Professor Joshua Mitts on Prediction Markets and Informed Trading

Prediction markets like Kalshi and Polymarket are booming. With about $24 billion in monthly global trading volume in April 2026 alone, prediction markets offer users the ability to bet on the occurrence of events as large as election outcomes and as small as high temperatures on a given day in a specific location. But Joshua Mitts, David J. Greenwald Professor of Law, says that a considerable amount of trading appears to be based on insider information. 

In a new paper, Mitts and co-author Moran Ofir of the University of Haifa analyzed all Polymarket contracts from February 2024 to February 2026 and found more than 210,000 suspicious bets. “These bets achieved a nearly 70% win rate, well in excess of chance,” says Mitts. “We estimate these traders earned $143 million in aggregate profit. There is a problem. The scope of that problem is open for debate.”

Below, Mitts discusses the pair’s research, the legal frameworks currently governing prediction markets, and recommendations for reforms to address informed trading on the platforms.

How did your analysis identify problematic trades?  

The criteria for our analysis are the size of the bet relative to your betting/trading history, the size of the bet relative to others’ bets, the profitability of the trade, and the directional imbalance of the trade. The fifth criterion is that you place the trade very close to the expiration of the market. For example, when the U.S. launched strikes on Iran on February 28, a trader placed a very large bet on “US strikes Iran by February 28” about 70 minutes before the news wire report of the strikes. That’s unlikely to be a random chance.

There is an important caveat that this analysis is just a statistical screen, notwithstanding all the effort we put in to argue this is not random trading; we don’t know in any given case that the individuals in question acquired information improperly. 

What are some examples of informed trading in prediction markets, and what are some of the issues that can arise from it? 

I think we all agree that it’s a problem for a special forces soldier who’s about to conduct a very sensitive military operation to trade on that event, particularly on a public blockchain-based system like Polymarket, where the trades are immediately broadcast to the entire world. 

Many informed trading cases are election-related. Polymarket traders built very sophisticated, state-by-state, county-by-county models as to who was going to win the 2024 presidential election. We would classify that as nonpublic information because it’s information the rest of the world didn’t have. But that’s a very different kind of information than “The U.S. is about to bomb Iran, and now’s the time to go trade on it.”

The vast majority of informed trading activity is very difficult, if not impossible, to detect. You’ll see someone making money, but you won’t be able to prove the information they had or that it was misappropriated—if you even know who they are. We identified multiple suspicious trades predicting the date of Taylor Swift’s engagement announcement, which, unlike the special forces operation to oust Nicolás Maduro from power in Venezuela, did not involve classified government information. The case of the special forces officer who traded on the Maduro removal is the exception, not the norm. He used a personal email address, and it was very straightforward to trace him. But these prediction market platforms also allow completely anonymous trading. 

What is the value of prediction markets beyond individual wins?

Defenders of prediction markets will argue that it’s actually quite useful for businesses to know the odds of a peace agreement in the Middle East or of war breaking out. There are geopolitical risk firms who essentially do this kind of forecasting. You could say that this is a free service that’s being provided by these traders.

In the case of elections, a defensible argument for prediction markets is that they inform the electorate and the candidates as to where people’s views are at a particular point in time. If the opinion poll has Kamala Harris up 6% but the prediction market looks completely different, maybe the Harris campaign should listen to the prediction market. 

When you get into a purely entertainment setting, like whether Taylor Swift is going to be engaged in the coming month, it becomes a lot harder to identify social value. It’s also harder to identify social costs. Sometimes people just want to have fun and play a game. We don’t say this game is OK and that game is not; we just let people play games.

Some states argue that prediction markets are gambling and therefore regulation should fall to the state. How are the prediction markets currently regulated, and what are some of the challenges in determining regulatory responsibility?

The Commodity Futures Trading Commission (CFTC) has the statutory authority to regulate prediction market contracts as financial contracts. There’s ongoing litigation between the states and the federal government over the extent of the CFTC’s authority. The states traditionally have had the authority to regulate gambling. But prediction market activity is conducted on digital platforms that are not housed within any given state. So it’s a bit of a square peg and a round hole to say New Jersey should be able to regulate platforms that are not housed in New Jersey even though they may have New Jersey users. 

Another important difference is that these markets are not traditional gambling markets. There is generally not a “house” that is taking the opposite side of a trade. The platforms match people who want to bet in one direction against those who want to bet in the other direction. That makes prediction markets much more like traditional financial markets.

Traditionally, U.S. regulators have said that when it comes to financial instruments like derivatives, the concerns are systemic stability—the idea that there is financial risk building up in the system. We haven’t said, Let’s protect the average consumer from buying a dangerous product. Given that paradigm, it comes as no surprise that the default assumption among U.S. regulators is that other than stamping out some misconduct, the prediction markets are basically fine. 

How should informed trading be addressed in prediction markets?

In our paper, we explain that we want to induce the platforms to conduct their own surveillance and be more aggressive. These platforms are the intermediary—much like broker-dealers, which have regulatory obligations and have been held responsible for market manipulation. Regulations could enforce heightened liability standards, where the platforms have an obligation and duty to engage in surveillance. If a platform knows that there’s a user who’s repeatedly exploiting insider information and then doesn’t make efforts to investigate, regulations could assign some legal liability to the platform. It’s in the interest of prediction markets to root out misconduct—it’s even in their interest to be liable for not rooting out misconduct—because if you become known as the platform where all the insiders trade, then your average customer won’t want to trade there.

There also could be policy interventions at the organizational level, throughout corporate America and the federal government. Organizations have ethics rules about trading on nonpublic information. For example, a Google engineer who used corporate information to trade on Polymarket was indicted by federal prosecutors in New York for insider trading. Using nonpublic proprietary information on material corporate events to trade on prediction markets probably already is considered a form of securities fraud, and rules prohibiting trading of this type are often enforced not only by the Securities and Exchange Commission but also by corporate compliance departments. Thinking about insider trading as a problem of employee compliance is the paradigm that will be most effective for dealing with the problem.

What has been the reaction to your proposals for reining in informed trading? 

A lot of people have agreed with the paper. The disagreements are either from people who say informed trading is not a problem or people who say we’re not going far enough in suggesting regulation and that these markets are bad because they are taking advantage of consumers. We are not saying prediction markets are bad. I’m a skeptic of regulators coming in and saying, “You’re having too much fun. It’s bad for you.” There’s a paternalism to that, and I struggle with drawing that line. At the same time, we should be looking at ways to address misconduct.

This interview has been edited and condensed.