Reg hack ponders mysterious whiteboard symbols
By Burke Hansen in Palm Desert
Published Tuesday 22nd May 2007 15:02 GMT

House of Cards
Ever since the DARPA "Terror Casino" went down in flames a couple of years ago in the face of Congressional scorn, the idea that prediction markets can evaluate risk better than pundits or policymakers has been pushed nearly out of the public imagination.
The idea still has adherents among economists, however, and companies like Intrade and Betfair continue to carry the torch across the pond, offering virtual trading floors with contracts covering any event imaginable.
A rare gambling conference relatively close to home, with the leading academics in the field in attendance - a no-brainer. But really, what to make of a conference with titles like: "An Examination of the Determinants of Biased Behavior in a Market for State Contingent Claims"? Or this: "Non-Expected Utility Models and Heterogeneity in Risk Attitudes: Towards an Explanation of Gambling Outcomes for Individuals and Markets"? Better yet, this page turner: "Inferring Risk Preferences Using Synthetic Win Bets in Horse Betting 'Exotic' Markets"?
Got that?
We didn't either, but, in the interests of demystifying the cult-like status prediction markets have in certain quarters, we decided to take the plunge.
Steven Levitt, of Freakonomics fame, got the party started with a keynote address that gave his general opinion of the importance of gambling and prediction markets - namely, that in the grand scheme of things, they really aren't all that important. A bit of a dribbler for a kickoff speech, true, but the point that prediction markets will remain on the fringe of policy analysis - more intellectual curiosity than serious policy formation tool - seems accurate. Levitt showed more interest in using limited prediction markets within organizations as a way to uncover information that members or employees are unwilling to disclose face to face.
Companies such as BestBuy have started to utilize in-house prediction markets - where employees can win prizes by gambling on the outcome of certain company problems - to enhance the possibility that undiscovered employee knowledge will appear on the in-house market. Fair enough, but whether the information in the contracts is specific enough to be valuable was a question I tried without success to ask of Mr Levitt - BestBuy CEO's favorite contract on its internal market was one that bet whether or not a particular store would open on time. The answer was no - useful to know perhaps, but not exactly the kind of information that will solve the problem of the store not opening on time.
There was a healthy dose of Kool Aid going around among some of the others, however. Since the conference was sponsored by Intrade, that didn't really come as a surprise. Assertions about the accuracy of prediction markets got tossed around without always being backed up by the statistics themselves. That doesn't mean that stats necessarily do not exist, but it would have been nice to hear fewer bald assertions about the accuracy of the models used.
In the afternoon session, Bob Erikson of Columbia University did one of the better presentations I saw, which compared the Iowa Election Market (IEM) with the predictions of pollsters for presidential elections going back to 1988.
The comparison between prediction markets and polls was clever because it really cut to the heart of the argument for the accuracy of prediction markets - namely, that people will be more honest when they have to put their money where their mouth is. The results were mixed, however; markets did somewhat better in some measurements, but in the most important question of a basic win/lose call, the pollsters have so far outperformed the market.
Bill Eadington of the University of Nevada, Reno - the world's leading expert on the gaming industry - gave a great overview during lunch of the gaming industry as a whole and where it is poised to go in the next few years. It cut a swath all the way from the explosive growth of gaming in emerging markets to problems in neuro and behavioral economics and addictive behavior. The break from the prediction markets was as informative as it was refreshing.
Vasiliki Makropoulou of Athens University followed Professor Erikson with an interesting paper on price-setting in fixed-odds betting and the favorite-long shot bias; though I admit, my math wasn't strong enough to follow everything she said. Still, the insights into odds making were worthwhile.
The day wrapped up with a quirky and interesting presentation about bookmaker William Hill and the weight-loss bets it sometimes offers. Surprisingly, only 20 per cent of those who bet William Hill they can lose a certain amount of weight in a certain amount of time actually win, even though they have complete control of the outcome. The study demonstrates nicely how people tend to overestimate their ability to change their future behaviors, underestimating their own sloth. It has important repercussions in health care, for one - instead of penalizing smokers by raising their insurance rates over the long term, why not make them pay an enormous license up front to make them better understand the long term implications of their behavior?
Singapore does something like this with their gaming industry by forcing gamblers to pay hefty fees to enter casinos, or pay for an expensive yearly license. It's a worthy idea, and reason enough for economists to keep studying a market that gets little respect. ®
Burke Hansen, attorney at large, heads a San Francisco law office

Date Posted: 31-May-2007