After reviewing 1.72 million accounts, 210,322 markets, and $13.76 billion in trading volume, the study found only 3.14% of accounts qualified as “skilled winners,” whose trades consistently predicted price moves and final outcomes. These traders, along with market makers, captured more than 30% of all gains despite representing less than 3.5% of accounts. Study Questions Polymarket Crowd Wisdom Model A new academic working paper from researchers at London Business School and Yale challenged the popular belief that prediction markets succeed because of broad crowd wisdom. After examining every transaction on Polymarket between 2023 and 2025, the researchers concluded that a very small minority of users appears to drive much of the platform’s price discovery and forecasting accuracy. The study analysed a vast dataset that included 1.72 million user accounts, 210,322 separate markets, and around $13.76 billion in total trading volume. Despite the platform’s size and activity, the researchers found that only 3.14% of accounts could be classified as “skilled winners.” These were traders whose buying and selling patterns consistently predicted short-term price movements as well as the final outcomes of the events being traded. According to the paper, this small group of skilled traders, together with market makers, captured more than 30% of all gains on the platform, even though they represented less than 3.5% of total accounts. The findings suggest that a concentrated minority of informed participants may be responsible for much of the market’s efficiency, rather than millions of ordinary users collectively producing accurate prices. The researchers also argued that raw profits alone are not enough to prove trading skill. To test whether winning traders were genuinely talented or simply fortunate, they ran simulations that randomly reversed buy and sell decisions 10,000 times across historical trades. This method found that only 12% of the highest earners overlapped with the group identified as genuinely skilled. In contrast, around 60% of so-called “lucky winners” later slipped into losses when their performance was tested on a different sample of markets. The paper also found that 67% of accounts labelled as unlucky or unskilled losers absorbed the platform’s overall losses. This suggests that many users may be consistently funding the gains of a much smaller and more informed group of traders. Researchers additionally reviewed possible insider trading behaviour. They identified 1,950 accounts that were opened shortly before a specific event and then became inactive once that event ended. These accounts moved prices seven to twelve times more per dollar traded than skilled traders. However, the authors said this behaviour was too concentrated in isolated markets to explain the platform’s forecasting success. Insider accounts identified in ‘Maduro Out’ contract The authors concluded that the platform’s accuracy reflects “the wisdom of an informed minority, not the wisdom of the crowd.”
