Only 3% of investors predict the market correctly

Recently, a US Army soldier was prosecuted for betting on the capture of Venezuelan President Nicolas Maduro. In the weeks before Mr. Maduro’s arrest, the soldier used classified information to bet on the Polymarket platform that US forces would enter Venezuela and Mr. Maduro would lose office. As a result, this person won 400,000 USD. This is considered an isolated scandal in the prediction market.

Polymarket is the world’s largest decentralized prediction market platform, built on blockchain, allowing users to bet with cryptocurrency on the outcome of political, sporting or economic events. This platform and the prediction market are creating profound effects on the general psychology of investors, gradually becoming one of the “compasses” for the global financial community to refer to before making decisions.

However, a new study suggests that this could be a more worrying sign: a small group of “informed” traders become the price-makers in the market, while the majority of investors lose money.

The study is part of an academic manuscript by researchers from London Business School and Yale, aimed at testing the industry’s core assumption that prediction markets are powered by the “wisdom of crowds”. By analyzing all transactions on Polymarket from 2023 to 2025, the authors concluded that in fact it is not the crowd, but a small group of knowledgeable and well-informed traders who are driving the price.

They analyzed 1.72 million accounts with a total trading volume of 13.76 billion USD and discovered that only about 3% of traders are people who understand and place orders correctly, so they are the ones who decide whether the price is right or wrong. This group often predicts accurately and acts early.

On the contrary, most of the remaining 97% cannot do the above. They play the role of providing liquidity and creating trading volume for the market, but overall they are on the losing side.

An important issue is how to distinguish between skill and luck. With more than 1 million participants, it is easy for many people to win big just by luck.

To eliminate this factor, the authors rerun each trade 10,000 times, keeping all factors unchanged except the trade direction. Same market, same time, same amount, but buying or selling is decided by tossing a coin. This creates a benchmark: If the actual results consistently outperform the “coin toss” model, that’s skill; If not, it’s luck.

The results showed that among the biggest winners by absolute profit, only about 12% actually beat this benchmark. Many people win just because of luck while about 60% of them become losers.

However, a skilled trading team plays an important role in making the markets more accurate. As their trading weight increases, prices often move closer to the correct outcome, especially in the period close to the event. They are also the fastest to react when there is new information, such as announcements from the US Federal Reserve (Fed) or corporate earnings reports, while most other traders do not react consistently.

However, this informational advantage also poses a big problem: what happens when that information is not public or not authorized to be used? Both Polymarket and Kalshi claim that trading on inside information is prohibited, but research shows the risk is real.

An example given is the US arrest of the President of Venezuela in January. Before the event happened, 3 new accounts placed large bets on the possibility of Mr. Maduro being overthrown, when the market at that time only priced the probability at about 10%. These accounts place very large orders before the price can react. When the event happened, they earned more than 630,000 USD. After that, two accounts stopped trading completely, the remaining account was almost inactive. While there is no evidence of wrongdoing, this is an interesting case.

Trades based on inside information, if they occur, have a much stronger impact than regular trades – perhaps 7 to 12 times the magnitude of the price impact. However, they occur rarely and only focus on a few specific events. Most of the time, market accuracy still depends on skilled traders who consistently predict better than the masses.

The study’s conclusions challenge the popular view that prediction markets operate efficiently thanks to collective intelligence. Reality shows that the important thing is not what the majority thinks, but who among them really has the information and understanding.