The ‘Outsider Trading’ Problem

There are worse things you could do than drive to a Parisian airport and mess around with a temperature sensor on the airfield—for only a few minutes! But why do that at all?To make money, maybe. This is the scheme that some mysterious person (or persons) reportedly carried out in April to win a bet on the prediction-market platform Polymarket, which allows people to wager on the highest temperature in Paris on any given day, not unlike weather derivatives. The popular version of the story says that the shadowy figure warmed the sensor with a hair dryer—a photo showing this turned out to be AI generated—but it could have been anything. A lighter, a hand warmer.By artificially boosting the number on the sensor, this person won about $20,000. The French police are investigating because you can’t trespass at the airport or tamper with its equipment, but Polymarket apparently didn’t take the winnings back. (It did change the airport that it relies on for Paris-temperature readings; the French government, however, blocked access to the platform this week.)Read: A technology for a low-trust societyOver the past couple of years, plenty of concern has been expressed about how prediction markets may enable—and create the incentive for—insider trading. Some of the many people who have wagered on markets corresponding to when a war would start, when a regime would change, or even who would be at Taylor Swift’s wedding may actually have known the outcome before it happened. “Mention markets,” which allow participants to bet on words that a public figure might use in a speech, or how long the speech will be, can obviously be gamed by people close to the figure in question—just last week, ABC News reported that a teleprompter operator at the White House has made more than $100,000 by using Kalshi to wager on President Trump’s speeches. (“The White House has strict ethics guidelines that we expect all staffers and officials to follow,” a White House spokesperson told the outlet, while confirming that the employee had been placed on leave.)The issues with insider trading are clear: It’s illegal, for starters, and the platforms ban it, anyway. (Although insider trading also arguably serves the purpose of prediction markets, which is to channel the “wisdom of the crowd” to forecast the event most likely to happen.) People can also easily identify and avoid the markets most susceptible to it. But what about the hair-dryer sort of trading, when an apparent outsider changes the course of events to cash in on a given outcome? “I think of it as probably a bigger issue or problem than insider trading would be,” Vincent Grégoire, a professor of finance at HEC Montréal, told me. The Parisian-airport incident didn’t hurt anybody, but future ones might. (You can imagine a bunch of ways in which someone might create a dangerous situation while, for example, attempting to delay an airplane.)Read: Insider trading is going to get people killedSome experts call this type of fraud “manipulation of the physical world”—a riff on market manipulation. I asked others for catchier ideas. Joshua Mitts, a professor at Columbia Law School, wasn’t sure that manipulation was the right word, because it has a specific definition for financial markets, but he still kept saying it accidentally as we spoke. I agree that the term is a little lacking. Manipulating the physical world is what I do every moment of every day as I breathe and walk. I would call this type of fraud “outsider trading.”Outsider trading is also against platform rules. Polymarket and Kalshi both prohibit any type of market manipulation. When I reached out to Polymarket to ask about what had happened in Paris, a spokesperson relayed a statement to me from Olivia Chalos, its deputy chief legal officer, who wrote that Polymarket “maintains monitoring, review, enforcement, and reporting processes to address potential violations.”But it’s obviously a bit of a whack-a-mole situation. The U.S. Commodity Futures Trading Commission recently proposed more federal rules for prediction-market platforms to ease the problem. One of the big suggestions was to prohibit the listing of markets that would “create perverse financial incentives”—for instance, markets on wildfires or a market on whether a famous ape named Little Joe would escape from a zoo (for a second time). Polymarket includes a note on that market explaining which law a person would be breaking if they attempted to free the gorilla themselves, and that they could receive a life sentence in prison for doing so. That may have been the first time a prediction-market platform has featured such a warning, the Bloomberg columnist Matt Levine noted in a May newsletter, “but I have not done a comprehensive survey and I doubt it’ll be the last.”The experts I spoke with said that the most appealing markets for this kind of trickery would rely on a single number or simple, discrete outcome that a normal person could cheaply affect—the temperature reading in Paris, for example. Caleb Davies, an internet-famous prediction-market trader, highlighted another when he wrote about his experience losing thousands of dollars betting on Spotify’s streaming charts, which scammers had apparently figured out how to game by artificially juicing how many times a given song was played. Davies says the experience led him to an epiphany. He’s no longer just betting on the data, he said: “I’m also betting that nobody is going to mess with it.” And that’s a bad bet.Davies posted a screenshot to X to show that he’d removed about $250,000 from his Kalshi account, and he says he’s stopped using it entirely. “I love prediction markets,” he told me. “I’d love to keep playing in them. But right now, both platforms are garbage, I think.” When I contacted the company for comment, the Kalshi spokesperson Jacki McGavick said that Kalshi “takes market manipulation or fraudulent activity as seriously as any federally regulated financial exchange” and is investigating the Spotify incident.For the sake of conversation, Davies scrolled through Kalshi while we were on the phone, looking for opportunities for manipulation. Spotify charts were a good one, he knew. Rotten Tomatoes scores would be better (more money), though a little harder to game. (Nevertheless, New York magazine’s Matt Stieb figured out how to do it.) I pointed out to him that a market was open for the timing of this year’s first human case of screwworm in the United States. You could expose yourself to some infected livestock. But he wasn’t interested. “I think I’d just rather not make money than do that,” he said. “I’ve read about that—it doesn’t sound very appealing.”Read: It’s not gambling, it’s “girl math”Right now, the platforms—which are very much in advertising mode—carry a lot of markets that don’t have a ton of trading volume. They want as many as they can to attract new users and generate revenue. Also, a lot of them are funny: Most people probably recognize that it’s a waste of money to wager on whether Little Joe the gorilla will escape, but it’s fun to talk about. As more outsider-trading stories surface, though, people will likely be more hesitant to participate in markets that seem easy to rig. This problem, even more than insider trading, may start to limit what Polymarket and Kalshi can offer.There are plenty of things the platforms could do to intervene in outsider trading, Joshua Della Vedova, an associate professor of finance at the University of San Diego’s business school, told me. They could limit the size of bets in markets in which the result would be easy to rig. They could look for suspicious trades using some of the same methods that detect insider trading. His preferred method looks for “excess accuracy.” If someone bets that an Olivia Rodrigo song is going to hit No. 1, that makes sense—but if they also specifically bet “no” on Drake and Charli XCX and Bad Bunny, then they’re demonstrating a suspicious amount of confidence. That confidence could come from having an inside line, or it could come from using bots to spike streams artificially. Platforms could also disallow markets that are resolved by checking a single number from a single source, he suggested. That would have solved the issue at the Paris airport, for example, and it’s not as if Polymarket couldn’t have used an aggregation of weather data, rather than just one reading.Without some change, the platforms may continue to send ripples through reality—beyond just the goofy markets that they choose to list. A Wall Street Journal report found that Polymarket had been encouraging people to share videos with faked versions of their Polymarket dashboards, showing made-up earnings on imaginary bets that never happened. Weirder to me were the staged viral clips through which Polymarket and Kalshi looped themselves into the New York Knicks’ run to their first NBA Championship in more than 50 years.In one of those clips, a purported Knicks fan outside Madison Square Garden yelled, apparently improvisationally, “My mayor Muslim, my bagels Jewish, my Christian Dior, Knicks in four.” Later, when it turned out that he hadn’t made this up at all—someone had posted it as a TikTok comment weeks before and parts of the rhyme were even older—and that the mic he’d yelled into had been Kalshi branded, the whole thing a Kalshi stunt, nobody seemed to care. The chant became real. It was everywhere.The platforms pitch themselves as truth machines, which could mean a couple of different things. Are they finding it or making it? At some point, it may be difficult to identify the difference between reflecting our reality and creating a new one—or doing both.