PredictIt aims to find out if political bettors could be more accurate than pollsters
Last week's Iowa caucuses marked the beginning of the U.S. presidential nomination process, and with it came a flood of media reports about the momentum of candidates, predictions about which candidate would be next to exit, and who would eventually win their party's nomination.
It also brought about a frenzy of political futures trading on PredictIt, a political prediction market. As results came in last Monday, traders who had bet big on Republican favorite Donald Trump started dumping shares in an attempt to limit their losses. Meanwhile, on the Democratic side, the share price of a Bernie Sanders victory was all over the map. As early results came in, it appeared that Hillary Clinton would win as expected. Then Sanders started surging, and his shares got as high as 48¢ before plummeting down to 1¢ as he eventually lost by a razor-thin margin.
How it works: There is a "Yes" and a "No" side of each market. For instance, in the "Will Donald Trump win the Republican nomination for president?" market, bettors can buy shares in either "Yes" or "No." (As of 10:30 a.m. on Feb. 8, Yes shares were selling for 39¢ and No shares were selling for 63¢.) When the nomination is decided, the value of shares for the winning side increases to $1, while shares in the losing side become worthless.
Selling shares: You don't have to hold the shares until the election is decided to cash out. Many bettors try to time the market by buying low and selling high. For instance, shares in Marco Rubio to win the Republican nomination jumped from 30¢ before the Iowa Caucuses to as high as 55¢ last week. A trader who had shares in Rubio to win could have nearly doubled his money by selling those shares if he believed the markets were overestimating Rubio's chances.
What does it cost?: Every time you make a profitable trade, PredictIt takes a 10% cut of the profit. So a trader who buys 100 shares of Bernie Sanders to win New Hampshire's Democratic primary at 95¢ for $95 would see shares increase in value to $100 if he does in fact win the primary. PredictIt would then take a $.50 commission, and the trader would be credited with $99.50 to his account. The site also takes another 5% of any profits when traders withdraw money from the site. Of course there's also the risk that the value of your shares will decrease, and shares on the losing side become worthless.
Why an $850 limit per market?: PredictIt modeled itself after the Iowa Electronic Market, which received a no-action letter from the CFTC in 1992, granting permission for individuals to invest up to $500 per market. The $850 limit was determined based on an adjustment for inflation over 22 years when PredictIt requested the same treatment from the CFTC in 2014.
Demographics: "Most of our users are between the ages of 21 and 39, and the majority are male," says Aristotle Chief Marketing Officer Brandi Travis. "We've got some that are 18-20, and we've got 80-year-olds, using it. But that's the majority of our traders."
Typical deposit: "Our average deposit is about $100," says Travis. "People put their money in and they start trading little by little, but once they make their second deposit, they tend to put more in."
PredictIt has more than 60,000 users, with 19,000 currently active on the site. That number is bound to grow as election season picks up and people learn more about the site through media reports and social media. And just like most other forms of gambling, the prospect of betting on elections is gaining more mainstream support. In 2014, Nevada State Senator Richard Segerblom introduced legislation that would allow Nevada sportsbooks to take wagers on political elections, something South Point Casino bookmaker Jimmy Vaccaro has been advocating for years.
If you're wondering how it's possible to make real-money wagers on the outcome of political events in the U.S. and not run afoul of federal gambling laws — especially since Nevada sportsbooks are banned from accepting those types of bets — you're not alone. Brandi Travis, the chief marketing officer for Aristotle, a political software and data company that services PredictIt, says that user concerns about legality were very common when the site launched in October 2014. But as the site has grown and its markets have been featured in mainstream media reports, those questions have become much less frequent.
Victoria University of Wellington in New Zealand is using PredictIt as an academic exercise; it proposed the "creation of a small-scale, not-for-profit, online market for event contracts in the U.S. for educational purposes." It received a no-action letter from the U.S. Commodity Futures Trading Commission, meaning that the CFTC would not seek to shut down the site if it operated under certain parameters.
The site is modeled after the Iowa Electronic Markets, which have allowed political wonks to trade election results futures since 1992. Intrade, a site that became quite popular over the last two presidential elections, also traded political futures, but was shut down in 2012 because it did not ask the CFTC for a no-action letter.
Many economists and political pundits believe that prediction markets are more accurate than polls, though the evidence isn't exactly clear. In 2008, Intrade's presidential election markets indicated a toss-up was likely when all signs indicated that Barack Obama had at least a 60% chance of winning the election. Nate Silver suggested it was the work of "some idiot degenerate gambler," but the ability to manipulate markets was a huge weakness at Intrade.
(PredictIt doesn't have the same vulnerability, due to an $850 individual maximum on any particular market.)
Robert S. Erickson, a professor of political science at Columbia University, and Christopher Wlezian, a professor of government at Temple University, wrote in 2009 that prediction markets in the early 1900s were remarkably accurate — but over the last 30-40 years, polls have had an outsized impact on the prediction markets, making them less effective when the polling is off. (Wlezian is now the Hogg Professor of Government at the University of Texas.) Results from last week's Iowa Caucuses seem to support that conclusion; with Donald Trump and Hillary Clinton holding leads in the polls in Iowa, both held big leads in the PredictIt markets for Iowa as well. Clinton did squeak out a win, but had the markets been accurate, the price on Bernie Sanders would have equaled Clinton's. On the Republican side, Ted Cruz defeated Trump, despite the prediction markets giving Trump roughly a 70% chance of winning.
"Although the performance of these new infant markets may not always live up to their promise, the future may be brighter," wrote Erickson and Wlezian in a presentation for the Annual Meeting of the American Association for Public Opinion Research in 2009. "With growing interest and higher volume of trading, the performance of current trading markets may surpass that of its ancestral forebearers from the pre-poll era. By theory, the promise is that election markets, informed by polls plus other information, should perform better than polls alone."
Whether theory ends up becoming reality in 2016 remains to be seen, but one thing is for sure: Political bettors at PredictIt will have plenty of opportunities to be right (or wrong) in 2016. There are markets for all the upcoming primary states. There are markets for who will win each party's nomination. There are presidential general election markets. There are even markets on whether there will be a federal minimum wage hike in 2016, what President Obama's approval rating will be at the end of the week, and whether Caitlyn Jenner will speak at either party's convention this summer.
"We're adding new markets almost every day," says Travis. "We're starting to add more international markets, as well. I would definitely keep track of the international markets, because we've been incredibly accurate on those."
For a humorous take on betting on politics, check out the video below from The Daily Show's Jan. 19, 2016 episode, featuring an appearance by Vaccaro.