Big-point players in tennis: NOT a myth

There really are big-point players in tennis. Just found a couple of statistical references to support this claim.

Watching this year’s Wimbledon, Rafael Nadal always looked in charge of his semi final against Andy Murray. Yet, there was a time late in the third set, with Murray down 0-2 on his way to a 0-3 whipping, when Murray had actually won more points than Nadal. Rafa was winning the points that mattered.

Similar claims in other sports have turned out to be false. For instance, baseball long believed in “clutch hitters”, batters who perform especially well in important situations. However, Bill James, the spiritual father of sports statistics, showed that this was simply not supported by the data. Similarly, fans long believed that basketball players have “hot hands”, when they are “in the zone” and sink every attempt. Statistical analysis showed that “hot hands” were fully explained by chance. Is tennis really different?

One reason for believeing tennis is different is comes from this (superb) New Yorker article on the state of the doubles game. The relevant sections say:

The doubles tour might no longer exist, if not for Etienne de Villiers, the chairman of the men’s tour at the time. De Villiers had previously worked at Walt Disney International, so he understood the need for better marketing. The doubles tour could survive, he said, but only if the players agreed to some compromises. The game would be streamlined. Most matches would be kept to two sets, with a “match tie break” in place of the third set. If a game went to 40-40 the next point would decide it, there would be no more endless ads and dueces. (Grand slams would stick with the traditional scoring).

The new format has few fans among the players. Martina Navratilova says it is a “bullshit excuse”. Leander Paes calls it as “Russian roulette”, and Luke Jensen dismisses it as “tennis in a microwave”. Jensen believes that the shorter format favours weaker teams, “Anyone can win one set”.

Oddly enough, though, the statistics don’t bear this out. Not long after the changes were made, Wayne asked Carl Morris, a mathematician at Harvard, to calculate their effect on a team’s chances. In shorter matches, Morris concluded, the likelihood of an upset could increase by as much as five percentage points. And yet, when the ATP later reviewed the tour’s statistics, it found that the best players had improved their records. The new format offered “no second chances”, as Bob Bryan put it, but that wasn’t necessarily a bad thing. “The one thing we didn’t figure in is that the better teams are clutch” Wayne says. “On those big points, they come through”.

That said, this is a roundabout way of making a simple point. My friend Sriram Subramaniam suggested a comparison of % break points won with % other points won. One would expect Rafa to win more break points than Murray. Unfortunately, Google didn’t turn up this specific analysis. The closest thing to this analysis that a few mintues of Googling turned up is this paper by a Franc Klaassen of the University of Amsterdam.

He shows that there really are big points, and that seeded players play better on big points than unseeded players. He observes that seeded players facing a break point on their serve have the same win % as on other points, and that unseeded players have a lower win %, suggesting that it is more about weaker players choking than better players raising their game. He also shows that serving first in a set, or serving with new balls, has no impact. He doesn’t make any conclusions about champions like Rafa or Federer as opposed to the general pool of seeded players; his dataset is small, coming only from Wimbledon 92-95.

Calling for tennis’ Bill James to mine the vast amount of data generated by the ATP tour...