Discussion about this post

User's avatar
Overcoming Bias Commenter's avatar

One last point: I will note that, in spite of the amount of money riding on this last election, the market on TradeSports was leaning heavily against the Democrats taking the Senate almost until the end. In other words, the prediction market was wrong. I've seen the markets fail in other instances, too.

One assumption we have made throughout this entire discussion is that prediction markets are high quality sources of information. We don't actually know that. We have been assuming that.

I would argue that, in fact, they are at best good quality ways of combining current available information, and even then only assuming the Efficient Market Hypothesis is true, and I have my doubts that the EMH is more than "largely" true.

The markets may be able to tell me what "we all know", but I suspect they aren't very good at telling us what no one knows, and that is often much more important. They are also not very good at telling you when they are combining good information versus when they are just combining everyone's bad information -- it is not obvious how to distinguish an interesting prediction from mere combination of lots of limited (and incorrect) information. We need a lot more studies, preferably with data from the fairly large betting markets that now exist.

Expand full comment
Overcoming Bias Commenter's avatar

There exist examples of prediction market prices which are widely used, accepted and reasonably well understood. Every time the news announcer says, "The price of oil rose two dollars today on fears of middle east unrest", he is quoting the result of a prediction market. There's no reason he couldn't say, "The odds of President Bush's impeachment rose 10% today after a new round of cricism of the president's policies." In fact I noted the betting markets being mentioned quite a bit this past election cycle, not quite yet hitting the major media but often mentioned in blogs and occasionally in newspaper articles.

Here's a cool idea: imagine if the Democratic or Republican parties ran their own prediction markets to see who among their main candidates has the best odds of winning the 2008 presidential election? It could help the party by reducing the impact of extremists in the early days of the election and giving candidates a boost who have the best chance in the fall primary.

Expand full comment
11 more comments...

No posts