Bastiat’s Bastions

For some years, I have been telling people in the Nicholls community about how markets are used to predict election results–and how they out-predict polls! Here is why markets, such as the Iowa Electronic Markets, work so well, compared to a single poll.  Just last year I wrote a blog post here on this subject, but now, the Iowa Electronic Markets are in the news (in my previous blog post I also discussed how markets could have been used to predict terrorist events, but this effort was squashed by our brilliant –not!–politicians in Washington).

Suppose you poll a very large group of potential voters about how they will vote in a few months (or days or weeks). You will get some results, but sometimes voters will not be very truthful. Some years ago, many voters who were supporting David Duke would not tell anyone, including pollsters, of their support for the former Klan leader.  Duke “under-polled” by a large margin.

The political markets are betting markets–good, old-fashioned wagers. Candidates’ “shares” get valued in the market between $0-$1, and the price reflects the market’s estimate of the candidate’s chances of success. If the bettor believes that candidate A has a better chance of winning than the candidate’s current price suggests, the bettor can buy up shares of candidate A, expecting to cash in after the votes are totaled.

A market makes forecasts much like the way the so-called “Delphi-technique” of forecasts(from Wikipedia) are made. The Delphi method takes the forecasts from experts and uses an average of middle value for the consensus.

There are several differences, though. Markets take middle values of forecasts, with each forecaster’s prediction weighted by the number of shares purchased. The more sure a forecaster gets, the more shares the forecaster buys. Forecasters select themselves instead of being picked by a researcher. Forecasters stand to gain if they make good predictions. and to lose if their predictions are lousy, and so, have a financial stake in the quality of their predictions. Traders who are especially good at forecasting, outpredicting the market, return to the market to play again. Those who lose, by being worse than the market at forecasting events, lose money and tend not to bet as much in the future, learning from their experience. The markets should just keep getting better at predicting as time goes by.

I should point out that sports betting lines and parimutuel betting odds are really determined in the same way and represent good forecasts of the outcomes of sporting events and horse races. By the same token, markets for oil and other depletable natural resources provide best forecasts of future scarcity of those resources.

While markets do better than polls, markets use information from polls, as the bettors incorporate poll information into their forecasts, and so, the polls end up being important in the process. Still, the markets pull more information together than what can be found in a single poll, and by using all information efficiently, weighing it for its reliability, markets provide the most accurate predictions of such future events.

-Morris Coats 

This entry was posted on Tuesday, October 30th, 2007 at 9:49 am and is filed under General, Political Economy, Sports. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.