Michael Giberson
Prediction markets (a category which includes gambling on event outcomes) work well in some cases and less well in others. Justin Wolfers and Andrew Leigh have an overview of prediction markets appearing in The Melborne Review, which states the pertinent point: “attempts to set up markets on topics where there are insiders with substantial information advantages have typically failed” because the cadre of highly informed insiders will tend to drive out the partly-informed public.
But if that is the case, then why all the gambling/prediction market interest in the fates of two fictional characters – Harry Potter and Tony Soprano – and their equally fictional associates?
These fictional worlds do not produce widely dispersed bits of information that can be usefully aggregated by a market. Careful study of, say, the first six Harry Potter books and reading interviews with J.K. Rowling may produce some sense of what will happen, but ultimately whether “Harry Potter must die!” (as Marginal Revolution suggested) or “Harry Potter will NOT die!” (as Chris. F. Masse at Midas Oracle interprets prediction market prices to suggest) will depend on what Rowling wants to say through the book and how she decides to do it.
I don’t think there is sufficient information available to form a good probability estimate. So the betting is all based on emotion, some bettors will be lucky and others not. An informed insider can enter the market and clean up.
A lot of people enjoy betting on entertainment events (a category broad enough to include everything from the Super Bowl winner to week-end movie box office totals to the survivor/winner in the TV show “Hell’s Kitchen), and far be it from me to want to squash people’s fun. Fun is good, and good for business if you are a prediction market maker.
But if you are an investor trying to maximize long term returns and have no inside information, this is a case where Kelly’s criterion for betting comes into play. Kelly’s criterion for setting bet size can be described as “edge over odds,” which implies, as Wikipedia explains: “If the gambler has no edge, … then the gambler should bet nothing.”
NOTES: The Wolfers and Leigh piece provides a good general background on prediction markets. HT to Midas Oracle for the heads up on the Wolfers and Leigh article. Also at Marginal Revolution: Tony Soprano must live, citing a Washington Post story among others.
[A lightly edited version of this post is also up at the prediction market group blog Midas Oracle under the title Harry Potter will NOT die? Don’t Bet on It.]