Article:A Passer Rating for soccer

There was an article in the paper today about Billy Beane working on metrics for football (soccer), and it had four examples, all of which would be reasonably crude and subject to noise (one was how many times the player wins the ball defensively, which I imagine comes out a lot higher for the ball-winning CB than the covering CB - it would favour Vidic and Terry over Ferdinand and Carvalho).

I don't think people in the press here understand at all how significant it's going to be if statisticians can crack football like they have baseball, because a team free from the weight of years of football prejudice and ignorance (anybody who saw the brilliant, hilarious tv programme Big Ron Manager knows what I'm talking about) that passes for conventional wisdom would be at a huge advantage. And it would be bound to improve our understanding of the game. It's a fascinating area, and I wish everybody involved in it didn't have such an incentive to keep their work secret.

So here's my suggestion for a refined soccer metric, which I'm calling Passer Rating after the gridiron thing. It would need ProZone-style data, which means that I can't do it personally and this is just a thought experiment: 1. Divide the pitch up into rough sections - defensive 3rd, middle 3rd, attacking 3rd would work, but the more sections the better. Probably at least 12 sections would be needed, but I'll go with thirds for the examples.. 2. Look at the % of successfully completed passes for all players in that area of the pitch 3. Now the clever bit - weight each player's INDIVIDUAL passing completion % by the area of the field in which the pass was made to create a quality stat - passer rating. So if the average completion % in the defensive third is 85%, and a player is completing 75% in that third, then that means minus points for their rating. If the average for the attacking third is 65%, and a player's completing 80% (like, say, Kenwyne Jones) then they get big positive points. You could start the rating at 0 or at 100 or whatever, as long as it was standard for everybody. Dirk Kuyt and Justin Hoyte are coming out of it badly, whatever you do.

Good, no? So Nicky Butt-at-Man-U style passengers who rack up reasonable passing %s by stroking easy ball around when the team are 3-0 against Norwegian minnows up in Europe would be shown up - because the bad passes they made in easy areas would hurt their rating more than a striker trying a flick-on inside the box. Whereas a demigod like Zidane would have an amazing rating because he completed an amazing amount of passes in tight areas.

And to make it more effective, you could even do: 4. An extra weighting to take into account average passing % at a given score (I bet teams that are 3-0 down complete more passes than teams at 0-0) 5. Look at how the presence of certain players affects others passer ratings - I bet having a good Drogba/Crouch-style target man would help team-mates ratings. 6. You could weight it by opponent - I bet teams would have lower passer ratings against Benitez and Mourinho teams because of their great pressing. Completing a lot of passes in attacking areas against them should count for more than against Derby. Doing this would show up flat-track bullies 7. Maybe do one where you take out passes over 20 yards from wide, or something, so you could remove the distorting effect that wingers going for crosses instead of passes would have.

Don't you think that would be an interesting thing to have? I think it would genuinely be revealing. I'd bet the list of top 20 in the prem would have mostly players you'd expect - Carrick, Fabregas, Xabi - but one or two surprises (like maybe Arca, who is classy and very under-rated) and the position of driving players like Lampard in the list would be interesting. And I'd LOVE to see a historical list of this sort to see where Zidane in 98 and 2000 stood..