It’s interesting to compare the World Cup odds generated by the Soccer Power Index against those offered by handicappers, like the consensus odds at the website oddsportal.com. There are some significant — and systematic — discrepancies.
The table below presents the probability of each country winning and advancing from its World Cup group, reaching the semifinals of the knockout stage, and winning the whole tournament, as determined from our most recent run of about 20,000 SPI simulations.* These probabilities are then translated into odds — 6.0 means that a team’s odds of winning the World Cup are 6-1 against — and compared against the average betting lines published at oddsportal.com. We then have two additional calculations: EV is the Expected Value, net of the original wager, on a bet of $100. Kelly is the Kelly Criterion, which is essentially the expected value of the wager divided by the odds, and which is designed to provide a suggestion as to how much of his bankroll a gambler should place on any one bet. (Note that the Kelly Criterion is often regarded as too aggressive.) Teams are sorted by the odds they are now assigned of winning the tournament by the consensus of bookmakers.
There are some real discrepancies here. Spain, for instance, are given nearly identical odds to Brazil. Since the teams are regarded about equally strongly by most ratings systems, that would seem to make some sense. But Spain lost its first match in a group where two other teams won theirs. That’s a lot of trouble to dig out of, even for a team like Spain — and our program thinks they have only even odds of doing so. Brazil, by contrast, although they have some tough matches ahead, is 85 percent likely to survive its group.
More generally, in fact, the bookmakers’ odds seem to be too slow to catch up to the matches that have been played so far. France, remarkably, is still being offered with odds of only about 120:1 against winning the tournament, even though it has just one point between two matches: they are about 30:1 against even making it out of their group. Meanwhile, teams like Paraguay and Ghana, who are very likely to advance, do not seem to be getting enough credit for it.
Thirteen teams, overall, are figured by SPI to offer some kind of wager that is better than breakeven. They are — in descending order of Kelly — Uruguay, the Netherlands, Chile, Brazil, Mexico, Ivory Coast, South Korea, Ghana, Paraguay, Portgual, Slovenia, Nigeria and Denmark. A bet on Argentina is also within the margin of sampling error of breakeven.
What do these teams have in common? Very few of them are from Europe, and the exceptions — the Netherlands, Portugal, Slovenia and Denmark — are small countries which average only about 8 million inhabitants each. In contrast, some of the larger European countries, like Spain and Italy, offer some of the least favorable bets. The handicappers may very well know their markets — but the popularity of teams in those portions of the world where people have enough disposable income to bet on football seems to be as much of a factor as the actual talent on the field. (Outside of Europe, wealthy United States and Japan also appear to offer somewhat unfavorable odds.)
All five South American countries, by contrast, offer odds which SPI regards as breakeven or better. And it’s not as though South America hasn’t shown its resolve in the tournament; they are undefeated in seven matches thus far, including five wins, whereas Europe is a middling 5-5-5. Europe did very, very well in the 2006 tournament, which was played in Germany, but they haven’t performed as strongly in tournaments staged outside of their continent. It may be that people are too entrenched in their perceptions about the relative strength of different continents based on the previous World Cup, while ignoring counterexamples like 2002.
Or — it may very well be that SPI is wrong. It’s certainly harder to rate international football teams, which rarely play competitive matches outside of their confederations, than it would be teams in another sport. I don’t know that I’d bet my livelihood on the notion that Uruguay are only 11:1 underdogs against winning the tournament, for instance — although getting 40:1 (and one bookie is offering them at 80:1) seems like a very good price. But the biases seem systemic enough — toward Europe and against the rest of the world; toward wealthier, more populous nations and against their opposites; toward pre-tournament expectations and against the predicament that the teams actually find themselves in — that there are probably some significant inefficiencies in the prices the bookmakers have established.
* A couple of teams failed to win the World Cup in any of the 20,000 simulations. Their odds are determined algorithmically, based on the proportion of the time in which they advanced to the semifinals.