Do 3% of traders determine market direction? Research reveals: Prediction market accuracy does not rely on the "wisdom of the crowd"

A while ago, a U.S. soldier was arrested for allegedly using classified information to place bets on Polymarket, which seemed like an isolated scandal in the prediction market. However, recent research suggests that this soldier may just be the tip of the iceberg. He represents a small group of "smart money" with access to insider information, the very people who effectively drive price trends in prediction markets like Polymarket; surrounding them are the vast majority of retail investors who are constantly losing money and wailing in despair. This research is a working paper co-authored by four scholars from London Business School and Yale University, directly challenging the core claim of prediction markets: that markets are accurate because they aggregate the "wisdom of the crowd" from a vast number of participants. The research team meticulously analyzed every transaction on Polymarket from 2023 to 2025. After examining a massive dataset of 1.72 million accounts and $13.76 billion in trading volume, the scholars reached a brutal conclusion: it is actually a small group of "informed traders" who truly drive market prices. The results show that only 3% of traders dominate the vast majority of "Price Discovery," meaning it is these 3% who push prices step-by-step toward the final correct prediction. This minority of traders not only accurately predicts outcomes but also guides prices in the right direction. As for the remaining 97% of retail investors, they possess almost no predictive ability; they merely provide liquidity and contribute to trading volume, but overall, they are on the losing side of their bets against the few informed traders. Real skill or just pure luck? However, distinguishing between a "true betting god" and someone who is just "lucky" among millions of Polymarket users is no easy task. To filter out the luck factor, the research team simulated each trader's betting behavior 10,000 times, keeping all conditions constant—the same market, the same timing, the same bet amounts—with the only difference being that the betting direction was determined by a coin flip. This established a baseline for judging whether a trader possesses "real skill." If a trader's actual profit outperforms the coin-flip results, it is genuine skill; otherwise, it is purely luck. The results show that among the most profitable winners on paper, only 12% performed better than the "coin flip" results. Even more brutally, about 60% of these "lucky winners" immediately became losing traders once they switched to a different set of prediction events. Retail investors are "behind the curve" on market trends It is undeniable that this small elite group has indeed improved the accuracy of prediction markets. The study found that when these skilled players account for a larger proportion of trading, prices tend to be closer to the ultimate truth, a phenomenon that is particularly evident in the final sprint before the event results are revealed. Furthermore, they are highly sensitive to information. Whether it is the Fed releasing major policy decisions or companies announcing earnings, this smart money always reacts immediately and adjusts their positions rapidly; in contrast, average retail investors often appear a step behind when faced with these major events that can shake the market, and they almost always fail to take action. The shadow of insider trading lingers, profiting from precise strikes However, this "information lead" also triggers legal and ethical gray areas: what if this information is not public at all, or even classified information that should not have been leaked? The study cites one of the most controversial cases: the U.S.-led military operation in January 2026 to overthrow Venezuelan President Nicolás Maduro. Hours before the operation began, three mysterious accounts appeared on Polymarket, aggressively increasing their positions on the "Maduro will step down" contract. At the time, the market believed the probability was only 10%, but these accounts placed bets worth tens of thousands of dollars before the price volatility occurred. When news of the coup broke and the event became reality, these three accounts collectively raked in over $630,000. After cashing out, two of the accounts immediately stopped trading, and the third account also went almost dormant. Although there is currently no hard evidence proving these accounts were involved in illegal activities, this clearly illustrates the dramatic impact of "non-public information" on prediction market prices. The study found that funds suspected of insider trading have 7 to 12 times the price-driving power of ordinary professional traders. However, the study also notes that such extreme insider trading is relatively rare, usually concentrated on a few specific events, and is not the main driver of daily market operations. Most of the time, market accuracy still relies on those experienced hands who profit steadily over the long term. In summary, this heavyweight research completely overturns the romantic fairy tale that "prediction markets rely on the wisdom of the crowd." Prediction markets are highly accurate not because they aggregate