Wednesday, May 18, 2011

Intrade Uptick Lowdown: Update

With regard to the previous post, I discovered that Nate Silver agrees with me, which is always reassuring. He also makes I point I didn't make, which is that small increases in approval may translate into large increases in probability of reelection, because a lot of elections are close.

Reading some of the chat on Intrade's website, I realized that I had overlooked something. The number I quoted was the probability that Obama gets elected president. But for that to happen, two things must happen. First, he must get the nomination. Then, given that he has gotten the nomination, he must get elected. The Intrade markets estimate the probability of him getting the nomination at 92%. (For example, he might for some reason decide not to run. Or he might get caught up in some scandal. Or have health problems. None of these things seems very likely, but they could easily add up to 8%.) If his overall probability of getting reelected is, say, 60%, that implies his probability of being reelected given that he has gotten the nomination is 65% (.92 x .65 = .60). And 65% is a lot closer to my completely subjective and unreliable estimate of the probability that he could defeat the Republican candidate, which was around 2:1, or 67%.

So perhaps it's just as well that I didn't buy in. On the other hand, my point from before still holds: Whatever the true probability was before Osama died, it's got to have gone up a little.

Monday, May 9, 2011

Chronicle of a Death Not Foretold

Intrade is an online "prediction market." The basic idea is that markets aggregate a lot of information from many dispersed individuals trying hard to make money, and so they can usually make better predictions than individuals can. That's why most economists consider the stock market valuation of a company to be the best measure of its value.

Intrade capitalizes on this idea by setting up speculative markets in particular outcomes to be predicted. For example, they might sell shares that (to simplify slightly) will pay $1 if the Higgs Boson subatomic particle is discovered by December 31, 2011, and zero if it isn't. Then if shares are selling for $0.20, as they are, you can interpret that as the market estimating that there is a 20% chance of the Higgs Boson being discovered in 2011. (There are some big caveats here, including whether enough people are participating in this market to give you a good estimate.)

Of course, one of the things people are most interested in forecasting is election results. Recently the Intrade market was predicting the Obama had around a 60% chance of being reelected. That seemed a little low to me, so I briefly thought that I should buy some shares, but of course I didn't.