Iowa Electronic Markets predict presidential winner

If you want to know who will win the upcoming Presidential election, and who will control the House and the Senate, don't follow polls, follow the money.

I'm not talking about following the twisted trail of campaign contributions. Although this may yield some interesting information, I'm saying that you should see how people are betting the election will turn out. And look at people who are betting with real money.

I suppose there are places in the world where you can really bet on the outcome of the election. In the US, I'd speculate that it's illegal to bet on the outcome of an election -- at least I'd hope that's how things are. Allowing betting on an election opens up too many opportunities for fraud.

But there is one place where you can bet -- err invest -- a modest amount ($500) of money on the outcome of this or many different elections -- The Iowa Electronic Markets.

The IEM has been forecasting Presidential elections for the past 16 years, and it has been doing a very good job of forecasting winners. One study referred to on the IEM's site says that a selection of polls has yielded an average error of 1.9 percent over the years, while the IEM results have had an average error of about 1.5 percent.

Why does the IEM work better than polls? One obvious reason is that it's so easy to lie to a poll taker. Another is that polls are easily slanted to show whatever the poll taker wants.

Q. "Will you vote for the smart, photogenic candidate, or the dumb, ugly one?"

A. "Smart and photogenic, of course." (Actually, not, in a million years but I'm not telling you that...)

Also, a lot of polls aren't based on good sample populations. For example, even the valuable polls are suspect since they are based on a self-selected population -- and you can vote more than once if you really want to.

But when it comes to money, things change. Sure, I may like the smart, photogenic candidate, but if I look at the odds give in the IEM, and I think I can make money by "betting" on the dumb ugly one, I'll forget what the polltaker thinks and bet on the dumb ugly one.

And the crucial thing is that this applies to everyone else. Sure, you may be very loyal and continually bet heavily on your candidate, but there enough greedy people out there who can swamp even the largest band of loyal partisans.

So although the IEM thus provides a good forum to let the "intelligence of the masses" cast it's vote, the IEM isn't perfect. For one, the market is fairly small. From what I could tell, there were no more than a thousand participants.

Another is that the market isn't "perfect". For example, in the Bush-Kerry presidential race, I saw a few cases where two different markets (vote-share and winner-take-all) predicted two outcomes. This smells like an obvious arbitrage opportunity. However, as far as I understood, it was impossible to short shares in either market. Also, the payoffs in the two markets were so dissimilar that it was really impossible to effectively arbitrage the two markets with the $500 limited investment. That's another "imperfection". Participants are limited to speculate with only $500.

But even with these limitations, I would say the IEM is a much better predictor of an election winner than polls. And the IEM is not only a useful predictor of elections. It's also useful to predict such things as Federal Reserve policy, Microsoft share prices, or Google IPO values.

There are also similar, more practical alternative futures markets. A small investor can also speculate (or if you prefer hedge risk) with a or I'll cover these in an upcoming column.