Category Archives: Win probability

The 5 Biggest Comebacks of the 2013 ATP Season

Everybody loves a big comeback, but some of the best come-from-behind wins on the ATP tour this year were such unheralded matchups that they’ve already fallen out of the spotlight.  While everyone else ranks NadalDjokovic matches in their year-end lists, let’s look at the five matches in which the winner had to climb out of the biggest hole.

To do this, I ranked every match this season by Comeback Factor (CF), a stat that identifies the lowest ebb in the match for the eventual winner.  If a player breaks serve to open the match and sails to victory, his chance of winning never falls below 50%.  But if he goes down a set and a break, his odds fall much lower.  If the latter player comes back to win, his CF is much higher.

1. Indian Wells Masters R64: Gilles Simon d. Paolo Lorenzi 6-3 3-6 7-5 (win probability graph)

Lorenzi went up a double break in the final set by winning the first four games on the trot.  Simon held twice to force the Italian to serve for it at 5-2.  Lorenzi went up 40-15 in that service game, earning two match points, before losing four points in a row and dropping serve.  At 5-4, Simon broke him to 15, then broke again to love to seal the final set, 7-5.

At 5-2 40-15 in the 3rd set, Lorenzi’s chance of winning was about 99.8%, the highest recorded in a match this year by a player who didn’t end up winning.

2. Queen’s Club R64: Ivan Dodig d. James Ward 6-7(8) 7-6(2) 7-6(2) (win probability graph)

Dodig fought back from nearly the same hole that Simon found himself in, but did so in the second set instead of the third.  Ward won the first set in a tight tiebreak, then earned an early break in the second.  He held on until he served at 5-3, when he reached 40-15.  Dodig won the next four points to erase the break, improving his probability of winning from 0.5% to 21.1%.

Amazingly, the scenario repeated itself in the third set after Dodig won the second in a tiebreak.  Ward went up a break and served for the match again at 5-4, but failed to generate another match point.  The Croatian won a pair of points from 30-30 in that game, then sealed the match in yet another tiebreak.

Dodig wasn’t so lucky a couple of months later, when he nearly upset Juan Martin del Potro in Montreal.  In this year’s 7th-biggest comeback, Delpo came back from a double-break hole in the third set to deny Dodig a place in the third round.

3. Madrid Masters R64: Mikhail Youzhny d. Fabio Fognini 7-6(4) 2-6 7-6(5) (win probability graph)

Fognini never had the double break that led to such disaster for Lorenzi and Ward, but he did have something neither of those men did: a triple match point.  At 3-3 in the deciding set, Fognini broke the Russian then consolidated, leading to a chance to serve for the match at 5-4.  After winning his first three points for a 40-0 advantage, his win probability climbed as high as 99.1%.

It wouldn’t go any higher.  Youzhny won 12 of the next 13 points, breaking the Italian, holding his own serve to love, then earning two match points of his own on the Fognini serve before Fabio gathered himself sufficiently to force a tiebreak.  Fognini kept up his streakiness to the end, claiming a minibreak to open the tiebreak, dropping five points in a row, and fighting back to 5-5 before finally losing the match.

4. Roland Garros R32: Tommy Robredo d. Gael Monfils 2-6 6-7(5) 6-2 7-6(3) 6-2 (win probability graph)

Monfils won the first two sets, which you would think put Robredo at enough of a disadvantage.  But the Spaniard’s lowest ebb didn’t come until the fourth set.  He lost serve in the seventh game, and after fighting off a match point at 3-5, he needed to break serve just to stay alive.

The Frenchman went up 40-15, earning two more match points and a win probability of 98.9%.  Robredo won four straight points to get back on serve, easily held, and even challenged Monfils’s own serve (to 0-30) before landing in a tiebreak.  He won that breaker and, compared to the fourth set, won the fifth with ease.

5. Australian Open QF: David Ferrer d. Nicolas Almagro 4-6 4-6 7-5 7-6(4) 6-2 (win probability graph)

After Robredo beat Monfils, he faced Almagro in the 4th round and Ferrer in the quarters.  Conicidentally, those are the two men who, at the Australian Open, gave 2013 its fifth-biggest comeback.

As in Robredo did in his comeback, Ferrer dropped the first two sets.  Unlike his countryman, he found himself in the most danger in the third set.  Almagro broke in the seventh game of the third set and reached 5-4, an opportunity to serve for the match.  But here, history (or something) got in the way. Almagro reached his highest chance of winning, 98.7%, at 15-0, before Ferrer fought his way to 15-40, Almagro got back to deuce, but Ferrer won the game.

Almagro earned more chances to serve for the match, but his odds of winning would never again be so high.  After breaking in yet another seventh game, Nico served for it at 5-4 and again at 6-5.  At 6-5, he reached 15-0 and a win probability of 97.4%, but from that point on, it was all Ferrer.


Filed under Win probability

Raonic, del Potro, and the Importance of One Point

In last night’s Coupe Rogers match between Milos Raonic and Juan Martin del Potro, one point stands out from the rest.

Raonic won the first set, then Delpo broke early in the second.  With del Potro serving at 4-3, Raonic earned a break point with a winner at the net.  Replays clearly show that he touched the net.  Had the chair umpire seen it in real time, Delpo would have been awarded the point.

The Argentine never recovered, losing the next nine points and the match.

The net touch, and the point Milos didn’t deserve, was clearly a turning point in the match.  But how important was it, really?

If we assume that the two men were equal and that both players win 75% of service points (not true in Delpo’s case yesterday, but reasonable for two big servers on hard courts), here is a summary of Raonic’s probability of winning at various stages of the match:

  • After winning the first set: 75.0%
  • With Delpo serving 4-3, 00-00: 52.4%
  • With Delpo serving 4-3, 40-40: 53.9%
  • After winning the “touch” point: 58.9%
  • If Delpo had won that point: 51.8%
  • After winning the “touch” game: 75.0%
  • After holding serve for 5-4: 76.3%

The controversial point was, clearly, very important.  The difference between winning it and losing it was 7%, a magnitude that doesn’t happen very often in a tennis match, especially outside of tiebreaks.

But the real story here is the next point.  Remember that under normal circumstances, del Potro is a huge server and Raonic does not have a strong return of serve.  (I say “normal circumstances” because somehow, Raonic won 50% of return points in this match.)

If a server is winning 75% of points on his own racquet, his probability of winning a game from break point down is still 67.5%.  There’s a 25% chance he’ll lose the game on the next point, of course, but a 75% chance he’ll get back to deuce, where his serve gives him a 90% chance of winning the game.

The touch point increased Raonic’s chances of winning from 53.9% to 58.9%.  The next point upped his odds from 58.9% to 75.0%.  Which one do you think was more important?

Another way of looking at this to consider what would’ve happened had there been no video replay, and no chance of del Potro spotting the touch and arguing with the umpire about it.  Normal Delpo would’ve stepped back to the line and hit a service winner.  Five minutes later he would’ve held serve again and the two men would’ve played a third set.

It’s easy to look back at this match and conclude that the net touch was the difference in the match.  But no: It was the reaction to the touch–the controversy itself–that had a much greater impact.


Filed under Win probability