him credit for (perhaps he’s been playing some doubles lately), and that he would benefit by coming in more.

In most cases, those 8 points couldn’t be less relevant.  Think back to Sunday’s Nadal-Djokovic match.  Much of the time Nadal came forward, it was in response to a Djokovic drop shot.  In other words, Nadal came forward on the defense!  I’m guessing he lost most of those points.  On the flip side, imagine Del Potro cracking a serve out wide, then coming in behind it to hit a swinging volley winner.  That’s 1-for-1 on the net point tally, but it doesn’t say a thing about Delpo’s deftness of touch around the net.

A framework

Let’s imagine that we suddenly had access to Hawkeye’s shot-by-shot data.  We’d know the hit point for every ball of every point of every match where the Hawkeye system was installed.  (Drool.)

If we knew that, we could come up with a fairly simple model to estimate the likelihood of winning a point from any position on the court, against a certain quality of shot.  Standing at the middle of the baseline smacking a 60 mph service return, you might have a 70% chance of winning the point.  Stuck in the backhand corner after your opponent has cracked a 90mph groundstroke, and it might be more like 20%.

The details of the model aren’t important.  What matters is that, with a certain data set, we could estimate the probability of winning a point given a variety of conditions.

Extending this framework to analyze a tactic like net-rushing wouldn’t be that complicated.  Let’s say Nadal is standing on the baseline with a 70% chance of winning the point.  No matter what he does afterward, he will probably hit a forehand into one corner or the other, after which we can once again estimate his probability of winning the point.  From there, he has two choices: Come forward, or stay back.  Some game theory might get involved, since his opponent will probably see him approach the net and may change his own strategy accordingly.

Again, we can work out the details when there is data to play with.  Given these relatively simple figures, we could estimate Nadal’s probability of winning the point coming forward behind his forehand and staying back after hitting the shot.

The numbers would give a better way of judging whether a particular play is advisable.  For Nadal, it may turn out that staying back is always smarter–after all, the numbers will probably tell us that, from any given position, he has a better chance than nearly anyone else of winning the point.  But, say, Ivo Karlovic may be better off coming in behind the exact same shot from the same position.  There’s a continuum between the extremes, of course, and we’ll need to know a lot more before we know what that looks like.

In the meantime, I’d still like to see Nadal come forward–and I’ll try harder to remember the times when his opponent goes for a blistering passing shot and misses.