## DataLab

Statkeepers at Wimbledon have a reputation for generosity. Often what looks like a player’s bad mistake appears to the tennis tournament’s official match-trackers to be the result of great play by the opponent. After tracking one of this year’s men’s semifinals on Friday, and looking at other independent match scoring, I think the reputation is deserved.

I watched every point of the Roger Federer-Milos Raonic semifinal, either at a desk in the press center, from Centre Court or via online replay. (You can replay the match using WatchESPN or, depending where you are, the BBC iPlayer. Federer won in straight sets, and faces Novak Djokovic in the final on Sunday.) After every point, I noted in a spreadsheet who won, and whether it ended with a winner, a forced error or an unforced error. Then I cross-referenced my point-by-point stats with those kept by the tournament’s official scorers and compiled by its official data-provider, IBM.

We read the endings of 92 percent of points exactly the same way. When we differed, it was usually because what looked to me like a forced error was scored as a winner, or what looked to me like an unforced error was scored as a forced error. Our disputes meant that the players combined for 7 percent more winners by their count, and for 21 percent more unforced errors by my count.

Wimbledon’s generosity of scoring spirit is corroborated by a count I helped conduct last year and by tennis fans who scored eight other matches at Wimbledon since 2006 as part of a charting project. The observant fans consistently found more unforced errors than showed up in the official counts, an average of 46 percent more and always at least 10 percent more.

Unlike me, the other scorers counted more winners than the official scorers did — 9 percent more, on average. Still, the official stats made it look like the matches were played much more cleanly than the independent counts suggest: 1.8 winners for every unforced error in the official scorecards, compared to a ratio of 1.4 in the unofficial counts. Combining those eight matches with Federer-Raonic, here’s how the two views differ:

As far as stats disputes go, this isn’t a particularly momentous one. Winners and unforced errors aren’t kept for every match, and they aren’t archived on websites that track stats year-round, such as the ATP’s, ITF’s and WTA’s. Generous scoring doesn’t affect the result of matches. It merely paints the quality of tennis at Wimbledon as a little better than it is — a little cleaner, with less messy human error involved.

There’s room for disagreement because there’s no single official guideline for how to score points. The All England Lawn Tennis and Croquet Club, which hosts Wimbledon, denied my request to sit in on a training session for scorers, some of whom in past years have been young amateur tennis players.

The general thinking behind scoring is that a winner is a shot that goes in and then bounces twice or otherwise goes out of play before the opponent gets a racket on the ball. An error is a shot that goes out. A forced error was forced by the quality of the opponent’s shot or court positioning, while the player who makes the error is at fault on unforced errors.

That leaves plenty of room for interpretation — and disagreement.

I re-watched the points where the official scorer and I disagreed in Federer’s defeat of Raonic and found a philosophical split over what constitutes a forced error. Seven times I saw an unforced error that was scored as a forced error. Five of the seven times, the error immediately followed a service return. All of the returns were good — hit hard or placed well, or both — but the server, in my judgment, still had time to set up for the next shot, and often missed while going for an aggressive one.

For instance, in Federer’s first service game, he backed up to field a Raonic slice return, and then redirected it with his forehand into the opposite corner. It went just long. It was scored as a forced error. That looked to me like an unforced error: The only person who forced Federer to go for that shot was Federer.

Just once did I see a forced error where the scorer saw an unforced error. Federer approached the net behind a shot deep into Raonic’s backhand corner. Raonic had time to set up for an attempt at a passing shot, but sliced the ball into the net. I followed the principle that when an opponent is at net most errors are forced, because of the need to come up with a great shot to counter the opponent’s dominant position. The scorer tended to follow that principle, but not on this point.

Winners generally are less subjective, and the subject of smaller disputes between the official and unofficial counts. Five times, though, I scored a Raonic shot as forcing an error where the scorer awarded him a winner. Each time Federer got his racket on the ball, but wasn’t able to direct it anywhere near the net. It’s reasonable to count these as winners, though it takes a judgment call to decide whether a miss was close enough to going in to count as an error.

When I presented my results to the club, a spokesman responded with an email saying, “The labeling of errors as forced or unforced is very subjective so inevitably there will be differences between one score and another.”

Carl Bialik is FiveThirtyEight’s lead writer for news.

All Sports

All Wimbledon