This weekend's World Cup soccer action features not only Germany vs. Argentina, but also Paul the octopus vs. the statistical gurus. The verdict from the human experts who pore over soccer stats is that Argentina has a better chance of advancing in the competition. Despite that, the psychic cephalopod continues to cling to the German flag.
Part of the reason for Paul's fame may be because he's a homer. Although the octopus was born in England, he's currently ensconced at the Sea Life Aquarium in Oberhausen. His handlers have been setting up two plastic boxes in the aquarium, festooned with the flags of Germany and its World Cup opponent. Each box contains a mussel treat, and the flag on the first box that Paul goes for is considered his pick for the winner.
Paul's been 4-for-4 so far, siding with Germany except for the game that was lost to the Serbians. That may sound like an amazing feat, and that's why the octopus has stirred such a worldwide sensation, particularly after "predicting" Germany's win over England last weekend. But it's not as if he's won the lottery.
The chances of correctly predicting win-loss outcomes four times in a row, using a purely random process like coin flips, would be 6.25 percent. Thus, if there are 100 kooky experiments to pick four World Cup winners, about six of them should come up with the right result. It's just that you don't hear about the kooky predictions that bomb - such as Petty the Hippo's wrong guess in the Germany-Serbia game, or Anton the Monkey's goof in Ghana vs. Germany, or Leon the Porcupine's bad prediction for Germany vs. Australia, or Lissy the Fox's flameout in the Germany-England match.
When you put together all those predictions at the Chemnitz Zoo in Germany, the animals went 0-for-4 - which is just as amazing, statistically speaking, as Paul's performance in Oberhausen.
Prediction markets: Advantage Argentina
Will humans beat Paul's predictive powers? Some humans will, some won't. But if you're wondering which team you should bet on, you can pick up some cues from the oddsmakers, prediction markets and network analysis. All of those indicators currently favor Argentina to beat Germany, and Brazil to triumph in the end.
Some folks have compared prediction markets to gambling, and they do work in similarly mysterious ways: Online markets give users the opportunity to "invest" (usually with play money) in the outcome of particular propositions: Will the Republicans take control of the Senate? Who'll win the best-picture Oscar? Who'll win the Super Bowl? If you're on the right side of the proposition, you win the maximum payoff. If you're wrong, you lose the investment.
You do have the ability to buy and sell "shares" as long as the proposition is still in play. For example, you might have bought Germany's Inkling Markets shares on Tuesday, when it was trading at 44.32 points. Today it's listed at 48.81, so you could theoretically turn around and sell your shares for a profit. But if you really believe Germany will win, you'll want to hang onto it for the 100-point payoff this weekend.
Prediction markets have proved their worth in presidential elections: Research from the University of Iowa, which runs the only real-money political prediction market in the country, indicates that such markets perform at least as well as traditional polling methods. Most recently, they tracked the rise of Barack Obama over his rivals (with the exception of that New Hampshire setback, which the Iowa Electronic Markets also correctly reflected).
Network analysis: Brazil in the end?
The World Cup serves as a great laboratory for studying the performance of predictions. For example, "Freakonomics" guru Steven Levitt was the co-author of a 2006 paper that looked at market efficiency in World Cup wagering. (The conclusion was that savvy traders could take advantage of the inefficient prices set by market makers.) More recently, Northwestern University's Luis Amaral and his colleagues developed a ranking system for soccer teams as well as for individual players that could be used to project performance trends.
Amaral and his colleagues have now set up a website that ranks all the World Cup teams, based not on the actual outcome of games, but on the connectedness and efficiency of team networks. The method isn't perfect: Only five of Amaral's top eight teams are in the actual knockout round - though, to be fair, some of those top-rated teams had to knock out others to get into the final eight.
Despite the discrepancies, the network analysis agrees with the investors' verdict in two important respects: Argentina (No. 2 on Amaral's list) has a significantly higher rating than Germany, and Brazil has just a bit of an edge over Argentina for the No. 1 spot.
That doesn't mean the tournament's outcome is set in stone. Anything can happen, especially in a traditionally low-scoring game like soccer. But the prognostication does set up an interesting interspecies duel this weekend - and if Amaral's ratings are right, Brazil will be in a heck of a game against Argentina a week from Sunday.
Which side are you on? Team Octopus or Team Geek? Oh ... did you think I was talking about soccer? OK, then: Feel free to leave your comments on the prediction game, or the actual World Cup games if you like. And keep a baleful eye on NBCSports.com for World Cup updates.
More soccer science:
- Penalty kicks may be predictable
- Devuvuzelator squelches the vuvuzela
- Altitude could cause soccer players to overshoot
- TechNewsDaily: How tech could help World Cup refs
- Physics Today: Power and spin in the beautiful game