05 April 2013

'Sabo'-metric look at scoring stat

Courtesy Real Salt Lake

There are a lot of statistics in soccer these days, and most of the time, I’m all for it. Finding numbers to illustrate key points or tracking long-running streaks can provide great talking points. But I get really frustrated when statistics are quoted out of context.

One such statistic caught my eye in this week’s MLSsoccer.com piece by Simon Borg, a friend and colleague with whom I love to argue and debate. In writing about Real Salt Lake forward Alvaro Saborío, Simon passed on a couple of statistics cited by Real Salt Lake GM Garth Lagerwey, presumably to indicate that Saborío is a clutch goalscorer and essential to his team’s success. The big statistic?

Real Salt Lake is 31-5-6 in all competitions when Saborío scores.

I have seen this stat used before in game notes, on television broadcasts, and in articles. But I have never found it to be particularly enlightening. After all, soccer is a low-scoring game. Isn’t it pretty obvious that a player scoring a goal is going to increase his team’s chances of winning? Don’t most teams have a high winning percentage when scoring at least one goal? Even if a team’s record when an individual scores is better than the team’s record when anybody scores, is it really because that player scored or is it perhaps because the team happened to play better defense in those games?

I asked Simon for his thoughts on the statistic, and he thought it was useful because it illustrated “the sense of Lagerwey and [head coach Jason] Kreis that [Saborío] is a clutch scorer. … There are plenty of scorers out there; not all of them show up when it counts.”

That’s certainly fair enough. I rate Saborío very highly and definitely consider him a key ingredient of Real Salt Lake’s success. I have enjoyed my conversations with both Lagerwey and Kreis, and I have an excellent relationship with the RSL communications staff. But I remained unconvinced that the statistic itself is all that indicative. I expected that most teams would have a winning record when their leading scorer found the net. I had no proof, however, so I set out to look at a few cases to test my theory.

In MLS last year, 13 players scored at least 10 goals. For each player, I went through his game-by-game log (thank you, Peter Hirdt and Elias Sports Bureau) and totaled his team’s record when he scores. I then looked up each team’s regular-season record when it scored at least one goal.

The win percentages based on individual scoring ranged from .500 (Will Bruin) to .857 (Steven Lenhart), with an average of .751 and a median of .773. Most of the numbers clustered between .750 and .789, and most players’ teams had suffered at least two losses when they scored.

The team numbers were, across the board, lower than the individual numbers, ranging from .500 (New England) to .818 (Real Salt Lake) and averaging .705.

Next, the crux of the issue: the difference between a team’s winning percentage when scoring any goal and its winning percentage when a particular individual scores. The average difference was .046, meaning a team wins 4.6% more of its games when its 10-goal striker scores than when just anybody scores. The highest number belonged to Steven Lenhart (.857 when he scored vs. 719 when any Earthquake scored), while the lowest number belonged to Will Bruin (.500 when he scored vs. .667 when any Dynamo player scored).

Player - 2012 seasonInd%Tm%Diff.
Steven Lenhart, SJ.857 (5-0-2).719 (19-5-8)+.138
Kenny Cooper, NY
.846 (10-1-2).714 (16-4-8)+.132
Saer Sene, NE.611 (4-2-3).500 (9-9-5)+.111
Robbie Keane, LA.750 (8-2-2).655 (16-7-6)+.095
Thierry Henry, NY.800 (8-2-0).714 (16-4-8)+.086
Chris Pontius, DC.833 (7-1-1).750 (17-4-5)+.083
Chris Wondolowski, SJ.789 (13-2-4).719 (19-5-8)+.071
Eddie Johnson, SEA.750 (8-2-2).704 (15-4-8)+.046
Fredy Montero, SEA.750 (6-1-3).704 (15-4-8)+.046
Alan Gordon, SJ.708 (7-2-3).719 (19-5-8)-.010
Kei Kamara, KC.773 (8-2-1).788 (18-3-5)-.016
Alvaro Saborio, RSL.800 (8-2-0).818 (17-3-2)-.018
Will Bruin, HOU.500 (2-2-6).667 (14-5-8)-.167
Nine of the 13 individuals had helped their team to a better win percentage when scoring, but Saborío (RSL was 8-2 when he scored) was actually one of the four who did not, along with Bruin, Alan Gordon, and Kei Kamara. Ironically, RSL was 7-0 in the 2012 regular season when since-traded forward Fabian Espindola scored.

Interestingly, a team’s defense might make it harder for a forward to steal the show. Of the seven players with the highest difference in the above table, all seven played for teams that allowed at least 43 goals. In the bottom six, only one (Gordon) played for a team that allowed at least 43 goals.

I would like to see Elias or Opta take a crack at these numbers and consider a much wider range of players and teams. But the data so far does not give me any reason to change my conclusion: Real Salt Lake does have a better chance of winning if Saborío scores, but not a significantly better chance than if anybody else scores.

No comments:

Post a Comment