Jan. 3, 2012 at 4:58 PM ET
Last updated 12:45 p.m. ET Jan. 4:
The pundits portrayed the Iowa GOP presidential caucuses as a tight three-way race, but market traders settled on an order of finish even before Tuesday's voting began. They had former Massachusetts Gov. Mitt Romney first, U.S. Rep. Ron Paul second, and former U.S. Sen. Rick Santorum third.
The contest turned out to be tighter than even the pundits expected: Romney led Santorum by just eight votes, with Paul not too far behind in third. But when it comes to the Republican nomination, Romney holds a far more commanding lead, at least for now.
Iowa's precinct-level party caucuses were convened to select delegates to go to the county-level conventions, but the real reason why GOP presidential candidates spent the past few months shuttling around the Hawkeye State was because this was the first real-world test of strength in an actual 2012 election. The same could be said for the pollsters and the prediction markets.
Most of the political handicapping you've heard about is based on traditional polling, which tallies up voters' preferences as expressed during telephone interviews. But for more than two decades, researchers have been experimenting with market techniques for sizing up political propositions — and so far, they've found that open market exchanges have been at least as accurate as the pollsters in predicting the results of elections.
Justin Wolfers, an economist at Penn's Wharton School who's been studying prediction markets for years, expected the accuracy record to hold for Tuesday's results, even though caucus outcomes are notoriously difficult to predict.
"All the usual ways we have for gathering information, including polls, are going to be much more screwy," he told me on Tuesday. "Someone's going to be surprised tomorrow. But ... empirical evidence suggests that prediction markets will be the least bad predictor."
How political markets work
Prediction markets let investors purchase "shares" in a particular proposition — for example, that Romney would finish either first or second in the Iowa caucus vote. Over the course of the campaign, they can sell off those shares and buy different ones, depending on how they gauge the candidates' chances. The price fluctuates based on supply and demand, but when the election occurs, the traders who backed the winning proposition get a set payoff. The traders who backed the wrong horse get nothing.
The idea is that markets provide a good vehicle for distilling the "wisdom of crowds," particularly when traders put their money or prestige behind their prediction. It's a lot like online gambling, but totally legal.
The nation's longest-running political prediction market happens to be headquartered in the Hawkeye State: Traders on the Iowa Electronic Markets buy and sell shares with real money, in hopes of reaping a $1-per-share return. Their account is limited to $500, so nobody makes a fortune. But the exercise is a valuable research tool for economists at the University of Iowa's Tippie College of Business, which manages the markets under a no-action arrangement with federal regulators.
Inkling Markets offers similar political trading, but with play money rather than real money at stake. And then there are offshore betting markets such as InTrade and Betfair. A website known as PredictWise.com aggregates the InTrade and Betfair statistics for political propositions.
On all these markets, the stated percentage or value for a given proposition reflects the probability that the proposition will come true. That's different from political polls, which reflect voter preferences in percentage terms. Thus, the numbers you see on the political markets don't exactly track the numbers you see in polling results. But the rankings on Tuesday afternoon were the same as what was seen in the polls: Romney first, followed by Paul and Santorum. Here's how the markets looked at 4:45 p.m. ET Tuesday afternoon, with 100 being the maximum value:
• Iowa Electronic Markets: 81.8 cents for Romney, 69.6 cents for Paul, 51.8 cents for "Rest of Field" (primarily Santorum). This is a market to predict who'll be in the top two.
• Inkling Markets: 73.2 percent for Romney, 15.3 for "Anyone Else" (primarily Paul), 10.1 for Santorum.
• Betfair: 48.3 percent for Romney, 29.4 for Paul, 23.1 for Santorum.
• InTrade: 49.3 percent for Romney, 25.6 for Paul, 24.2 for Santorum.
The projected outcome, as reflected in the markets, closely tracked what the polls were suggesting, Wolfers said. "The prediction markets are somewhat more bullish about Mitt Romney," he observed. If Romney didn't lead the pack, that would have been a surprise for the pundits — and for the traders as well.
There are a few additional twists to the trading. For example, much was made of Santorum's late surge in the polls, but the trading suggested that the surge hit its peak a few days ago and fell back somewhat.
After the top three, the GOP field's share values dropped dramatically. University of Iowa spokesman Tom Snee noted that within the past few days, traders had U.S. Rep. Michele Bachmann's shares at 0.2 cents, compared with businessman Herman Cain's 0.3 cents. "What they're saying is that Bachmann has less of a chance to win than someone who's not running anymore," Snee said. (Cain's shares have sunk since then, however.)
The big picture for the GOP nomination
For now, the traders on all the markets give Romney the overwhelming edge for the Republican nomination. The IEM had him at 81 cents on Tuesday, Inkling had him at 85.06 percent, and Betfair and InTrade had him at 74.4 and 79.8 percent, respectively.
If Romney faltered in Iowa, these prices could have changed overnight. Literally.
Joyce Berg, the IEM's director, noted that four years ago, Hillary Clinton was the market favorite before the Iowa caucuses, but was displaced by Barack Obama afterward. Clinton made an unexpected resurgence in the 2008 New Hampshire primary but eventually lost ground again to Obama. "What we're seeing is that markets are very good aggregators of news," Berg said. "In our markets, people appear to trade with their heads, not their hearts."
The IEM's traders are primed to respond quickly to the results of the caucuses, just as traders on the New York Stock Exchange respond quickly to the latest unemployment figures. So the fact that Romney has been on top for the past few months is no guarantee, in the political world or in the marketplace. "If he did not finish in the top two, that would be news," Berg said. "You would expect that to show up in the Republican convention market."
Wolfers said the fact that the GOP field was so wide in the run-up to the Iowa caucuses made this "one of the most unstable markets we've ever seen," in large part because the different voting constituencies didn't want to waste their votes on someone who wasn't perceived as a potential winner. Evangelical Christians, for example, might have wavered between several candidates based on the perception of their chances. Now there's a good chance that the evangelical vote will coalesce behind Santorum.
"It's not a standard two-horse race," Wolfers observed. "The dynamics when you have strategic voting are very, very volatile."
How do you think the Iowa caucus results compared with the pre-election predictions? Weigh in with your comments below.
Update for 5:45 p.m. ET Jan. 3: Mashable and Global Point Research are checking out whether positive sentiments expressed in Twitter tweets can be correlated with the Iowa caucus results. Analysis of data collected between Dec. 27 and 30 shows a big spike for Santorum, which would mesh with his surge of support in the polls and prediction markets.
Update for 9:28 p.m. ET Jan. 3: The betting on InTrade and Betfair shifted quickly toward Ron Paul, making him the betting favorite. Romney fell to No. 2, with Santorum rated No. 3. The order on the IEM and Inkling Markets was unchanged: Romney, Paul, Santorum. Inkling, meanwhile, closed its market for the night at 9 p.m. ET, so there'll be no chance for folks to change their prediction.
Update for 11:25 p.m. ET Jan. 3: By this time it was clear that Ron Paul will finish third, but Romney and Santorum were neck-and-neck for first and second. InTrade and Betfair favored Romney to squeak out a victory. The IEM's traders basically valued Romney's and Santorum's shares at the maximum price of $1, in recognition of the fact that they were both assured of finishing in the top two. It was possible to make a quick buck by betting on Santorum: Someone who invested $51.80 in 100 of his shares on Tuesday afternoon will get $100 on Wednesday.
Update for 3:40 a.m. ET Jan. 4: I've updated this report with the complete results from Iowa, headlined by Romney's eight-vote edge and Santorum's surprisingly strong surge.
Update for 12:45 p.m. ET Jan. 4: One of the night's other big winners was Justin Wolfers, the prediction-market expert from Penn's Wharton School. FiveThirtyEight.com's polling guru, Nate Silver, bet Wolfers a steak dinner that Santorum would be the top vote-getter in the Iowa caucuses, but Wolfers prevailed along with Romney by an eight-vote margin. Wolfers is a pescetarian who eats fish but not steak, so Silver will have to pay off with seafood. "Am I allowed to request a recount in Iowa if I have a seafood dinner riding on the outcome?" he asked in a Twitter tweet. (Sorry, Nate, there'll be no recount, even if seafood is at steak ... er, stake.)
More about prediction markets:
For the latest results from Iowa and their significance for the 2012 presidential campaign, check in with NBCPolitics.com.
Alan Boyle is msnbc.com's science editor. Connect with the Cosmic Log community by "liking" the log's Facebook page, following @b0yle on Twitter and adding the Cosmic Log page to your Google+ presence. You can also check out "The Case for Pluto," my book about the controversial dwarf planet and the search for new worlds.