A recent scandal involving a Green Beret who was arrested for wagering on a classified US raid may be more than an isolated incident, according to a new study. The research suggests that this individual may be an extreme example of a small group of informed traders who actually influence prices on platforms like Polymarket, while the majority of participants incur losses around them.
The study, conducted by researchers from the London Business School and Yale, analyzed 1.72 million accounts and $13.76 billion in trading volume on Polymarket between 2023 and 2025. The findings indicate that just 3% of traders are responsible for the majority of price discovery, meaning they are the ones who drive prices towards the correct outcome.
These informed traders consistently make accurate predictions and move prices in the right direction, whereas the remaining 97% of traders do not. The latter group provides liquidity and generates volume but ultimately loses money to the informed minority.
Distinguishing between skill and luck is a challenging task, as many traders may accumulate significant winnings by chance alone. To address this issue, the researchers simulated each trader's bets 10,000 times, keeping all parameters the same except the direction of the trade. This allowed them to establish a benchmark for what each trader's profits would look like without any real edge. The results show that among the biggest winners, only 12% outperformed the benchmark, and many apparent winners did not sustain their performance over time.
The activity of skilled traders improves market accuracy, particularly in the final stages before an event's resolution. They are also the first to react to new information, adjusting their positions in response to events like Federal Reserve announcements or corporate earnings.
However, the same edge that makes skilled traders valuable to price discovery raises concerns when they possess non-public information. Both Polymarket and Kalshi have stated that trading on non-public information is strictly prohibited. The study cites the US removal of Nicolás Maduro from power in Venezuela as a concrete example of this risk. In the days leading up to the operation, three newly created Polymarket accounts placed unusually large bets on a contract asking whether Maduro would be removed, with the market pricing the odds at around 10%.
The accounts collectively made over $630,000 when the raid occurred, and two of them stopped trading soon after. While there is no evidence of wrongdoing on these accounts, the incident highlights the potential risks associated with insider trades. The study's findings challenge the notion that prediction markets work due to the collective knowledge of their participants. Instead, they suggest that the accuracy of these markets depends on the presence of informed traders who consistently outperform the crowd.