wherein DF travels to Deutschland for the 2006 world cup to follow the US men's national soccer team

Saturday, March 04, 2006

The strength and weakness of World Cup groups: some numbers

Which is the real group of death? This question draws lots of attention because it points out one of the major issues of the World Cup draw: that its focus on geography can result in massive differences in the quality of competition from group to group. This is part of the game; ex ante, no team has a much greater or worse chance to be in a GoD. However, once the groups are drawn the debate tends to focus on which teams are better than others in the putative GoDs, and this doesn’t really get very far. We could debate all day about whether the US is better in its group than Serbia is in its, or Cote d’Ivoire versus Ghana, ad nauseam.

So I’ve tried to use the ELO ratings to provide a more determinate look at the relative difficulty of the eight groups, with some interesting results. I’ve averaged the ELO ratings of each team in each group to give some sense of their average degree of difficulty. This isn’t a foolproof method, of course. The ELO ratings aren’t guarantors of current quality and certainly don’t predict how good teams will be on the eve of the WC. Also, averages like this can be deceiving. A group with three excellent teams and one poor one might be much more unfair than a group with two superior and two inferior teams, even though the average ELO ratings of both those groups might appear equal. Anyway, with those caveats in mind, here’s what I’ve found:

Group A………..26.25
Group B………..25.75
Group C………..16.5
Group D………..28
Group E………..21.5
Group F………..15.75
Group G………..37
Group H………..33.25

The average difficulty for a group was 25.5. So we have a couple groups (A and B) right at the mean; one group a bit tougher than the mean (E); one group a bit easier than the mean (D); two much more difficult (C and F) and two much easier (G and H).

So compare this to the conventional wisdom. I looked at the groups and initially thought there were some major disparities. Just looking at the teams, I perceived groups C and E to be clearly the hardest; and groups G and H to be clearly the weakest. According to ELO, E may not be as hard as all of us US fans seem to believe. The numbers could be deceptive, though; Ghana’s got a very bad ELO rating (67), which skews the average upward of what I think it should be.

Also, I didn’t rate F as a very difficult group. I think any one of Croatia, Japan, and Australia should be thrilled to make the second round, and one of them is guaranteed that. Not so, according to the ELO averages, which has F as the real Group of Death, even harder than C (which includes Argentina, Holland, Serbia, and Cote d’Ivoire). The conventional wisdom that G and H are weaklings seems right, though. Lucky bastards.


Blogger DF said...

I cross-posted this information on a thread on bigsoccer.com and another user pointed out that it might make more sense to use the total points in ELO rankings rather than the rankings themselves. This poster was kind enough to actually run those numbers, which yielded basically the same results in an interesting confirmation of my original conclusions:


4:02 PM, March 05, 2006


Post a Comment

<< Home