So a friend posted a link to this blog post over on the Freakonomics blog: McCain Pulls Ahead Where It Really Matters
He had mentioned at dinner that night that Senator McCain had, for the first time, traded at a higher price than Senator Obama was trading. I didn't think anything of it at the time. But then I got home and read the actual blog post.
Two things in the post set me off. One, was the title of the post - "McCain Pulls Ahead Where It Really Matters"; and, two, was a comment in the post - "This is news.".
I disagreed - vehemently - with both assertions.
InTrade is NOT "where it really matters". Where it "really matters" is the results in each individual state. Where it "really matters" is the electoral college. Where it "really matters" is the voters, and the ground game, and the debates, and fifty other things - but InTrade is definitely NOT where it matters.
The national polls? Don't matter. Presidential elections aren't about the popular vote. InTrade prices? Don't matter. They fluctuate, in an extremely reactive manner, mostly to the current state polling.
If you want something with predictive value look at pollster.com or even electoral-vote.com. Both seem to me to have better predictive value. Even better, take a look at fivethirtyeight.com. One of their stated goals is coming up with some sort of paradigm for looking at information and using that to accurately predict who will win the election in the fall. What they want to do is correctly predict the outcome as far in the future as possible.
So I replied to his posting by saying "I'm trying to remember the last time InTrade decided a presidential election, and I'm drawing a blank." My friend replied in turn by posting a link to an article from pollingreport.com that contained this quote: "Indeed, by 2004 the Intrade market model went stratospheric in predictive accuracy as the market favorite won the electoral votes of every single state in that year’s U.S. presidential election."
The pollingreport.com article doesn't reference any actual data, so (despite my reluctance) I'm going to have to believe that at some point before the 2004 election the InTrade market favorite for each state was the winner of that state's electoral votes. But when was that person the "favorite"? The day before the election? The day of the election? And what does that have to do with the predictive value of the InTrade market right now? If you looked at the InTrade market on January 1st, 2008, was it predicting an Obama-McCain contest for president?
Here's the data:
http://data.intrade.com/...
http://data.intrade.com/...
http://data.intrade.com/...
http://data.intrade.com/...
I think you'll find out the answer pretty quickly.
I think if you look at polls on a random date at some point a long way out from the election, sometimes they'll be right and sometimes they'll be wrong. And same thing with the "prediction markets". But they aren't "predictive". They are a snapshot in time.
If I was going to say anything was predictive, I would say it's state by state polling. And, even that is pretty unreliable this far out. Yes, in some cases they are absolutely correct. I'm pretty sure both state polling and InTrade are correctly predicting the outcome in the presidential election for the state of Alabama. And, yes, they both become more accurate the closer you get to the elections. Furthermore, there are contentions that "predictive markets" "outperform" polling - especially the closer you get to an election. But you know what? They just closed trading on "Hurricane Ike to make first Landfall as Category 2 or Higher in State X" too. It was extremely predictive the night of September 12th. On September 7th, not so much.
In my opinion, the Freakonomics guys are pushing a meme. That's all it is. It has no basis in reality, and the fact that they are pushing it in contravention to actual facts tends to piss me off.
But if you look here:
http://electoralmap.net/...
you'll find data that is in opposition to the Freakonomics post.
Is it relevant?
Well, I tend to like it!! But I don't think it's predictive.
A lot of things are going to happen in the next 51 days. I'm sure I won't like some of them. And some of them I may. But between the debates, and the commercials, and the speeches, and the interviews, and the gaffes, and too many editorials, and way too many posts to read, I'm sure we'll eventually find out who the president is going to be. I just don't think InTrade will EVER be "where it really matters". And I don't think it will ever be "news".