| Updated:
Prediction markets like Polymarket and Kalshi are more accurate than Wall Street analysts in forecasting corporate earnings. But it’s not for the reason you think, writes Dr Roberto Gomez-Cram
Who will be the next UK prime minister? Will Arsenal win the Premier League? Will the UK win Eurovision? What will be the highest temperature in London today?
These are questions people debate every day. On prediction markets, they are also questions on which people stake real money from thousands to millions. The sector has grown explosively, with trading volume reportedly increasing 130-fold from 2024 to 2025 and monthly volume reaching $13bn by the end of 2025. Its two leading platforms, Polymarket and Kalshi, have been valued at $9bn and $11bn.
Prediction markets enable participants to trade on real-world events, and price the probabilities that these events occur. Events cover a wide range including politics, sports, cryptocurrency, finance, economy, culture, technology and weather. Each market is framed around a clear question with a Yes or No outcome. For example, “Will Arsenal win the Premier League this season”. Participants can take a position by buying either Yes or No, which pays off 1 USD if the outcome occurs and 0 otherwise. The prices can be approximately interpreted, as the predicted probability of an outcome will happen.
The accuracy of prediction markets like Polymarket
Polymarket CEO Shayne Coplan has called prediction markets “the most accurate thing we have as mankind right now, until someone else creates some sort of a super crystal ball”. This claim may be an exaggeration, but it is shown in different fields that the forecasts generated by prediction markets, which are financially incentivized and updated in real time, can outperform benchmarks like polls and experts.
Prediction markets gained widespread attention during the 2024 US presidential election. On the eve of the vote, Polymarket and Kalshi gave Donald Trump a slight edge, with implied probabilities of around 55 per cent and 52 per cent, while the New York Times poll showed Kamala Harris narrowly ahead. Trump’s eventual victory highlighted the ability of prediction markets to aggregate dispersed information more accurately.
Our recent research shows that prediction markets are more accurate than Wall Street analysts in forecasting corporate earnings. In September 2025, Polymarket introduced earnings markets based on whether firms will beat analyst forecasts. For example, “Will BP beat quarterly earnings?”. In a recent paper with myself, Howard Kung and Yunhan Guo of London Business School and Theis Ingerslev Jensen of Yale University, we find that one day before earnings announcements, prediction markets correctly forecast 78 per cent of outcomes, compared with 62 per cent for Wall Street analysts.
The wisdom of the informed minority, not the crowd
Why are prediction markets accurate? The usual answer is the wisdom of crowds, whereby markets aggregate information from a large and diverse pool of participants. However, in a separate paper, we challenge that view using a large-scale dataset of 1.7m accounts and $13.7bn in trading from 2023 to 2025.
In a large-scale analysis of 1.7m accounts and $13.7bn in trading from 2023 to 2025, we find that market accuracy is not driven by the “wisdom of crowds”, but by a small, highly informed minority. About three per cent of accounts earn profits persistently and generate the bulk of price discovery. Their trades predict future prices and final outcomes, make prices more accurate throughout a market’s lifespan and react to news the moment it arrives. While the majority generate most of the trading volume, they don’t improve accuracy and their losses fund the informed minority.
The future of prediction markets
The romantic idea of crowd wisdom may be overstated, but the future still looks promising. Prediction markets offer timely forecasts across many domains and a real-time measure of expectations for journalists, policymakers, companies and investors. For financial markets, they create ways to trade specific events that are otherwise difficult to isolate. As liquidity deepens and prediction markets move closer to mainstream finance, their prices are likely to become part of how society measures uncertainty.
Dr Roberto Gomez-Cram is assistant professor of finance at London Business School
