Category: Forecasting

Forecasting the Davis Cup Finals

It took more than a year to decide on a new format, but barely a week to make the draw. With 12 countries qualifying for the inaugural Davis Cup Finals in home-and-away ties earlier in month, the field of 18 is set. Using the ITF’s own system to rank countries, the 18 teams were divided into three “pots,” then assigned to the six round-robin groups that will kick off the tournament this November in Madrid.

The new format sounds complicated, but as round-robin events go, it’s easy enough to understand. Each of the six round-robin groups will send a winning team to the quarter-finals. Two second-place sides will also advance to the final eight, as determined by matches won, then sets won, and so on as necessary, until John Isner and Ivo Karlovic stand back to back to determine which one is really taller. From that point, it’s an eight-team knock-out tournament.

Here are the groups, as determined by yesterday’s draw, with seeded countries indicated:

  • Group A: France (1), Serbia, Japan
  • Group B: Croatia (2), Spain, Russia
  • Group C: Argentina (3), Germany, Chile
  • Group D: Belgium (4), Australia, Colombia
  • Group E: Great Britain (5), Kazakhstan, Netherlands
  • Group F: United States (6), Italy, Canada

The ITF ranking system considers the last four years of Davis Cup results, so Spain’s brief exit from the World Group makes the seedings a bit wonky. As it turns out, not only is it a top team (Croatia) who will have to deal with early ties against the Spaniards, the entire Group B trio constitutes a group of death. Russia would be an up-and-coming squad in any format, and it is clearly the most dangerous of the six lowest-ranked sides.

Madrid to Monte Carlo

Last week, I introduced a more accurate, predictive rating system for Davis Cup, involving surface-specific Elo ratings for the players likely to compete. Those rankings put Spain at the top, Croatia second, Russia fifth, and fourth-seeded Belgium 14th in the 18-team field.

Now that we have a draw, we can use those ratings to run Monte Carlo simulations of the entire Davis Cup carnival Finals. As in my post last week, I’m estimating that singles players have a 75% chance of playing at any given opportunity and doubles players have an 85% chance. Those are just guesses–there’s no data involved in this step. Surely some teams are more fragile than others, perhaps because their stars are particularly susceptible to injury or just uninterested in the next event. I’ve excluded Andy Murray, but for the moment, I’m keeping Novak Djokovic and Alexander Zverev in the mix.

(We’re using Elo ratings for each individual player, which means the simulation is telling us what would be likely to happen if it were played today. Things will change between now and November, even if every eligible player shows up. A proper forecast that takes the time lag into account would probably give a slight boost for younger teams [whose players will have nine months to mature] and a penalty for older ones [who are more likely to be hit by injury]. And overall, it would shift all of the championship probabilities a bit toward the mean.)

Here are the results of 100,000 simulations of the draw, with percentages given for each country’s chance of winning their group, then reaching each of the knock-out rounds:

Country  Group     QF     SF      F      W  
ESP      46.1%  59.1%  41.9%  30.3%  19.3%  
FRA      54.2%  66.6%  40.6%  25.1%  14.6%  
AUS      74.5%  84.4%  46.0%  23.8%  12.1%  
USA      53.0%  65.5%  36.8%  19.7%  10.4%  
CRO      31.0%  43.0%  27.2%  17.8%   9.8%  
GER      52.5%  67.9%  39.7%  17.6%   7.7%  
RUS      22.9%  33.1%  19.5%  12.0%   6.1%  
SRB      33.0%  47.9%  24.1%  12.6%   6.0%  
GBR      66.8%  78.7%  35.9%  12.5%   4.4%  
ARG      39.7%  56.6%  28.6%  10.4%   3.8%  
ITA      24.3%  35.9%  14.6%   5.5%   2.1%  
CAN      22.7%  33.4%  13.1%   4.9%   1.8%  
JPN      12.8%  19.5%   7.2%   2.8%   0.9%  
BEL      20.3%  32.0%   8.5%   2.1%   0.6%  
NED      21.7%  35.5%   8.6%   1.7%   0.3%  
CHI       7.8%  12.9%   3.4%   0.6%   0.1%  
KAZ      11.5%  19.0%   3.2%   0.5%   0.1%  
COL       5.1%   8.9%   1.2%   0.1%   0.0%

Spain is our clear favorite, despite their path through the group of death. Five teams have a better chance of winning their group and reaching the quarters than the Spaniards do, but their chances in the single-elimination rounds make the difference. At the other extreme, Australia seems to be the biggest beneficiary of draw luck. My rankings put them sixth, and they landed in a group with Belgium (the lowest-rated seed) and Colombia (the weakest team in the field). Their good fortune makes them the most likely country to reach the final four, even if Spain and France have a better chance of advancing to the championship tie.

Less randomness, more Spain

What if we run the simulation one step earlier in the process? That is to say, ignore yesterday’s draw and see what each country’s chances were before their round-robin assignments were determined. For this simulation, we’ll keep the ITF’s seeds, so Spain is still a floater. Here’s how it looked ahead of the ceremony:

Country  Group     QF     SF      F      W  
ESP      63.0%  75.9%  52.9%  35.0%  22.6%  
FRA      56.8%  70.8%  43.9%  25.7%  14.5%  
CRO      55.5%  69.4%  42.2%  25.1%  13.5%  
USA      51.3%  65.6%  38.5%  19.8%  10.0%  
AUS      48.3%  62.9%  34.8%  17.7%   8.5%  
RUS      40.6%  53.5%  30.2%  15.8%   7.9%  
SRB      42.9%  55.8%  28.3%  13.5%   5.9%  
GER      42.0%  55.7%  27.3%  12.5%   5.4%  
ARG      35.9%  49.1%  20.9%   7.9%   2.8%  
ITA      33.6%  47.1%  19.2%   7.2%   2.5%  
GBR      34.9%  48.3%  20.3%   7.5%   2.5%  
CAN      24.5%  35.5%  14.3%   5.3%   1.9%  
JPN      19.8%  29.4%  10.6%   3.6%   1.1%  
BEL      20.9%  30.4%   7.5%   1.8%   0.4%  
NED       9.5%  15.5%   3.5%   0.7%   0.1%  
CHI       7.9%  13.3%   2.6%   0.4%   0.1%  
KAZ       8.4%  14.1%   2.1%   0.3%   0.0%  
COL       4.3%   7.5%   1.1%   0.2%   0.0%

With the “group of death” out of the picture, Croatia jumps from fifth to third, swapping places with Australia. The defending champs lost the most from the draw, while Spain suffered a bit as well.

Elo in charge

Another variation is to ignore the ITF rankings and generate the entire draw based on my Elo-based ratings. In this case, the top six seeds would be Spain, Croatia, France, USA, Russia, and Australia, in that order. Argentina and Great Britain would fall to the middle group, and Belgium would drop to the bottom third. Here’s how that simulation looks:

Country  Group     QF     SF      F      W  
ESP      71.6%  82.8%  57.3%  38.0%  24.1%  
FRA      64.6%  77.6%  45.8%  26.7%  14.4%  
CRO      63.1%  76.3%  45.8%  25.6%  13.6%  
USA      59.7%  73.3%  41.1%  20.2%  10.2%  
RUS      58.6%  71.2%  37.0%  19.7%   9.5%  
AUS      57.7%  71.4%  37.7%  17.7%   8.8%  
SRB      37.1%  53.0%  26.1%  12.1%   5.3%  
GER      35.3%  52.3%  24.5%  10.9%   4.6%  
ARG      28.0%  44.2%  17.5%   6.4%   2.2%  
ITA      27.4%  43.6%  16.9%   6.2%   2.1%  
GBR      27.0%  43.1%  16.5%   6.0%   2.0%  
CAN      26.7%  41.8%  16.0%   5.8%   2.0%  
JPN      15.9%  23.6%   8.1%   2.6%   0.8%  
BEL       9.4%  15.1%   3.9%   0.9%   0.2%  
NED       6.5%  10.8%   2.3%   0.5%   0.1%  
CHI       5.3%   9.0%   1.8%   0.3%   0.1%  
KAZ       3.2%   5.8%   0.9%   0.1%   0.0%  
COL       3.1%   5.2%   0.8%   0.1%   0.0%

The big winners in the Elo scenario are the Russians, who gain a seed and avoid a round-robin encounter with either Spain or Croatia. Australia gets a seed as well, but the benefit of protection from the powerhouses isn’t as valuable as the luck than shone on the Aussies in the actual draw.

Imagine a world with no rankings

Finally, let’s see what happens if we ignore the rankings altogether. It would be unusual for the tournament to take such an approach, but if there’s ever a time to have a tennis event with no seedings, this is it. The existing rankings are far too dependent on years-old results, leaving young teams at a disadvantage. And my system, while more accurate, doesn’t quite feel appropriate either. It is based on individual player ratings, and this is a team event.

Whatever the likelihood of a ranking-free draw in the Davis Cup future, here’s what a simulation looks like with completely random assignment of nations into round-robin groups:

Country  Group     QF     SF      F      W  
ESP      62.8%  75.4%  52.4%  34.8%  22.5%  
FRA      54.8%  68.6%  42.6%  25.0%  13.9%  
CRO      53.4%  67.2%  41.0%  23.6%  13.0%  
USA      48.8%  62.9%  35.9%  19.1%   9.7%  
RUS      47.9%  61.0%  34.8%  18.5%   9.3%  
AUS      47.1%  61.1%  34.1%  17.6%   8.5%  
SRB      41.5%  54.3%  28.0%  13.5%   6.1%  
GER      40.3%  53.6%  26.7%  12.3%   5.3%  
ARG      31.9%  44.9%  18.8%   7.2%   2.6%  
ITA      31.5%  44.2%  18.6%   7.1%   2.5%  
GBR      30.7%  43.4%  17.6%   6.5%   2.3%  
CAN      30.4%  42.7%  17.4%   6.4%   2.2%  
JPN      25.9%  36.4%  13.5%   4.6%   1.4%  
BEL      17.2%  25.9%   7.2%   1.8%   0.4%  
NED      12.5%  20.0%   4.6%   0.9%   0.2%  
CHI      10.4%  16.9%   3.5%   0.6%   0.1%  
KAZ       7.0%  11.8%   1.9%   0.3%   0.0%  
COL       5.9%   9.7%   1.5%   0.2%   0.0%

Round-robin formats do a decent job of surfacing the best teams, so the fully random approach doesn’t give us wildly different results than the seeded simulations. The main effect of the no-seed version is to give the weakest sides a slightly better chance at advancing past the group stage, since there is a better chance for them to avoid strong round-robin competition.

Madrid or Maldives redux

Some top players are likely to skip the event. Zverev has said he’ll be in the Maldives, and Djokovic has hinted he may miss the tournament as well. The new three-rubber format means that teams will suffer a bit less from the absence of a singles star, assuming he also isn’t one of the best doubles options as well. Still, both Germany and Serbia would much rather head to the party with a top-three singles player on their side.

Here are the results of the intial simulation–based on the actual draw–but without Djokovic or Zverev:

Country  Group     QF     SF      F      W  
ESP      46.5%  59.5%  44.0%  33.2%  21.3%  
FRA      68.2%  79.3%  49.6%  30.6%  17.8%  
AUS      74.3%  84.5%  46.1%  24.2%  12.6%  
USA      53.4%  66.2%  37.5%  20.4%  10.8%  
CRO      30.3%  42.5%  28.4%  19.6%  10.8%  
RUS      23.2%  33.6%  21.1%  13.8%   7.0%  
GBR      67.0%  79.0%  40.9%  14.6%   5.2%  
ARG      52.1%  66.9%  35.5%  12.9%   4.9%  
GER      36.4%  52.3%  23.3%   7.2%   2.2%  
ITA      24.2%  35.9%  14.5%   5.7%   2.2%  
CAN      22.4%  33.2%  13.4%   5.2%   2.0%  
JPN      19.4%  31.7%  11.5%   4.8%   1.6%  
BEL      20.5%  32.4%   8.6%   2.3%   0.6%  
SRB      12.4%  21.1%   6.0%   1.9%   0.5%  
NED      21.6%  35.5%   9.8%   2.0%   0.4%  
CHI      11.4%  18.5%   4.9%   0.9%   0.2%  
KAZ      11.3%  19.1%   3.8%   0.5%   0.1%  
COL       5.2%   9.0%   1.2%   0.2%   0.0%

Germany’s chances of winning the inaugural Pique Cup would fall from 7.7% to 2.2%, and Serbia’s odds drop from 6.0% to 0.5%. Argentina and France, the seeded teams sharing groups with Germany and Serbia, respectively, would be the biggest gainers from such high-profile absences.

Anybody’s game

I’ve been skeptical of the new Davis Cup, and while I remain unconvinced that it’s an improvement, I find myself getting excited for the weeklong tennis hootenanny in Madrid. These simulations were even more encouraging. As always, the ranking and seeding isn’t the way I’d do it, but in this format, the differences are minimal. The event format will give us a chance to see plenty of tennis from every qualifying nation, and the high level of competition from most of these countries ensures that most teams have a shot at going all the way.

Top Seed Upsets in ATP 250s

Italian translation at settesei.it

In a typical week, no one would notice if Fabio Fognini, Karen Khachanov, and Lucas Pouille combined to go 0-3. This week is different, as those three men held the top seeds at the ATP events in Cordoba, Sofia, and Montpellier. After their first-round byes, each of them lost in the second round, to Aljaz Bedene, Matteo Berrettini, and Marcos Baghdatis, respectively. At least two of the top seeds pushed their opponents to three sets, while Fognini lasted only 71 minutes.

This is not the first time a trio of number one seeds have suffered first-match upsets in the same week. Amazingly, it’s not even the first such occurrence in this very week on the calendar. Two years ago, when the South American event was played in Quito, the results were the same: top seeds Marin Cilic, Ivo Karlovic, and Dominic Thiem all failed to win a match. Thiem’s vanquisher, Nikoloz Basilashvili, even extended the streak the following week, heading to Memphis and handing Karlovic his second straight second-round ouster.

Predictable upsets?

Focusing on these losses, it’s natural to wonder whether top seeds are particularly fragile in this sort of tournament. There’s certainly a logic to it. The number one seed at an ATP 250 is usually ranked in the top 20, and is the sort of player who might have considered taking the week off. He knows that more ranking points are available at slams and Masters, so winning a smaller event isn’t his highest priority. His opponent, on the other hand, is competing every chance he gets, and the points on offer at a smaller event could make a big difference in his standing. Further, he has already played–and won–his first-round match, so he might be performing better than usual, or the conditions might suit him particularly well.

Let’s put it to the test. Since 2010, not counting this week’s carnage, I found 267 non-Masters events at which a top seed got a first-round bye and completed his second-round match. (Additionally, there have been three retirements and one withdrawal; only one of those resulted in a loss for the top seed.) The number one seeds had a median rank of 10, and the underdogs had a median rank of 89. Based on my surface-weighted Elo ratings at the time of each match, the favorites should have won 81.5% of the time. That’s better than this week’s trio of top-seeded losers, who were 64% (Fognini), 80% (Khachanov), and 69% (Pouille) favorites.

As it happened, the unseeded challengers were more successful than expected. The favorites won only 76.8% of those matches–a rate low enough that there is only a 3% probability it is due to chance alone. It’s not an overwhelming effect–certainly not enough that we should have predicted this week’s results–but it seems that a few of the top seeds are showing up unmotivated and a handful of the underdogs are playing better than expected.

Riding the wave

What about the underdog winners? Once they’ve defeated the top seed, how many capitalize on the opportunity? Berrettini came back to beat Fernando Verdasco in his quarter-final match today, while Baghdatis and Bedene play later. My forecasts believe that, of the three, Bedene has the best chance of claiming a title, though still less than a one-in-five shot at doing so.

In our subset of 267 matches, the underdog won 66 of them. More than half the time, though, that was the end of the run. 38 of the 66 (58%) fell in the quarter-finals. Another 17 lost in the semis. Whatever works so well for these underdogs in the second round disappears afterward. In the 105 matches contested by these 66 men in the quarter-finals and beyond, Elo thinks they should have won 44.9% of them. Instead, they managed only 42.3%.

There’s still a bit of hope. Five men knocked out the top seed in the second round and went on to win the entire tournament. One of those was a challenger we’ve already mentioned: Estrella, who knocked out Karlovic and went on to hoist the trophy in Quito two years ago. Maybe there’s some magic in week six. This week’s trio of underdogs would surely love to think so.

Picking Favorites With Better Davis Cup Rankings

Yesterday, the ITF announced the seedings for the first new-look Davis Cup Finals, to be held in Madrid this November. The 18-country field was completed by the 12 home-and-way ties contested last weekend. Those 12 winners will join France, Croatia, Spain, and USA (last year’s semi-finalists) along with the two wild cards, recent champions Argentina and Great Britain.

The six nations who skipped the qualifying round will make up five of the top six seeds. (Spain is 7th, while Belgium, who had to qualify, is 4th.) The preliminary round of the November event will feature six round-robin groups of three, each consisting of one top-six seed, a second country ranked 7-12, and a third ranked 13-18. Seeding really matters, as a top position (deserved or not!) guarantees that a side will avoid dangerous opponents like last year’s finalists France and Croatia. Even the difference between 12 and 13 could prove decisive, as a 7-through-12 spot ensures that a nation will steer clear of the always-strong Spaniards, who are seeded 7th.

The seeds are based on the Davis Cup’s ranking system, which relies entirely on previous Davis Cup results. While the formula is long-winded, the concept is simple: A country gets more points for advancing further each season, and recent years are worth the most. The last four years of competition are taken into consideration. It’s not how I would do it, but the results aren’t bad. Four or five of the top six seeds will field strong sides, and one of the exceptions–Great Britain–would have done so had Andy Murray’s hip cooperated. Spain is obviously misranked, but given the limitations of the Davis Cup ranking system, it’s understandable, as the 2011 champions spent 2015 and 2016 languishing outside the World Group.

We can do better

The Davis Cup rankings have several flaws. First, they rely heavily on a lot of old results. If we’re interested in how teams will compete in November, it doesn’t matter how well a side fared three or four years ago, especially if some of their best players are no longer in the mix. Second, they don’t reflect the change in format. Until last year, doubles represented one rubber in a best-of-five-match tie. A good doubles pair helped, but it wasn’t particularly necessary. Now, there are only two singles matches alongside the doubles rubber. The quality of a nation’s doubles team is more important than it used to be.

Let’s see what happens to the rankings when we generate a more forward-looking rating system. Using singles and doubles Elo, I’m going to make a few assumptions:

  • Each country’s top two singles players have a 75% chance of participating (due to the possibility of injury, fatigue, or indifference), and if either one doesn’t take part, the country’s third-best player will replace him.
  • Same idea for doubles, but the top two doubles players have an 85% chance of showing up, to be replaced by the third-best doubles player if necessary.
  • The three matches are equally important. (This isn’t technically true–the third match is likely to be necessary less than half the time, though when it does decide the tie, it is twice as important as the other two matches.)
  • Andy Murray won’t play.

Those assumptions allow us to combine the singles and doubles Elo ratings of the best players of each nation. The result is a weighted rating for each side, one that has a lot of bones to pick with the official Davis Cup rankings.

Forward-looking rankings

The following table shows the 18 countries at the Davis Cup finals along with the 12 losing qualifiers. For each team, I’ve listed their Davis Cup ranking, and their finals seed (if applicable). To demonstrate my results, I’ve shown each nation’s weighted Elo rank and rating and their hard-court Elo rank and rating. The table is sorted by hard-court Elo:

Country  DC Rank  Seed  Elo Rank   Elo  sElo Rank  sElo  
ESP            7     7         1  1936          1  1891  
CRO            2     2         2  1898          2  1849  
FRA            1     1         3  1880          3  1845  
USA            6     6         4  1876          4  1835  
RUS           21    17         7  1855          5  1827  
AUS            9     9         5  1857          6  1820  
SRB            8     8         8  1849          7  1808  
GER           11    11         6  1855          8  1799  
AUT           16              10  1800          9  1766  
ARG            3     3         9  1803         10  1755  
                                                         
Country  DC Rank  Seed  Elo Rank   Elo  sElo Rank  sElo  
GBR            5     5        11  1796         11  1750  
SUI           24              14  1763         12  1749  
ITA           10    10        12  1780         13  1745  
CAN           14    13        13  1777         14  1744  
JPN           17    14        15  1735         15  1719  
BEL            4     4        17  1688         16  1673  
CZE           13              16  1712         17  1661  
NED           19    16        18  1685         18  1643  
BRA           28              20  1659         19  1638  
IND           20              21  1652         20  1621  
                                                         
Country  DC Rank  Seed  Elo Rank   Elo  sElo Rank  sElo  
SVK           29              22  1645         21  1617  
CHI           22    18        19  1682         22  1609  
KAZ           12    12        26  1582         23  1574  
COL           18    15        24  1597         24  1551  
SWE           15              27  1570         25  1542  
BIH           27              28  1552         26  1540  
POR           26              23  1610         27  1535  
HUN           23              25  1583         28  1533  
UZB           25              29  1491         29  1489  
CHN           30              30  1468         30  1465

Spain is the comfortable favorite, regardless of whether we look at overall Elo or hard-court Elo. When the draw is conducted, we’ll see which top-six seed is unlucky enough to end up with the Spaniards in their group, and whether the hosts will remain the favorite.

The biggest mismatch between the Davis Cup rankings and my Elo-based approach is in our assessment of the Russian squad. Daniil Medvedev is up to sixth in my singles Elo ratings, with Karen Khachanov at 10th. Those ratings might be a little aggressive, but as it stands, Russia is the only player with two top-ten Elo singles players. Spain is close, with Rafael Nadal ranked 2nd and Roberto Bautista Agut 11th, and the hosts have the additional advantage of a deep reservoir of doubles talent from which to choose.

In the opposite direction, my rankings do not forecast good things for the Belgians. David Goffin has fallen out of the Elo top 20, and there are no superstar doubles players to pick up the slack. In a just world, Spain and Belgium will land in the same round-robin group–preferably one without the Russians as well.

Madrid or Maldives

The results I’ve shown assume that every top singles player has the same chance of participating. That’s certainly not the case, with high-profile stars like Alexander Zverev telling the press that they’ll be spending the week on holiday in the Maldives. Some teams are heavily dependent on one singles player who could make or break their chances with a decision or an injury.

As it stands, Germany is 8th in the surface-weighted Elo. If we take Zverev entirely out of the mix, they drop to a tie for 14th with Japan. It’s something the German side would prefer to avoid, but it’s not catastrophic, partly because the Germans were never among the favorites, and partly because Zverev could play only one singles rubber per tie and the doubles replacements are competent.

Even more reliant on a single player is the Serbian side, which qualified last weekend without the help of their most dangerous threat, Novak Djokovic. With Djokovic, the Serbs rank 7th–a case where my surface Elo ratings almost agree with the official rankings. But without the 15-time major winner, the Serbs fall down to a tie with Belgium in 16th place. While the Serbs are unlikely to take home the trophy regardless, Novak would make a huge difference.

The draw will take place next Thursday. We’ll check back then to see which sides have the best forecasts, nine months out from the showdown in Madrid.

What I Should’ve Known About Playing Styles and Upsets

In the podcast Carl Bialik and I recorded yesterday, I mentioned a pet theory I’ve had for awhile, that upsets are more likely in matches between players with contrasting styles. The logic is fairly simple. If you have two counterpunchers going at it, the better counterpuncher will probably win. If two big servers face off, the better big server should have no problem. But if a big server plays a counterpuncher … then, all bets are off.

We’ve seen Rafael Nadal struggle against the likes of John Isner and Dustin Brown, and and we’ve seen big servers neutralized by their opposites, as in Marin Cilic’s 1-6 record against Gilles Simon. There are upsets when similar styles clash, as well, but as untested theories go, this one is appealing and not obviously flawed.

Then, to kick off the 2019 Australian Open, Reilly Opelka knocked out Isner. Playing styles don’t come much more evenly matched, and the veteran was the heavy favorite. It was a perfect example of the kind of match I would expect to follow the script, yet the underdog came out on top. They played four tiebreaks and there were only two breaks of serve, but Opelka didn’t even need the Australian Open’s new fifth-set 10-point tiebreak. While it’s just one match, of course, it suggested that I ought to look more closely at my assumptions.

After a couple of hours playing with data this afternoon, my theory is no longer untested … and it turned out to be flawed. Fortunately, it isn’t just another negative result. Playing style is related to upset likelihood, but not in the way I predicted.

Measuring predictability

Let me explain how I tested the idea, and we’ll work our way to the results. First, I used used Match Charting Project data to calculate aggression score for every ATP player with at least 10 charted matches since 2010. Aggression score is, essentially, the percentage of shots that end the point (by winner, unforced error, or inducing a forced error), as will serve as our proxy for playing style. That gives us a group of 106 players, from the conservative Simon and Yoshihito Nishioka with aggression scores around 13%, to the freewheeling Brown and Ivo Karlovic, with scores nearing 30%. I divided those 106 players into quartiles (by number of matches, not number of players, so each quartile contains between 21 and 31 players) so we could see how each general playing style fares against the others. Here are the groups:

(Aggression score conflates two things: big serving/big hitting and tactical aggression. Isner is sometimes not particularly aggressive, but because of his size and serve skill, he is able to end points so frequently that, statistically, he appears to be extremely aggressive. Accordingly, I’ll refer to “big servers” and “aggressive players” interchangeably, even though in reality, there are plenty of differences between the two groups.)

Limiting our view to these 106 men, I found just over 11,000 matches to evaluate and divided them into groups based on which quartiles the two players fell into. Each of the ten possible subsets of matches, like Q1 vs Q2, or Q4 vs Q4, contains at least 400 examples.

For every match, I used surface-adjusted Elo ratings to determine the likelihood that the favorite would win. That gives us pre-match odds that aren’t quite as accurate as what sportsbooks might offer, though they’re close.

Those pre-match odds are key to determining whether certain groups are more predictable than others. If there are 100 matches in which the favorite is given a 60% chance of winning, and the favorites win 70 of them, we’d say that the results were more predictable than expected. If the favorites win only 50, the results were less predictable.

Goodbye, pet theory

For the matches in each of the ten quartile-vs-quartile subsets, I calculated the average favorite’s chance of winning (“Fave Odds”), then compared that to the frequency with which the favorites went on to win (“Fave Win%”). The table below shows the results, along with the relationship between those two numbers (“Ratio”). A ratio of 1.0 means that matches within the subset are exactly as predictable as expected; higher ratios mean that the favorites were even better bets than the odds gave them credit for, and lower ratios indicate more upsets than expected.

[table id=1 /]

There’s a striking finding here: The largest ratio, marking the most predictable bucket of matches, is for the most conservative pairs of players, while the smallest ratio, pointing to the most frequent upsets, is for the most aggressive players.

Before analyzing the relationship, let’s check one more thing. The very best players aren’t evenly divided throughout the quartiles, since Q1 has two of the big four. Elo-based match predictions–one of the building blocks of these results–are tougher to get right for the best players and the most uneven matchups, so we need to be careful whenever the elites might be influencing our findings. Therefore, let’s look at the same numbers, but this time for only those matches in which the favorite has a 50% to 70% chance of winning. This way, we exclude many of the best players’ matchups and all of their more lopsided contests:

[table id=2 /]

We discard about 40% of our sample, but the predictability trend remains the generally the same. In both the overall sample and the narrower 50%- to 70%-favorite subset, the strongest relationship I could find was between the predictability ratio and the quartile of the less aggressive player. In other words, a counterpuncher is likely to have more predictable results–regardless of whether he faces a big server, a fellow counterpuncher, or anyone in between–than a more aggressive player.

Back to basics

My initial theory is clearly wrong. I expected to find that Q1 vs Q1 matches were more predictable than average, and I was right. But by my logic, I also guessed that Q4 vs Q4 matches went according to script, and that other pairings, like Q1 vs Q4, would be more upset-prone. I would have done better had I let the neighbor’s cat make my predictions for me.

Instead, we find that that matches with more aggressive players are more likely to result in surprises. That doesn’t sound so groundbreaking, and it’s something I should’ve seen coming. Big servers tend to hold serve more often and break serve less frequently, meaning that their matches end with narrower margins, opening the door for luck to play a larger role, especially when sets and matches are determined by tiebreaks.

After all this, you might be thinking that I’ve squandered my afternoon, plus another few minutes of your attention, arriving at something obvious and unremarkable. I agree that it’s not that exciting to proclaim that big servers are more influenced by luck. But there’s still a useful–even surprising–discovery buried here.

Exponential upset potential

We know that the most one-dimensional players are more subject than others to the ups and downs of luck, thanks to the narrow margins of tiebreaks. For a man who rarely breaks serve, no match is a guaranteed win; for a man who rarely gets broken, no opponent is impossible to beat. However, I would have expected that the unpredictability of big servers was already incorporated into our match predictions, via the Elo ratings of the big servers. If a player has unusually random results, we’d expect his rating to drift toward tour average. That’s one reason that it’s very difficult for poor returners to reach the very top of the rankings.

But apparently, that isn’t quite right. The randomness-driven Elo ratings of our big servers do a nearly perfect job of predicting match outcomes against counterpunchers, and they’re only a little bit too confident against the more middle-of-the-road players in Q2 and Q3. Against each other, though, upsets run rampant. That extremely volatile fraction of results–the tiebreak-packed outcomes when the biggest servers face off–only accounts for part of these players’ ratings.

We’re accustomed to getting unpredictable results from the most aggressive players, with their big serves, inconsistent returns, and short rallies. Today’s findings give us a better idea of when these do and do not occur. Against counterpunchers, things aren’t so unpredictable after all. But when big servers play each other, we expect the unexpected–and the results are even more unpredictable than that.

Daniil Medvedev’s Leading Elo Indicator

Italian translation at settesei.it

It is shaping up to be a breakthrough season for 22-year-old Russian Daniil Medvedev. His Tokyo title two weeks ago was his first at the ATP 500 level and his third on the season, after earlier triumphs in Sydney and Winston-Salem. The run in Japan was a particularly notable step, since he knocked out three top-20 players along the way. He had only four top-20 victories in the entire season leading up to Tokyo, and two of those were against the slumping Jack Sock.

His ATP ranking is rising alongside his results. The Winston-Salem title moved him into the top 40, and the Tokyo trophy resulted in a leap to No. 22. After a first-round win in Shanghai last week, Medvedev crept to his current career-high of No. 21. With a couple of wins in Moscow this week, he could overtake Milos Raonic and reach the top 20.

The improvement on the ATP ranking table is nothing next to the Russian’s race to the top of the Elo list. Last Monday, with the Japanese title in the books, Medevdev rose to No. 8 on my men’s Elo ranking. Since then, he has dropped two places but remains in the top ten, ahead of Marin Cilic, Kevin Anderson, and a host of others who outrank him on the official ATP list.

Given the discrepancy, what do we believe? Is Medvedev inside the top 10 or outside the top 20? Is Elo a leading indicator–that is to say, an early-warning signal for future ATP ranking milestones–or a misleading one? Elo is designed to be forward-looking, tuned to forecast upcoming match outcomes and weighting wins and losses based on the quality of the opponent. The official rankings explicitly consider a year’s worth of results, with no adjustments for quality of competition. In theory, Elo should be the better of the two measures for predicting longer-term results, but that assumes the algorithm works well, and that it doesn’t overreact to short-term successes. Let’s take a look at past differences between the two systems and see what the future might hold for the 22-year-old.

Precedents

Since 1988, 102 men have debuted in the ATP top ten. A slightly larger number, 113, have shown up in the top ten of my Elo ratings. There’s a very substantial overlap between the two, with 94 names appearing in both categories. Thus, 8 players have reached the ATP top ten without clearing the Elo threshold, while 19 have rated a spot in the Elo top ten without convincing the ATP computer to agree.

Here are the eight ATP top-tenners whose Elos have never merited the same status:

Player               ATP Top Ten Debut  ATP Top Ten Weeks  
Jonas Svensson                19910325                  5  
Nicolas Massu                 20040913                  2  
Radek Stepanek                20060710                 12  
Jurgen Melzer                 20110131                 14  
Juan Monaco                   20120723                  8  
Kevin Anderson                20151012                 31  
Pablo Carreno Busta           20170911                 17  
Lucas Pouille                 20180319                  1

A few of these players could still make progress on the Elo list, especially Kevin Anderson, who is currently 11th, a miniscule five points behind Medvedev.

Here is the longer list of Elo top-ten players without any weeks in the official top ten:

Player                 Elo Top Ten Debut  Elo Top Ten Weeks  
Carl Uwe Steeb                1989/05/22                  3  
Andrei Cherkasov              1990/12/11                  1  
Goran Prpic                   1991/05/20                  1  
David Wheaton                 1991/07/08                  9  
Jerome Golmard                1999/05/03                  2  
Dominik Hrbaty                2001/01/15                  2  
Jan Michael Gambill           2001/04/06                  6  
Nicolas Escude                2002/02/25                  4  
Younes El Aynaoui             2002/05/20                  2  
Paul Henri Mathieu            2002/10/14                  8  

Player                 Elo Top Ten Debut  Elo Top Ten Weeks
Agustin Calleri               2003/05/19                  2  
Taylor Dent                   2003/10/06                 10  
Andrei Pavel                  2004/05/10                  2  
Robby Ginepri                 2005/10/24                  1  
Ivo Karlovic                  2007/11/12                  3  
Roberto Bautista Agut         2016/02/22                  1  
Nick Kyrgios                  2016/03/04                 62  
Stefanos Tsitsipas            2018/08/13                  3  
Daniil Medvedev               2018/10/08                  2

* I define ‘weeks’ a little differently for Elo ratings, as ratings are generated only for those weeks with an ATP-level tournament or Davis Cup tie.

Most of these guys came very close to cracking the ATP top ten. For example, David Wheaton’s peak ranking was No. 12. With the exception of Nick Kyrgios, no one spent more than ten weeks in the Elo top ten without eventually reaching the same standard according to the ATP formula. This list shows that it’s possible to have a brief peak that cracks the Elo top ten but doesn’t last long enough to reflect the kind of success that the official ranking system was designed to reward. About one in six players with a top-ten Elo rating never reached the ATP top ten, though as we can see, the odds of remaining an Elo-only star fall quickly with each additional week in the top ten.

Kyrgios is a perfect example of the differences between the two approaches to player ranking. The Australian has recorded a number of high-profile upsets, which are the fastest way to climb the Elo list. But knocking out the second-ranked player in the world, as Kyrgios did to Novak Djokovic at Indian Wells last year, doesn’t have much impact on the ATP ranking when it happens in the fourth round. Usually, a player who can oust the elites will start piling up wins in a form that the official computer will appreciate. But Kyrgios, unlike just about every player in history with his talent, hasn’t done that.

In short, Elo will always elevate a few players to top-ten status even if they’ll never deserve the same treatment from the ATP formula. It’s too early to say whether Medvedev fits that mold. But where Elo really excels is identifying top players before the ATP computer does. Of the 94 cases since 1988 in which a man debuted in both top tens, Elo was first to anoint the player a top-tenner in 76 of them–better than 80%. The official rankings were first 10 times, and the two systems tied in the other eight instances. On average, players reached the Elo top ten about 32 weeks before the ATP top ten.

Here are the 11 most extreme gaps in which Elo got there first, along with the top-ten debuts of the Big Four:

Player               ATP Debut   Elo Debut  Week Diff  
Mariano Puerta      2005/07/25  2000/06/12        267  
Marc Rosset         1995/07/10  1990/11/05        244  
Fernando Gonzalez   2006/04/24  2002/10/07        185  
Guillermo Canas     2005/05/09  2002/08/05        144  
Mikhail Youzhny     2007/08/13  2004/11/15        143  
Gaston Gaudio       2004/06/07  2002/04/29        110  
Richard Gasquet     2007/07/09  2005/06/20        107  
Tomas Berdych       2006/10/23  2004/10/11        106  
Robin Soderling     2009/10/19  2007/10/08        106  
Mark Philippoussis  1999/03/29  1997/03/24        105  
Jack Sock           2017/11/06  2016/01/18         94  
                                                       
Player               ATP Debut   Elo Debut  Week Diff  
Roger Federer       2002/05/20  2001/02/19         65  
Andy Murray         2007/04/16  2006/08/21         34  
Novak Djokovic      2007/03/19  2006/07/31         33  
Rafael Nadal        2005/04/25  2005/02/21          9

And in case you’re curious, the ten cases in which the ATP computer beat Elo to the punch:

Player              ATP Debut   Elo Debut  Week Diff  
Stan Wawrinka      2008/05/12  2010/10/25        128  
David Ferrer       2006/01/30  2007/05/28         69  
Janko Tipsarevic   2011/11/14  2012/05/13         26  
Rainer Schuettler  2003/06/09  2003/08/25         11  
Tommy Robredo      2006/05/08  2006/07/24         11  
Fernando Verdasco  2009/02/02  2009/04/06          9  
Albert Costa       1997/04/21  1997/05/26          5  
Nicolas Almagro    2011/04/25  2011/05/22          4  
John Isner         2012/03/19  2012/04/15          4  
Jiri Novak         2002/10/14  2002/10/21          1

The 32-week average difference is suggestive. As I’ve noted, Elo ratings are optimized to forecast the near future, so at least in theory, they reflect each player’s level right now. The ATP algorithm tallies each man’s performance over 52 weeks, with equal weight given to the first and last weeks in that timeframe. Setting aside improvement and decline due to age, that means the ATP computer is telling us how each player was performing, on average, 26 weeks ago. If Medvedev continues to oust top-20 players on a regular basis and claims another 500-level title or two, he could well be 26 or 32 weeks away from a top-ten debut.

Elo isn’t designed to make long-term forecasts–the tools needed to do so, for the most part, have yet to be invented. And the system occasionally gives high ratings to players who don’t sustain them for very long. But in general, a superlative Elo rating is a sign that a similar mark on the ATP ranking list isn’t far behind. So far, Kyrgios has managed to defy the odds, but the smart money still points to an eventual ATP top-ten debut for Medvedev.

The Rosy Forecast of Arnya Sabalenka’s Elo Rating

Italian translation at settesei.it

It’s been almost two weeks since Aryna Sabalenka’s last title, and the next one is starting to feel overdue. With respect to Naomi Osaka’s ascent, the Belarussian is the hottest rising star on the women’s tour right now, with two titles in the last two months, plus two more finals earlier in the season. The 20-year-old is 8-4 against the top ten this year, with wins over Caroline Wozniacki, Petra Kvitova, Elina Svitolina, and Karolina Pliskova.

It takes time for all of these wins to show up in the WTA rankings. Sabalenka nudged into the top 20 after winning New Haven in August, and rose as high as 11th last Monday, though she is set to fall back to 14th after failing to defend her title in Tianjin this week. While the official ranking is a lagging indicator, Elo ratings react more quickly, especially to high-profile upsets like the ones Sabalenka has been recording almost every week.

Sabalenka’s Elo rating has rocketed to the top of the list. Through last week’s matches, she sits at second overall, behind Simona Halep, but closer to Halep than to third-place Wozniacki. After knocking out Caroline Garcia in Beijing last week, she briefly took over the Elo top spot before handing it back after her quarter-final loss to Qiang Wang. Still, an overall ranking of #2 is a lot more suggestive of future stardom than the WTA computer’s report of #11.

When Elo looks at hard court matches alone, it is even more optimistic, putting Sabalenka at the very top of the list. Elo would narrowly favor the Belarussian in a hard-court match against Halep and, assuming the draw treated both players equally, would make Sabalenka the early favorite for the 2019 Australian Open title.

What should we make of this? Is it time to appoint Sabalenka the next superstar, or ought we treat Elo ratings with more circumspection? Let’s take a look at players who have topped the Elo list in the past to get a better idea.

Since 1984, only 29 women (including Sabalenka) have reached the #1 or #2 spot on the overall Elo list. 19 of them got to #1 in the official rankings. Here are the other ten:

Player               Peak  
Petra Kvitova           2  
Conchita Martinez       2  
Jana Novotna            2  
Agnieszka Radwanska     2  
Elina Svitolina         3  
Gabriela Sabatini       3  
Elena Dementieva        3  
Samantha Stosur         4  
Johanna Konta           4  
Aryna Sabalenka        11

This is pretty good company. Svitolina could still reach #1, and several of the others were expected to attain even greater heights than they did. The only warning sign here is Johanna Konta, who isn’t the best comp for a young star, as she didn’t crack the top two until close to her 26th birthday.

The group of women who have ranked #1 on the hard-court specific Elo ranking table is even more select. Sabalenka is only the 17th player since 1984 to head the list, and 14 of the 17 have topped the official rankings as well. The only other exceptions are Svitolina and Konta.

If there’s ever a good time to anoint a 14th-ranked player the future of the sport, I’d say this is it. Elo isn’t perfect, and it’s possible that the algorithm has overreacted to a series of upsets in a season packed full of them. But if the system has made a mistake, it’s one that it doesn’t make very often. Sabalenka has only won four main-draw matches at majors, so maybe that 2019 Australian Open title is too much to ask. But in the long term, one grand slam title might be a mere harbinger of even greater things to come.

Forecasting the 2018 Laver Cup

Embed from Getty Images

Italian translation at settesei.it

It’s that time of year again: group selfies in suits, dodgy Davis Cup excuses, and a reminder that it takes more than six continents just to equal Europe. That’s right, it’s Laver Cup.

Last year, I worked out a forecast of the event, walking through a variety of ways in which captains Bjorn Borg and John McEnroe could use their rosters and ultimately predicting a 16-8 win for Team Europe. As it happened, both captains intelligently deployed their stars, and the result was 15-9. This year, the competitors are a little different and the home court has moved from Prague to Chicago, but the format remains the same.

Let’s start with a look at the rosters. I’ve included two additional players for reference: Juan Martin del Potro, scheduled to play for Team World, but withdrew; and Pierre Hugues Herbert, the doubles specialist Borg hasn’t realized he needs. Each player is shown alongside his surface-weighted singles Elo rating and surface-weighted doubles “D-Lo” rating:

EUROPE                       Singles Elo  Doubles D-Lo  
Novak Djokovic                      2137          1667  
Roger Federer*                      2097          1700  
Alexander Zverev                    1971          1690  
David Goffin                        1960          1582  
Grigor Dimitrov                     1928          1719  
Kyle Edmund                         1780          1542  
                                                        
WORLD                        Singles Elo  Doubles D-Lo  
Kevin Anderson                      1914          1692  
Nick Kyrgios                        1910          1668  
John Isner                          1887          1800  
Diego Sebastian Schwartzman         1814          1540  
Frances Tiafoe                      1772          1544  
Jack Sock                           1724          1925  
                                                        
ALSO                                                    
Juan Martin Del Potro               2062          1678  
Pierre Hugues Herbert               1691          1890

* Federer has played very little tour-level doubles for a long time. Last year I estimated his D-Lo at 1650; he played rather well last year, so I’ll bump him up to 1700 this time around.

Especially with Delpo on the sidelines, Europe looks to dominate the singles. The doubles leans in World’s favor, largely because Jack Sock is so good, especially in comparison with guys who have focused on singles.

Format review

Let’s do a quick refresher on the format. Laver Cup takes place over three days, each of which has three singles matches and one doubles match. Each player must play singles at least once, and no doubles pairing can repeat itself. Day 1 matches are worth one point each, day 2 matches are worth 2 points each, and day 3 matches are worth 3 points each. If there’s a 12-12 tie at the end of day 3, a single doubles set–in which a previously-used team may compete–will decide it all.

Given that format, the best way for the captains to use their rosters is to stick their three worst singles players on day 1 duty, then use their best three on both day 2 and day 3. For doubles, they should use their best doubles player every day, with the best partner on day 3, next best on day 2, and third best on day 1. As I’ve mentioned, Borg and McEnroe came close to this last year, although Borg didn’t use Rafael Nadal (his best doubles player) in day 3 doubles, and he generally overused Tomas Berdych. Both decisions are understandable, as Nadal may not have been physically able to play every possible match, and Berdych was in front of a Czech crowd.

Now that we know the captains will act in a reasonably savvy way, we can forecast the second edition with a little more confidence than the inaugural one.

The forecast

Nadal’s absence this year will hurt the Europe squad on both singles and doubles. Combined with a small step backward for Federer’s singles game, this year’s Laver Cup figures to be closer than last year. Recall that my forecast a year ago called for a 16-8 Europe victory, and the result was 15-9.

Assuming optimal usage, the 2018 forecast gives Europe a 67.6% chance of winning, with a most likely final score of 14-10. There’s a nearly one-in-ten shot that we’ll see a 12-12 tie, in which the superior doubles capabilities of Team World give them the edge, with a 70.7% probability of winning the tie-breaking set.

Were del Potro not so fragile, this could get even more interesting. Swap out Frances Tiafoe for the Tower of Tandil, and Europe’s chances fall to 56.8%, with a most likely final score of 13-11.

Nothing McEnroe could have done, short of going to medical school a few decades ago, could have put the Argentine on his team this week. But Borg has less of an excuse for failing to maximize the potential of his team. Unlike World, with its world-beating doubles specialist, Europe has a stunning singles roster that rarely takes to the doubles court. As we’ve seen, one doubles player can take the court three times, plus the potential 12-12 tie-breaking set. The specialist would need to play singles only once, on the low-leverage first day.

The obvious choice is Pierre Hugues Herbert, a top-five doubles player with the ability to play respectable singles as well. The Frenchman would be considerably more valuable than Kyle Edmund, who is a better singles player, but not good enough to be of much help to an already loaded side. (I made a similar point last year and illustrated it with Herbert’s partner, Nicolas Mahut. Since then, Herbert has taken the lead over his Mahut in both singles and doubles Elo ratings.)

When we sub in Herbert for Edmund, the simulation spits out the best result yet for Europe. Against the actual World team (that is, no Delpo), the hypothetical Europe squad would have a 74.6% chance of winning, with the likely final score between 14-10 and 15-9. Herbert and a mediocre partner would still be the underdogs in a tie-breaking final set against Sock and John Isner, but the presence of a legitimate doubles threat would narrow the odds to about 58/42.

We won’t get to see either Delpo or Herbert in Chicago this year, but we can expect a slightly more competitive Laver Cup than last year. Add in home court advantage, and the result is even less of a foregone conclusion. It’s no match for last week’s Davis Cup World Group play-offs, but I suspect it’ll make for more compelling viewing this weekend than the final rounds in Metz and St. Petersburg.

Two Servebots and Zero Tiebreaks

Embed from Getty Images

Isner had energy to burn since he never needed to count to seven.

Italian translation at settesei.it

There have been plenty of upsets at this year’s US Open, but they all pale in comparison with the surprise that John Isner and Milos Raonic delivered Sunday night in their fourth round match. Isner won, 3-6 6-3 6-4 3-6 6-2, failing to hold twice and breaking Raonic’s serve four times. Rarely has a tiebreak seemed so assured, and the two big men didn’t even get close.

In five previous meetings, Isner and Raonic have been more likely to deliver two tiebreaks than only one, and most of their matches were best-of-three, not the grand slam best-of-five format. In 13 previous sets, they had played 9 tiebreaks. In the last year, 45% of Isner’s sets have reached 6-6, while nearly a quarter of the Canadian’s have. One or the other of these guys is responsible for the longest match in history, the longest ever major semi-final, and the longest match in Olympics history. They are really, really good at holding serve, and really not-so-good at breaking.

Great expectations

The likelihood that Isner and Raonic would play a tiebreak depends on some basic assumptions. If Raonic served like he has for the last 52 weeks, that’s a service-point won percentage (SPW) of 72.8%, which is equivalent to holding 93% of the time. If we use Isner’s actual SPW from the match of 74.3%, that translates to a hold rate of 94.4%. If we choose Isner’s SPW from his previous meetings with Raonic of a whopping 76.5%, that gives us an implied hold rate of 96%. Those all sound high but, as we’ll see, the difference between them ends up affecting the probability quite a bit.

I’m going to run the numbers using three sets of assumptions:

  1. The head-to-head. In five matches (four of them on hard courts, the fifth at Wimbledon this year), Isner won 76.5% of service points, while Raonic won 71.4%. That’s equivalent to hold rates of 96.0% and 91.7%, respectively.
  2. The last 52 weeks (adjusted). Across all surfaces, going back to last year’s US Open, Isner has won 73.6% of service points, against Raonic’s 72.8%. Those numbers, however, are against average opponents. Both players, and especially Isner, have below-par return games. If we adjust each SPWs for the other player’s rate of return points won (RPW), we get 75.5% for Isner and 78.5% for Raonic. In game-level terms, those are hold rates of 95.3% and 97.1%.
  3. The match itself. On Sunday night, Isner won 74.3% of service points and Raonic won 68.8%. Using these numbers doesn’t give us a true prediction, since we couldn’t have known them ahead of time. But maybe, if we used every scrap of information available to us and put them all together in a really smart way, we could have gotten close to the true number. Those rates translate to hold percentages of 94.4% for Isner and 88.5% for Raonic.

Not enough tiebreaks

Apparently, the betting odds for at least one tiebreak in the match set the probability around 95%. That turns out to be in line with my predictions, though the specific assumptions affect the result quite a bit.

I’ve calculated a few likelihoods using each set of assumptions. The first, “p(No brk),” is the probability that the two men would simply hold serve for 12 games. It’s not the only way to reach a tiebreak, but it accounts for most of the possibilities. Next, “p(TB)” is the result of a Monte Carlo simulation to show the odds that any given set would result in a tiebreak. “eTB” is the expected number of tiebreaks if we knew that Isner and Raonic would play five sets. Finally, “p(1+ TB)” is the chance that the match would have at least one tiebreak in five sets.

Model   JI Hld  MR Hld  p(No brk)   p(TB)   eTB  p(1+ TB)  
H2H      96.0%   91.7%      46.5%   51.3%   2.6     97.3%  
Last52   95.3%   97.1%      62.8%   65.3%   3.3     99.5%  
Match    94.4%   88.5%      34.0%   41.2%   2.1     93.0%

Given how the big men played on Sunday, it isn’t unthinkable that they never got to 6-6. In large part because Isner’s return game brought Raonic’s SPW under 70%, each set had “only” a 41.2% chance of going to a tiebreak, and there was a 7% chance that a five-setter would have none. The other two sets of assumptions, though, point to the sort of tiebreak certainty reflected in the betting market … and just about anyone who has ever seen these two guys play tennis.

Perhaps the strangest aspect of all of this is that, in six previous matches at this year’s Open, Isner and Raonic combined for seven tiebreaks–at least one in five of their six matches–before their anticlimactic encounter. Knowing Isner, this is a blip, not a trend, and he’s sure to give us a breaker or two in his quarter-final against Juan Martin del Potro. His tournament record will likely show one or two tiebreaks in every match … except for the one against his fellow servebot. This must be why we stick with tennis: Every match has the potential to surprise us, even if we never really wanted to watch it.

Did Rafael Nadal Almost Lose a Set to David Ferrer?

Italian translation at settesei.it

In David Ferrer’s final grand slam, the draw gods handed him a doozy of a first-round assignment in Rafael Nadal. Ferrer has struggled all year, and no one seriously expected him to improve on his 6-24 career record against the King of Clay. In the end, he didn’t: Ferrer was forced to retire midway through the second set with a calf injury. But before his final Flushing exit, he gave Rafa a bit of a scare.

Nadal won the first set, 6-3. The second set was a bit messier: Ferrer broke to love in the opening game, Rafa broke him back in the next, and a bit later, Ferrer broke again to take a 3-2 lead. He maintained that one break advantage until he physically couldn’t continue. Leading 4-3 and serving the next game, he was been two holds away from leveling the match.

Does that mean Nadal “almost” lost the set? People on the internet argue about these things, and while I don’t understand why, I do love a good probability question. If it overlaps with semantics (yay sematics!), that’s a bonus.

Let’s forget the word choice for now and reframe the question: Ignoring the injury, what were Ferrer’s chances of winning the set? If we assume that both players were equal, it’s a simple thing to plug into my win probability model and–ta da!–we find that from 4*-3, Ferrer had a roughly 85% chance of winning the set.

But wait: I can already hear the Rafa fans screaming at me, these two players aren’t exactly equal. In the 102 points the Spanish duo played on Monday night, Ferrer won 38% on return and Nadal won 47%. For an entire five-set match, those rates work out to a 93% chance of Rafa winning. Maybe that’s not quite high enough, but it’s in the ballpark. Using those figures, Ferrer’s chance of hanging on to win the second set drop significantly, to 57.5%. When you’re winning barely half of your service points, your odds of securing a pair of holds are worse than a coin flip. Had Ferrer won the set, it’s more likely that he would’ve needed to either break Rafa again or come through in a tiebreak.

That’s a pretty big difference between our two initial estimates. 85% sounds good enough to qualify for “almost” (though one study quantifies the meaning of “almost” at above 90%), but 57.5% does not.

That doesn’t quite settle it, though. The win probability model takes all notions of streakiness out of the equation.  According to the formula, there’s no patches of good or bad play, no dips in motivation, so extra energy to finish off a set, etc. I’m not convinced any of those exist in any systematic manner, but it’s tough to settle the question either way. Therefore, if we have the ability to use data from real-life matches, we should.

And here, we can. Let’s start with Nadal. Going back to late 2011, I was able to identify 69 sets in which Rafa was returning down a break at 4-3. (There are probably more; my point-by-point dataset isn’t exhaustive, but the missing matches are mostly random, so the 69 should be representative of the last several years.) Of those 69, he came back to win 21, almost exactly 30%.

Ferrer has been more solid than Nadal’s opponents. (It helps that Ferrer only had to face Rafa once, while Nadal’s opponents had face him every time.) I found 122 sets in which Ferrer served at 4-3, leading by a break. He went on to win the set 109 of those times, or about 89%.

The 89% figure is definitely too high for our purposes: Not only was Ferrer a better player, on average, between 2012 and today, than he is now, but he also had the benefit of facing weaker opponents than Nadal in almost all of those 122 sets. 89%–not far from the theoretical 85% we started with–is a grossly optimistic upper limit.

Even if we take the average of Nadal’s and Ferrer’s real-life results–roughly 90% conversions for Ferru and 70% for Rafa’s opponents–80% is still overshooting the mark. As we’ve established, Ferrer’s numbers refer to a stronger version of the Spaniard, while Rafa is still near the level of his last half-decade. Even 80%, then, is overstating the chances that Nadal would’ve lost a set.

That leaves us with a range between 57%, which assumes Nadal would keep winning nearly half of Ferrer’s service points, and 80%, which is based on the experience of both players over the last several years. Ultimately, any final figure comes down to what we think about Ferrer’s level right now–not as good as it was even a couple of years ago, but at the same time, good enough to come within two games of taking a set from the top-ranked player in the world.

It would take a lot more work to come up with a more precise estimate, and even then, we’d still be stuck not only trying to establish Ferrer’s current ability level, but also his ability level in that set. Just as the word “almost” refers to a range of probabilities, I’m happy to call it a day with my own range. Taking all of these calculations together, we might settle on a narrower field of, say, 65-70%, or about two in three. There’s a good chance a healthy Ferrer would have taken that set from his long-time tormentor, but it was far from a sure thing … or even, given the usual meaning of the word, an “almost” sure thing.

Unseeded Serena and the Roland Garros Draw

In a wide-open women’s field at this year’s French Open, it seems fitting that one of the most dangerous players in the draw isn’t even seeded. Serena Williams has played only four matches–none of them on clay–since returning to tour after giving birth. As such, her official WTA ranking is No. 453, and her current match-play level is anyone’s guess.

Because her ranking is low, she needed to use the ‘special ranking’ rule to enter the tournament, and the rule doesn’t apply to seedings. (I’m not going to dive further into the debate about how the rule should work–I’ve written a lot about the rule in the past.) As an unseeded player, she could have drawn anyone in the first round; in that sense, she was a bit lucky to end up opposite another unseeded player, Kristyna Pliskova, in the first round. Her wider draw section is manageable as well, with a likely second-round match against 17th seed Ashleigh Barty and a possible third-rounder with 11th seed Julia Goerges. If she makes it to the round of 16, we’ll probably be treated to a big-hitting contest between Serena and Karolina Pliskova or Maria Sharapova.

According to my Elo-based forecast, a best guess about the level of post-pregnancy Serena is that she’s the 7th best overall player in the field, and 9th best on clay. That gives her about a 40% chance of winning her first three matches and reaching the second week, a 6.2% chance of making it to the final, and a 3.1% chance of adding yet another major title to her haul.

What if she were seeded? Seeds are a clear advantage for players who receive them, as a seeding protects against facing other top contenders until later rounds. By simulating the tournament with Serena seeded, we can get a sense of how much the WTA’s rule (and the French Federation’s decision not to seed her) impacts her chances.

Seeded 7th: Let’s imagine a bizarre world in which my Elo ratings were used for tournament seedings. In that case, Serena would be seeded 7th, knocking Caroline Garcia down to 8th and sending current 32nd seed Alize Cornet into the unseeded pool. That would be a clear advantage: 50/50 odds of reaching the fourth round, a 9% chance of playing in the final, and a 4.4% shot at the title, compared to 3.1% in reality.

Seeded 1st: If seeds were assigned based on protected ranking, Serena would be the top seed. You can’t get much more of an advantage than that: The top seed is protected from playing either of the other top-four seeds until the semifinals, for instance. (It’s no insurance against a meeting with 28th seed Sharapova, but Serena, of all people, isn’t worried about that.) Moving from 7th to 1st would give her another boost, but it’s a modest one: As the top seed, her chances of sticking around for the second week would still be 50/50, with 10.1% and 4.7% odds of reaching the final and winning the title, respectively.

Here’s a summary of Serena’s chances in the various seeding scenarios. The final column is “expected points”–a weighted average of the number of WTA ranking points she is expected to collect, given her likelihood of reaching each round.

Scenario     R16  Final  Title  ExpPts  
Actual     39.8%   6.2%   3.1%     273  
Unseeded*  34.4%   6.2%   3.0%     259  
Seeded 7   50.3%   9.0%   4.4%     356  
Seeded 1   50.5%  10.1%   4.7%     371

* the ‘unseeded’ scenario represents Serena’s chances as an unseeded entrant, given a random draw. She got a little lucky, avoiding top players until the 4th round, though her chances of making the final end up the same.

Seeds matter, though there’s only so much they can do. If Serena really is at a barely-top-ten level, she’s a long shot for the title regardless of whether there’s a number next to her name. If my model grossly underestimates her and she’s back at previous form–let’s not forget, she made the final the last time she played here, and won the title the year before that–then the rest of the field will once again look like a bunch of flies for her to swat away, regardless of which numbers they have next to their names.