Monday, June 07, 2010

2010 World Cup: Who Do the Rating Systems Predict to Advance?

Four years ago, I looked at a number of factors in an attempt to see which would best predict the 16 teams to advance out of the group stage: form, GDP, odds, qualifying record, population, etc. In the end, the best predictors were the ELO rankings and the betting odds, which both got 13/16 right. Both missed Australia/Ecuador/Ghana. Actually, ELO had Iran and Mexico with the exact same ranking, and I'm giving them credit for Mexico. So count it as 12.5 if you like. The much-maligned FIFA rankings only got 10/16, missing the same three as well as Italy/Switzerland/Ukraine (remember the Czechs and USA both being ranked in the top 5?). So it's no wonder that they changed the formula immediately afterward.

Since the FIFA rankings began in 1993, here's how both systems have done predicting the advancing teams:


ELO FIFA
2006 13 10
2002 10 12
1998 14 11
1994 13 12



Total 50 45
PCT 78.1% 70.3%

Who's favored to advance in South Africa? Let's take a look. I'm also including betting odds, SPI, Voros' simulations, and the Castrol Index.

2010 FIFA World Cup: Predicted to Advance


ELO FIFA Odds SPI Voros
Castrol
Group A MEX, FRA FRA, URU FRA, MEX URU, FRA FRA, MEX
FRA, URU
Group B ARG, SK ARG, GRE ARG, NIG ARG, NIG ARG, GRE
ARG, NIG
Group C ENG, USA ENG, USA ENG, USA ENG, USA ENG, USA
ENG, USA
Group D GER, SER GER, SER GER, SER GER, SER GER, SER
GER, SER
Group E NET, DEN NET, CAM NET, DEN NET, DEN
NET, DEN
NET, DEN
Group F ITA, PAR ITA, PAR ITA, PAR ITA, PAR ITA, PAR
ITA, PAR
Group G BRA, POR BRA, POR BRA, POR BRA, POR BRA, POR
BRA, CIV
Group H SPA, CHI SPA, CHI SPA, CHI SPA, CHI SPA, CHI
SPA, CHI

EDIT: SPI's final pre-World Cup sims showed Denmark advancing instead of Cameroon.

Some notes on how I determined this. For FIFA, I just took the highest rated teams to advance. For ELO, I'm using the current rankings as of today (which may change a bit before the tournament). So it's fair to say that FIFA may be at a disadvantage since they only update once per month, and don't take into account the pre-tourney friendlies. For the betting odds, I looked at several top sites and they were all the same. Some sites had Cameroon equal to Denmark for the runner up in group E, but I didn't see any that had them ahead. For Castrol, I looked at their group predictions, which somehow are different than their listed advancement chances for each team (otherwise Australia and Portugal would be listed).

All five have the exact same picks for groups C-D-F-H. Apparently Serbia is a bigger favorite than I'd thought. In fact, all five pick 12 of the same teams to advance (including the USA), which means that there are only differences of opinion on 4 spots. Here's the differences:

ELO MEX-SK-DEN-POR
FIFA URU-GRE-CAM-POR
Odds MEX-NIG-DEN-POR
SPI URU-NIG-CAM-POR
Voros
MEX-GRE-DEN-POR
Castrol URU-NIG-DEN-CIV

No love for South Africa, even at home. Despite virtually everyone being down on their chances, all six have France advancing from group A. Given Mexico's recent success at the World Cup group stage, it's surprising to see Uruguay with strong support. However, Mexico is ranked first in the group by ELO. On the other hand, Nate Silver's SPI ranks Uruguay as the group's top team. Quote: "Uruguay was a constant source of debate during the creation of the SPI." As probably the biggest difference between the systems, SPI may be judged solely by that pick. Also, I'll note that A is the only group with more than one team picked to finish first.

Group B is the only one with all four teams picked to advance by someone. If all three non-Argentina teams are evenly matched, will home continent advantage for Nigeria be the difference? South Korea has the WC experience, while Greece boast the greatest achievement (Euro 2004).

Denmark vs Cameroon is considered a toss up in terms of advancement by most of the betting sites, and may be the most up in the air spot of all. Nobody is giving Japan any chance, and that seems fair. Their coach said earlier this year that they could reach the semifinals, but reportedly offered to resign after losing to South Korea.

With Drogba's injury, Portugal is even more favored to move on. Even Castrol, the only one to pick Cote d'Ivoire, lists Portugal with a slightly better chance of advancing (51.7% vs 51.1%). Once again, I can't explain why it's listed that way.

Labels: , , , , ,

Comments on "2010 World Cup: Who Do the Rating Systems Predict to Advance?"

 

Blogger Dorian said ... (6:42 PM, June 07, 2010) : 

Great post! Very interesting! Thank you.

Did you see that several brokerage firms (UBS, JP Morgan) have written investment reports that predict the match winners, group advancers, etc. using quantitative methods? Would be fun to track their accuracy as well.

Also, FYI, Voros does have May 13th, 2010 rankings here:
http://vorosmccracken.com/nationalteamrankings051310.html

 

post a comment