As we move through election season, the polls start coming fast and furious; the pundits punditorate, the politicians spin, the news media pretend not to spin, bloggers blog, and everybody offers the inside scoop as to the outcome of the election. How is one to aggregate all this information into the best possible guess as to who will win the presidential election?
One excellent way to do it is to pay people to be right. This is exactly what is done at the Iowa Electronic Markets. If you are sure that John Kerry is going to be the next U.S. President, you can got to the IEM website and buy Kerry stock. Currently, for 46 cents one can buy an option to get $1 if Kerry wins. Bush costs about 54 cents. If you are sure Kerry is the man, this is an easy way to make money, $46 now will get you $100 in just a few months.
In fact, ignoring some niceties, most people should probably be willing to buy Kerry stock if they think the chance of him winning is more than 46%. The same goes for Bush. If people do start buying a stock, the price rises until one reaches the point where once again, as many buyers think Kerry has at least those odds of winning as sellers think he doesn’t. Thus the stock price is the probability of winning.
The market aggregates information based on your willingness to put money down on the proposition. This is not a poll. It is a probability based on the best guess of the investors. It lets the investors think through the punditry and the polls and the spin. And it rewards people who are right. As we move through the election, IEM should be high on your list of places to get info.
For myself, I view this as not much better than gambling, so I don’t particularly advise those opposed to gambling to dig in. But the numbers are useful nonetheless. If you want the best available guess, this is probably it. If you don’t think so, I know how you can make some money!
Here’s a graph of the last two months of prices. Kerry has been on top a couple times, but for most of the time, Bush has a slight lead. Also, this form of information gathering can be used for pretty much any question. The Pentagon got hammered a while back for contemplating using this method to figure out the probability of various bad things happening.
You can play for free at Newsfutures, although you can’t win any money either (obviously).
I played during the primary season — It was fun.
IEM has an impressive track record.
A version of the DARPA terror exchange is online here.
You can play for free at Newsfutures, although you can’t win any money either (obviously).
I played during the primary season — It was fun.
IEM has an impressive track record.
A version of the DARPA terror exchange is online here.
Interesting to note that Kerry’s lead evaporated right at the end of the Demo convention — right when he was getting the most amount of exposure.
TradeSports.com is another great resource to follow the politcal futures market. What I like best about TradeSports is their more obscure futures, like who will win the popular vote, which party will win the House and Senate, and the color of the terror alert code on election day.
Last fall, on the morning of California’s gubernatorial recall election, I scanned a half dozen newspapers for final polling results. Each media outlet was reporting that the final polling made the race too close to call. (In case you’ve forgotten, a slew of allegations about Arnold’s loose hands on movie sets arose five days prior to the election, casting doubt on previous polls showing an Arnold victory.)
But when I turned to Tradesports to see what those speaking with their money had to say, they obviously knew better — Schwarzenegger was trading at 95, Davis at 5.
I’d heard of Tradesports but never looked at it. It looks like a nice site. IEM also does other political markets, but it looks like Tradesports has far more. Of course, both sites should give the same predictions or you can make some really easy money off arbitrage.
So I’m thinking of the Church applications here. We do a futures market in number of converts active in one year and use the resulting info to adjust our missionary efforts?
We predict the date of the Second Coming?
I’m mostly in jest, but surely there’s some way we can think of that the untapped knowledge of the Saints ought to be tapped.
I think the IEM’s record is no better than any national survey poll. There’s a short article on this in the Atlantic this month (p. 35). Long story short, extrapolating from IEM results in July doesn’t do you any better than extrapolating from a July poll. As the race nears the finish line, all these predictors will get closer to being right.
Great, now y’all have given me another hobby. Just wait ’till my wife hears about this one…
John,
What does it mean to extrapolate from an opinion poll? Just take whoever wins in the poll and call it good? Since we’ve only had about 3 presidential elections, and each election is binary, I would be surprised if there was much the July extrapolation method could tell us.
I haven’t seen the Atlantic Monthly article, but I think the IEM has a paper on their site. I doubt it is much better.
Frank, all I’m saying is that the IEM is interesting, but is certainly not more reliable than a regular poll.
John,
I looked at the Atlantic Monthly sidebar. They compare July 1st prices to actual outcome vote percentages for the various candidates since 1988. This is misguided:
The biggest and most grievous flaw is that prices do not, repeat, do not, reflect outcome vote percentages. The IEM is the probability that the person has the highest number of votes, it is not an estimate of how many votes they have.
Second, they are absolutely correct to point out that the IEM does not reveal in advance who will win. Between July and November, most of the campaigning occurs. The IEM gives the best guess, better than any particular pundit or poll. But that does not mean that the best guess will be right. Fortunately, the IEM also gives the probability of each outcome. So even though the best guess is a Bush victory, there is still a huge chance of a Kerry win (46%). If you are sure they are wrong, you can make some good money! If you think the polls provide indiependently useful information, you can make some money!
Third, the sidebar does not actually compare the data with any polls taken in July. So by what measure should we judge this method no better than the polls?
Fourth, suppose we had a poll and the IEM. The poll tells us who the sample would vote for today. But this is not really the question we wish to answer. We want to know who will win. For that we need to know voter turnout, how voters vote, by state, in November. The pollsters try to measure “likely voters”, but never in a trnasparent way. They do polls in individual states, but how should I evaluate the relative probabilities?
The IEM figures all the above in and produces a nice clean probability. So, for example, the Democratic convention gave Kerry a boost in the polls. But candidates always get a boost in the polls post-convention, so this boost should not give us any new info unless it is atypical. In this case, it was too low, and so the odds of Kerry winning fell on the IEM. Poll followers would think Kerry’s odds just went up. The best guess is they really went down. This is because polls measure now, the IEM is focused on the election outcome.