The prediction market is said to be more accurate than opinion polls. Kevin Davie investigates.
I thought I’d buy Obama. Not the actual person, you understand, as I am sure he is not for sale, but his chances of becoming the next president of the United States.
This is known as a prediction market and there are celebrated examples. You buy a contract, say for 60c and should your man get the top job, you get a $1 payout.
Prediction markets are growing in favour and are used to predict all manner of outcomes. They are also increasingly being used as a business tool by companies such as HP and Google.
These markets are claimed by their supporters to be highly accurate predictors. Detractors say that they are no better in forecasting than any method that samples the wisdom of many.
They certainly appear superior, though, to opinion polls. One leading prediction market has significantly out-performed opinion polls in determining the outcome of the fierce contest between Barack Obama and Hillary Clinton.
Blogsite Caveat Bettor compared the performance of pollster Zogby with Intrade, a prediction market.
It found that Intrade was right in seven cases and wrong in three and Zogby wrong in seven and right in three. Both correctly predicted the outcome in 11 states.
Prediction markets are better predictors, it is suggested, because real money is involved.
The polls would have you believe that Obama and John McCain, the two presumptive nominees are within a few percentage points of each other. Top polls such as Gallup give a 48% chance to Obama versus 42% to McCain.
This is despite Obama usually pulling crowds of say 18 000 (a small event will draw 2 600), while McCain’s typical audience is a mere 600, the Wall Street Journal reported.
The fact that you pitch to sample some of Obama’s star appeal does not mean you will vote for him, of course, but the opinion polls are surprisingly close given that Obama has superstar status and McCain none.
Intrade, however, shows quite a different picture to the polls. At mid-week his ruling price was 61c. McCain was trailing badly at a cheap 34,5c.
But how do the prediction markets agree with one another? If they are as accurate as their supporters claim, you’d expect them to show similar pricing.
Intrade is Dublin-based and betting is illegal in the US, but the Iowa Electronic Market (IEM) at the University of Iowa has special permission from the authorities to operate its prediction market.
The two markets are only a click away on the internet. Both offer Obama options at 61c, but the IEM prices McCain at 39c compared with Intrade’s 34,5c. There is no obvious reason for this difference.
Intrade offers trades on everything from the weather to the price of fuel and whether Osama bin Laden will be captured or neutralised.
Contracts can be bought for political events in Australia, Canada, Germany, India, the United Kingdom, New Zealand and India, but not South Africa.
I checked the Yellow Pages and called Sportsbet, a Jo’burg bookmaker. Turns out it takes bets on Obama, but not Zuma. You can buy Obama to win from Sportsbet for R10 to get R15 back. This means investing 66 US cents to win $1, not too dissimilar to what you’d pay at Intrade. A bet on McCain costs R10 to get R28 back, about 30 US cents to make a dollar.
Prediction markets are a form of crowdsourcing, a hot new term that refers to using new media tools to source the wisdom of many. I canvassed my colleagues at the Mail & Guardian both verbally and by email to see if I could establish a market price for Zuma futures.
There appears to be no consensus on the Zuma price. Most respondents were bunched in two groups of similar size, one at 50c and the other at 95c. A few hovered at the 65c to 75c level.
My sense is that the middle price would be about 75c, which is quite low, actually, since unlike Obama he has no apparent competitor for the top job.
For Mike Lemon of Sportstrade, the Zuma bet would be “deep in the red”, meaning you’d have to bet a lot to get a small return but, he says things are so changeable here that you cannot be sure of the outcome.