Intrade: Not so Hot after all

September 18, 2008

Back in the early days of the Despot, I argued that Intrade was the place to go for election predictions:

President McCain futures are trading at 43.7 over at Intrade at the moment. It would be a stretch to call guessing about politics an efficient market at this point, but as a rule it’s always better to listen to people who are paid to be right than to people who are paid to say clever and interesting things. As far as I know, Intrade is the best game in town for the former.

I still believe that the principle here is correct. Certainly Las Vegas is far more reliable than ESPN when it comes to predicting the outcomes of sporting events, and anyone foolish enough to take stock tips from magazines should think again. The inefficiency of this particular market, however, is pretty staggering. This shouldn’t be all that surprising, as a market in which Americans cannot legally invest has a serious handicap when it comes to predicting U.S. elections. As it turns out, it’s fairly easy to see the nature of that inefficiency by looking at the presidential election markets over the past few weeks.

The basic problem is that Intrade’s Obama/McCain futures are far too closely tied to polling trends. Here is a graph of the Obama market over the past month:

Note that the last day or so isn’t reflected here (I’m not sure why). Obama to win is currently trading at 51.5. Now, here is a (hideous, blown up) graph of gallup polling over roughly the same period:

Finally, here is a projection from the folks at TNR, posted over at 538, of how the two convention bounces would affect tracking polls over this period, made back on August 15th:

So, what can we take away from this? First, the projection didn’t nail the size of the two bounces exactly, but it’s very close. The basic shape is identical, and they even did a fairly good job of predicting when Obama would retake a small lead.

Secondly, the price of Obama-to-win futures generally moved in step with the polls. It is difficult to compare the scales, since the futures price is essentially guessing the probability of Obama winning – by any margin – while the poll tracks how much he will win or lose by. Roughly speaking, though, we have a flatter version of the same curve; Obama went from a 60% favorite to a marginal underdog, and has just now bounced back to become a marginal favorite, with his price still rising. This is pretty much what the better trend lines have shown. Sites like 538 compile a lot of polling data, and massage the data so as to downplay short-term swings. But while sudden spikes based on the news cycle of a given day are eliminated, convention bounces aren’t – and shouldn’t be.

A convention bounce lasts much longer than most meaningless fluctuations. To put it differently, if there were a spike in a candidate’s polling of the size and duration generally seen after a convention at some other time, it would be very, very significant. So a trendline that didn’t register a convention-size bounce would be broken: the algorithm doesn’t know anything about conventions.

The same should not apply to a futures market. An efficient market would set the price of these futures taking all relevant, publicly available information into account. Polling data is incredibly important information, but it isn’t the only information. For instance, polling data, no matter how well it is finessed, will follow a very predictable pattern after conventions, hence the New Republic’s ability to predict that pattern. But behaving predictably is something a functioning market simply can’t do. So this market is broken.

Having thus opened the door for the idea that the well-informed observer can predict election outcomes better than the best (that I know of) existing market, I will follow up later today or tomorrow with my own attempt to do just that. (SPOILER ALERT: Obama’s going to win.)

Advertisements

One Response to “Intrade: Not so Hot after all”


  1. […] polls sent Obama’s chances of winning the electoral college vote soaring from 45% to 61.2%. As was discussed yesterday, no rational observer would actually have taken that first number to heart; McCain’s polling […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: