What if, with just two weeks to go until the presidential election — and the media obsessively following and deconstructing every number in every poll and every other metric to determine who might win — someone decided to mess with the numbers?
Intrade is an online prediction market where people can bet on the likelihood of various outcomes by buying shares in that outcome. For instance, one could buy shares in “Barack Obama to be re-elected president in 2012.” If, on Nov. 6, Obama won, that person would make money. If Mitt Romney won that person would lose money, but other bettors who bought “Mitt Romney to be elected president in 2012” would make money. As with the stock market, the more people buy shares in an outcome the more the price of those shares rises and, by Intrade’s metrics, so too the probability of that event happening.
Rigging the polls isn’t an option, but Intrade is more malleable. It may not be as closely watched as the polls, but it’s still a metric that reporters use, and someone with the time, money and inclination could push the numbers one way or the other to try to create a sense of momentum or a sense of failure.
Intrade forbids “Any attempt by you to interrupt the operation of Intrade, disrupt the operation of Intrade website, to manipulate Intrade market prices or act in a manner that may be injurious to Intrade” and doing so “may result in the immediate closure of your account, and/or unwinding of your trades and/or being prevented from accessing all or any part of Intrade or the website in future.”
But that hasn’t necessarily stopped people in the past.
In 2004, recalls Eric Zitzewitz, an associate professor of economics at Dartmouth College who has co-written several papers on prediction markets, “somebody, in the middle of the night, drove the price of the [George W.] Bush re-elect contract – which was probably trading about 60 at the time – down to 15.”
In 2007, he recalled, someone attempted to run up the odds of Hilary Clinton winning the presidency, buying up contracts until the odds of her being the next president were listed at 40 percent.
The most reported instance of manipulation was one month before the 2008 elections when several traders and reporters noticed price fluctuations of Sen. “John McCain to win the presidential election” contracts that seemed to be the result of someone trying to artificially inflate his odds of winning – buying up lots of “McCain to win” contracts while simultaneously selling lots of “Obama to win” contracts, resulting in the odds of McCain winning looking much better and those of Obama winning looking much worse.
The problem with all of those instances (as a model for our hypothetical manipulator) was that they moved the Intrade odds drastically over a very short period of time, and everybody noticed.
The 2004 manipulation was “particularly ham-fisted,” Zitzewitz said. It was so obviously unrealistic that other traders caught on and began exploiting the manipulation to make money, buying up the shares at the artificially deflated prices, raising the price, and selling them at the higher price. Essentially, the manipulator gave a lot of money to a lot of people.
Zitzewitz recalled: “You could look at the [Intrade precursor] Tradesports forum at the time, and you could see all these quotes from frustrated people where they were saying, you know, ‘I bought all I could, but my account’s tapped out and it’s going to take a little while to get more money in, so, you know, the rest of you guys, please profit from this,’ from this fool who’s attempting to manipulate the market.”
“I think it took about 24 hours for the price to come all the way back,” Zitzewitz said.
The 2007 attempt was “a little bit less obvious,” but still evident within the larger context. At the time, the odds of Democrats winning the primary were just over fifty percent, and the odds of Clinton winning the primary were lower. Taken all together, Zitzewitz explained, the numbers didn’t jive – they showed that Democrats would have “an over 80 percent chance of winning” if they nominated her, and that if they nominated anyone else, they would have less than 20 percent chance of winning.
The McCain manipulation was also discordant with related numbers, as well as the odds for the same events on other futures trading sites like Betfair.
The matter was heavily discussed on Intrade forums, and the company conducted an investigation into the matter. The now-late founder of Intrade, John Delaney, concluded that the fluctuations were not the result of someone trying to rig the system.
“The trading that caused the unusual price movements and discrepancies was principally due to a single ‘institutional’ member on Intrade,” Delaney explained on his blog. Intrade had contacted the company involved, and established that they were “using our markets in good faith” as a means of “[managing] certain risks.”
The lesson to be taken from all of this, Zitzewitz said, is that market manipulation would have to occur gradually. Dramatic changes in price tip off other investors who, then, trade against the exaggerated odds and drive the prices back down to normal.
To sustain an altered price for longer than 24 hours would also require a huge amount of money, said David Rothschild, an economist at Microsoft Research New York City and manager of PredictWise — a website that makes predictions in part based on futures markets.
Moreover, a successful manipulation would have to move the numbers not only on Intrade, but also those on other futures trading platforms like Betfair. Rothschild explained that his calculations look at Intrade and Betfair, along with a third market, the Iowa Electronics Market (IEM), because the combination enabled him to “mute any short term fluctuation or noise.”
Disparities between Intrade and Betfair are one way that a manipulation could become obvious, so a successful manipulation would move the numbers on both.
IEM, a futures market operated by the University of Iowa, limits account sizes to $500 — a small enough sum in relation to the amount of money circulating in the market to prevent any single bettor from affecting major change in any market. They also have a second market that acts as an arbitrage opportunity, allowing people to hedge their risks and preventing dramatic swings in prices. On IEM, people bet on a potentially different outcome than they do on Intrade and Betfair: That a candidate will win the popular vote, whereas at Intrade and Betfair, people bet on a candidate winning the election.
Rothschild also pointed out the importance of timing. Right after a major event such as a debate or a speech, a more significant movement in the polls wouldn’t necessarily seem that out of the ordinary — it would seem like opinion was changing as a result of the debate.
The McCain manipulation happened in the middle of the night in the United States, when it was obviously not a result of a campaign event.
Of course, it remains to be seen why anyone would want to manipulate the markets.
“I don’t think these markets are high enough profile that it’s probably a good use of money to do something like this,” said Koleman Strumpf, an economics professor at the University of Kansas Business School who has written on prediction markets. He noted that media more often covers poll results than Intrade results.
Moreover, he added, “It’s not clear that voters respond to these markets like that. If nobody really pays attention to these markets, not really clear why that would make a lot of sense.”
Joyce Berg, a professor at the University of Iowa who serves as the director of the IEM, had a similar take.
“It’s not clear to me which way you’d want to make these prices go,” Berg said. “Do you want to depress your candidate’s price so that other people sort of rally to the cause? Or do you want to inflate it because you think people join winners?”
Even assuming a successful manipulation of Intrade, “I don’t know that people are actually going to change the way they vote in response to our market,” said Thomas Reitz, another economics professor who sits on the steering committee for the IEM. He said poll outcomes seemed more likely to affect voter opinion, and something like advertising would be more effective “to create hype about a candidate.”
Steven Gjerstad, a professor of economics at Chapman University who studies the issue, argued that would-be manipulators face a steep challenge to overcome the confidence people have in their predictions.
“Presumably, people who attempt these sorts of manipulations think that by shifting prices they’ll shift what people believe about the probabilities of the different outcomes. They think that the map from prices to beliefs is very strong,” he said. “My interpretation of the ‘bounce-back’ is that beliefs are more resilient. They probably take a lot of factors into account other than prices and aren’t so easily manipulated by an attempt to manipulate prices. (That may be more true of problems like presidential elections or sporting events where knowledge is widely disbursed than with stocks, where beliefs are more easily manipulated by analysts).”