A recent scandal involving a Green Beret arrested for betting 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 significantly influence prices on platforms like Polymarket, while the majority of users incur losses. The study, conducted by researchers from London Business School and Yale, analyzed 1.72 million accounts and $13.76 billion in trading volume on Polymarket from 2023 to 2025. The findings indicate that just 3% of traders are responsible for the majority of price discovery, consistently predicting outcomes and moving prices in the right direction.

In contrast, the remaining 97% of traders primarily provide liquidity and generate volume, but tend to be on the losing side of trades against the informed minority. To distinguish between skill and luck, the researchers simulated each trader's bets 10,000 times, using the same markets, moments, and dollar amounts, but with the direction of the bet determined by a coin flip. The results showed that only 12% of the biggest winners by raw profit consistently outperformed the benchmark, suggesting that many apparent winners may have simply been lucky.

The study also found that the activity of skilled traders improves market accuracy, particularly in the final stages before an event's resolution. Moreover, these traders tend to be the first to react to new information, such as Federal Reserve announcements or corporate earnings. However, the same edge that makes skilled traders valuable to price discovery raises concerns when they may be trading on non-public information.

The researchers cite the example of the US removal of Nicolás Maduro from power in Venezuela, where three newly created Polymarket accounts placed unusually large bets on the event before it occurred, collectively making over $630,000. While there is no evidence of wrongdoing in this case, the study highlights the risk of insider trades and the need for platforms to ensure that trading on non-public information is strictly prohibited. Overall, the findings challenge the notion that prediction markets work due to the collective knowledge of their participants, instead suggesting that they are driven by the actions of a small group of informed traders.