Saturday, December 22, 2007

U of Iowa's electronic political/financial markets

This is a fascinating research project, using a financial instrument (futures) to predict outcomes in political markets. They've been doing this for a long time and their track record is impressive-- what one would expect if people are putting their money where their mouth is!

In a "vote share" market, the price of a candidate equates to the market's assessment of that candidate's expected share of the vote.

For example, as of today, the market predicts a four point victory for the Democratic candidate in November.

In a "winner-take-all" market, the price of a candidate equates to the market's assessment of that candidate's probability of victory.

For example, as of today, the market says that there is a 60% probability that a Democrat will win the presidency.

Those two markets are far more speculative since they're further in the future and conditional on the expected probabilities of given candidates winning the nominations in the highly contested races in each party.

More interesting, the market's assessment of each party's nomination outcomes...

Here's the historical graph/data for the Democrats and the Republicans looking back to March.

At present, Hillary has a 60% chance to win; Obama has 30%; Edwards and the rest of the field are 5% each. (The changes over time are also interesting. For example, Hillary's probability grew over the summer and has now faded somewhat.)

Romney has just passed Guiliani, 28-25%. Huckabee may have a bit over the last two weeks but is still at 18%. McCain has picked up the pace in recent weeks, getting to 16%. Ron Paul (presumably most/all of what now comprises "rest of field") has risen in the last two weeks to 9%. And Thompson has fallen spectacularly with a steady fade-- from leading for quite awhile before he was in the race and for a month afterwards to today's 4% probability.

Bookmark this. It's fun to check out every once in awhile!


At December 22, 2007 at 2:19 PM , Blogger Doug said...

I think it's pretty fun too, but it doesn't seem to have any predictive value. Mainly it just reflects the common wisdom on any given day. As I recall, Howard Dean was a prohibitive favorite about this time four years ago while Kerry was buried.

At December 22, 2007 at 3:25 PM , Blogger Eric Schansberg said...

Right. Part of that is the natural difference between predicting an event that is soon vs. future.

I don't recall the IEM's track record for any particular political event, but have a vague memory that it was a better predictor on the day before an election than the polls.

With the recent data, I'm surprised about the assessment that Paul is more probable than Thompson.


Post a Comment

Subscribe to Post Comments [Atom]

<< Home