Sunday, July 28, 2013

The Numbers Game: Are Corners Really Worthless?

I picked up The Numbers Game, by Chris Anderson and David Sally, while I was in Oxford earlier this summer. I really enjoyed it, and echo the positive reviews that the book has received. That said, I do have a few nits to pick, which I'll do over a series of posts.

Here I'll take on the claim that "corners are next to worthless," a conclusion which I think is vastly overstated. The Numbers Game argument rests on a chain of causality from corner kicks to shots to goals, and a finding that very few corners actually lead to goals.

In a blog post that first presented the analysis which appeared in the book Anderson and Sally explain their logic:
The data show that, on average, about 20.5% of all corners (that's 1 in 5) led to a shot by the team that took the corner within three touches of the ball. . . Of the corners that actually led to shots within three touches, what were the odds that the ball crossed the line? . . . On average, about 11% or 1 in 9 shots created from corners produce goals. Incidentally, that's also right around the average ratio for goals to all shots taken in soccer matches (the Reep ratio). . . The yield from corners ranges from 0 to .07. On average, the data show that a corner is good for (drumroll ....) 0.022 goals. This means that the average EPL team scores 1 goal from a corner about every 10 games. And this helps to explain the lack of a correlation between the number of corners and goal scoring. The infrequency of the goals from corners combined with a lack of dispersion between teams in corners per game lead conspire to make corners mostly, well, useless when it comes to scoring goals.
Let's now put the data on corner kicks into a bit different context.

In the Premier League in 2012/2013 there were 1,008 goals scored from 7,528 total shots.  So 13.4% of shots resulted in a goal scored, or each shot is worth 0.134 goals. This means that in the currency of goals, a shot is worth about 6 times as much as a corner (i.e., 0.134/0.022). Already, you can see that a corner is far from "next to worthless."

Let's now look at corners in the context of the larger category of which it is a part, -- the pass. According to Opta, in 2012/2013 the Premier League had 285,482 total passes. If we follow Anderson and Sally's approach with corners and define the three touches prior to a goal being scored as leading to the goal, then this means that 3,024 of the total passes led to goals (setting aside, for simplicity, unassisted goals).

Thus, in goal currency, each pass is thus worth 0.011 goals (i.e., 3,024/285,482). Put another way, corners are twice as likely to lead to a goal than passes generally. If corners are next to worthless, then passes must be even more worthless.

So how valuable are corners?

Simply looking at aggregate statistics, corners carry about 17% of the value of a shot, expressed in goal currency, but they carry 200% of the value of a pass. A more sophisticated analysis would consider the spread of value in goal currency across teams, and no doubt would find that for some teams, the value of the corner is far greater than the averages shown here.

So are corner kicks useless or next to worthless? Hardly.

1 comment:

  1. Your math is:
    # of passes leading to goals / all total passes
    You're comparing it to:
    # of corner kicks leading to goals /
    all corner kicks

    The problem is that corner kicks are very different than passes. Passes can be made anywhere on the field and are often made for control of the ball.

    Corner kicks on the other hand are supposed to be able to be somewhat effective attempts on goal as a result of defensive play. Your analysis shows the insight that it's twice as valuable as a random pass on the field, which to me doesn't seem that valuable.