By Fester:
I have some significant issues with political prediction markets in that they are good aggregators of conventional wisdom among a small, illiquid, and thinly trading sector of actors. They are a useful tool, a one-stop aggregator of opinion, but they are not infallible and they are not fully efficient. They should be a complement to polling, demographic and issue analysis and other political analytical techniques.
For instance, the AP called Obama clinched the nomination early yesterday afternoon. Delegate counters had projected that even in worse case scenario situations --- double digit losses for Obama in both Montana and South Dakota, he would pick up a sufficient number of pledged delegates to push him over the top once they were added to the new superdelegate committments he was getting. And these projections were occurring early yesterday or on Monday. No one was saying anything new as we knew this was a highly likely scenario since at least March 5 when Ohio, Texas, Vermont and Rhode Island ended up as a net delegate wash for Clinton.
And yet, as Matt Yglesias notes, the Intrade prediction market still was giving someone other than Obama a more than 5% probability of being the nominee. Given his age & health, death or disability is a very low probability event so the mechanism would most likely be a political mechanism of several hundred Democratic superdelegates flipping in the next eight weeks. I think that is amazingly unlikely. I think 5% is an order of magnitude too high to describe that probability.
An efficient market would arbitage that problem away, but these markets are too thin and unsophisticated to do that.
No comments:
Post a Comment