Predictive Markets

I’m somewhat fascinated by predictive markets. As a bit of a political junkie, I’ve followed InTrade during the primaries.

If you don’t know what a predictive market is, imagine that instead of trading a stock, you are trading on whether or not something will happen. If the thing actually does happen, then your share ends up being worth $100 and if it doesn’t happen your share ends up being worth $0. So the current stock price represents what people think the likelihood percentage of that event happening. For example, on InTrade you could trade on the likelihood of Elliot Spitzer being indicted on felony charges before June 30th. It is currently trading at 17.5, so the market doesn’t think it’s very likely (17.5% likely). However the contract on Hillary Clinton winning the West Virginia primary is trading at 97.6 so the market considers that a virtual certainty.

It’s interesting because on the actual day of a primary, often the networks really know who is going to win based on exit polls. But they won’t announce anything until the polls close. Clearly the people who actually trade on InTrade have some level of inside information, because you can watch the movement in real time and get a sense for who will win. It’s day trading for political junkies.

Predictive markets have been shown to generally be more accurate than polls. They aren’t perfect, but they are currently a very hot topic these days.

What’s really fascinating is when a company uses predictive markets to forecast internal business objectives. This is done at a small percentage of companies, but with real success. Google and Best Buy have used predictive markets extensively. The logic is that the folks “on the ground” really working probably have better information than one or two “experts”.

This is something I would love to do some experimenting with…