Back to top

Predictive Markets - Can't Wait Until 2008

Predictive Markets

Predictive Markets are a really neat idea. Basically, you let knowledgable people bet on the likelihood of a particular event happening: Bush getting re-elected, 10 inches of snow falling in one day on New York City by 2008, or Nikki letting me buy a new computer this year. Then the price of the contract for that particular event will then reflect the collective knowledge of the community about that event. If you have something that is hard to analzye or hard to predict, this is a great way to get a fairly reliable answer.

Greed is good

The quality of the prediction is the basic concept behind Efficient Market Theory. Efficient Market Theory says that when information costs and trading costs are low enough, markets will seize upon any inaccuracies in the price of a security and will then price that security at a price that the market judges collectively to be an accurate price. The Gordon Gekko version of it is "Greed is good" because it makes sure that market prices accurately reflect the knowledge about that security. Is oil relatively cheap? Efficient Market Theory says that's because it's more plentiful than the anti-oil activists would have you think. Is it getting more expensive? Perhaps it is getting less plentiful or otherwise difficult to deliver.

The use of efficient markets as an indicator is extremely useful. The "invisble hand" is also based partially on the idea that vendors will see prices rising on a certain good--following in my example, oil--and then use that rise in prices to decide to make a larger investment in exploration, refineries, or alternative technologies. Similarly, markets that are created solely for predictive purposes such the "George Bush Re-Election" contract I mentioned above are extremely useful themselves.

How reliable is this?

Well, in 2004 the various gambling and predictive market sites had Bush re-election running at a slightly positive likelihood of happening for basically the entire year before the election. Only on election day did the initial exit polls (which were wrong, again) swing the prediction markets into a Kerry win for a little while before the evening came and they swung back to Bush.

In a fairly different use, the Google Blog discussed internal use of prediction markets as it relates to product release dates and other difficult to predict issues.

How does this impact my life?

Well, let's say that you are one of the many people who is interested in the 2008 elections. The following are tradesports contracts for some relevant events. They are $1 contracts, so note the scale on the left hand side of the graph. The closer to 1 that the graph is, the more likely the market thinks the event is to happen.

John McCain winning Republican Nomination for President in 2008

Price for 2008 Republican Pres Nominee(Others on Request) at

Well, that's the most likely event out of the group of candidates for that nomination according to tradesports.

Hillary Clinton winning Democrat Nomination for President in 2008

Price for 2008 Democratic Pres Nominee(Others on Request) at

Republican Candidate to Win in 2008

Price for 2008 US Presidential Election Winner at

Democrat Candidate to Win in 2008

Price for 2008 US Presidential Election Winner at

Field Candidate to Win in 2008

Price for 2008 US Presidential Election Winner at


Well, for the McCain supporters of the world, it looks like 2008 will be a good year. If you like Hillary, she is probably going to get the nomination but is not as likely to be president. Interesting stuff folks, interesting stuff.

People Involved: