Chris Masse has posted a long and very thoughtful analysis of the recent election of Cardinal Ratzinger to the Papacy. He is appropriately critical of journalists and prediction-market promoters who now claim that the markets accurately predicted Ratzinger’s election. In fact the markets were almost useless, since meaningful public information was not available.
Chris also points, in Section 5 of his essay, to a string of other prediction-market failures, and argues that they too were difficult calls because of limited public information. I think that he is right; prediction markets aren’t crystal balls. At best they distill public information into the most accurate predictions possible. However, when information isn’t publicly available, the best prediction may be no prediction.
He also points out, almost in passing (Section 7), that current prediction-market exchanges make it bizarrely difficult for outsiders to obtain price data, even for expired markets — a sure sign, IMO, that the people running these markets do not fully understand their business (cf. eBay, which is similarly clueless).
And he makes some excellent points about contract design, particularly WRT to contracts having excessively long duration. That’s exactly right. For example, instead of having several contracts with long duration, each predicting the price of gold within a particular price range, there should be fewer contracts having shorter durations, with prediction ranges that are closer to current price levels — like traditional exchange-traded futures.
One of his suggestions that I do not find compelling is his recommendation (Section 8) to create a “think tank” of economists to evaluate specific prediction markets. I think the market will probably take care of such evaluations, as the exchanges learn with time how to create contracts that customers will be eager to trade. The exchanges have a big financial incentive to do that, after all.
But that is a quibble. Chris’s post is excellent and perhaps will spur some useful discussion about the evolving prediction-markets industry.
UPDATE: Chris has responded to my comments, in an addendum near the bottom of his post.
Unfortunately he does not provide a direct link to his response. However, it’s easy to find if you scroll down to the bottom of his post and look for the update dated 9/21. [Fixed by Chris, thanks!]
I may not have been clear enough in my post. I meant that the exchanges have a strong financial incentive to experiment with different contract configurations. My own trading experience suggest that a futures contract that expires in 1 year is likely to be much less interesting to traders than would be a similar contract that expired in three months. Most futures traders prefer to trade contracts whose expirations are a few months away as opposed to a full year or more away. The way for an exchange to cover a long time horizon is not to create only long-term contracts, but to create a series of relatively short-term contracts with staggered expiration dates — as is done in conventional futures markets. I suspect the prediction-market exchanges will figure this out for themselves, by trial and error, and probably do not need outside “experts” to help them.