Free market system can predict election results

A horse race, a trip to Atlantic City and the upcoming presidential election all share something in common, according to Wharton professors J. Scott Armstrong and Justin Wolfers.

Both Armstrong and Wolfers have developed models based on betting markets to forecast who will win the presidential election.

People bid on contracts for candidates, and the logic is that the amount people are willing to pay directly relates to the strength of their conviction that the candidate will win the election.

"Thus, these markets are sort of like a 'political stock market' in which you buy and sell shares" of President George W. Bush and Sen. John Kerry, "and the 'stock price' tells you something about which stock the market thinks is worth more -- and hence who is more likely to win the election," Wolfers said.

Apple pie and freedom aside, Wolfers said that this kind of model has been part of the American heritage for quite some time now.

"It's been happening since the days of Lincoln," he said. "You can go back to The New York Times about a century ago, and find that the pundits were fixated not on polls, which were yet to be invented, but instead on betting on the election."

Skipping ahead to 2004, an unusually intelligent parrot enters the mix. Armstrong's model can be found at www.pollyvote.com and features a combined forecast with an animated mascot known as Polly the Parrot running the show.

"Polly is an ideal collaborator," Armstrong said. "She has no political opinions about politics, and she listens to all objective forecasts. She then parrots back that information, thus providing one-stop shopping for those interested in who will win an election."

Polly has become quite popular. The Pollyvote Web site uses "Pollyisms" to explain political terminology and also displays "Pollygraphs," which are charts of the election standings.

"Her one skill that is unusual for parrots is that Polly can calculate averages," Armstrong said.

Wolfers said that using these betting models was "fairly natural" for him, citing his college job working for various sports bookies as well as his "betting blood."

"I'm an Australian, and we Aussies love to gamble," he said. "It has been said that we'll bet on two flies walking up a wall, and surely betting on elections is more interesting than that."

Wolfers is confident that his market approach will be fairly accurate.

"I would say that the simplest and easily most accurate way of predicting the likelihood of who will win this election is to simply log on to www.tradesports.com and look at the price of the 'Bush Contract,'" he said. "At present, this suggests a 58 percent chance of Bush winning and a 42 percent chance of John Kerry winning."

"The efficient markets hypothesis tells us that we should expect financial market prices to reflect all publicly available information," Wolfers added.

Armstrong is also willing to put his money where his parrot is.

"My expectation is that the polls will be least accurate, while the betting market will be among the most accurate," he said.

Still, as in all betting, it is important for people to know what they are getting into.

"Warning: Once you start tracking the election through the lens of the betting markets, it becomes very addictive," Wolfers said. "A fair few of the faculty now check the online markets a couple of times a day to get a sense of how the race is evolving ... and it's a nail-biter."

Whatever the outcome of the election, the losing candidate will not be alone in his predicament.

"Unfortunately, [Polly] will be out of work in November," Armstrong said.

Please note All comments are eligible for publication in The Daily Pennsylvanian.