While I’m on the subject of prediction markets, I might also add that they’re not even internally consistent, some of the time. Brian Weatherson has looked at arbitrage opportunities across markets, but they exist within markets as well. Here’s some free money for people trading at InTrade.
Consider: PRESIDENT.REP2008, the probability that a Republican will be the next president, can be bought in bulk at 37. If you want to be really safe, you can buy PRESIDENT.FIELD2008 at 1.7 as well, for a total cost of 37.7.
Then, on the short side, you can sell both NONDEM.PRES-GOVT.DEBT and NONDEM.PRES-TROOPS.IRAQ at 43.8. Those are contracts which expire at zero if a Democratic president is elected, and expire at somewhere between zero and 100 if a non-Democratic president is elected.
If a Democrat is elected, then, you lose 37.7 on your long but gain 43.8 on your short, for a net gain of just over 6. And if a non-Democrat is elected, you gain 62.3 on your long and lose a maximum of 56.2 (and quite possibly substantially less than that) on your short, for a net gain of at least 6.
Now there is an opportunity cost to tying up your money until after the elections. But if prediction markets were remotely as efficient as many people seem to think they are, these kind of obvious arbitrages wouldn’t exist.