A new academic study challenges the common belief that prediction market accuracy comes from the wisdom of the crowd. Instead, it finds that only a tiny slice of traders on Polymarket actually know what they are doing.
The working paper, titled “Prediction Market Accuracy: Crowd Wisdom or Informed Minority?” was published on SSRN on April 20, 2026, and revised five days later. Four researchers from London Business School and Yale University wrote it. They looked at the complete transaction history of Polymarket, which is the biggest prediction market by trading volume.
The numbers are huge. The study covers 98,906 events, 210,322 individual markets, and a staggering $13.76 billion in total trading volume. That came from about 1.72 million accounts.
Skill vs. Luck
The researchers used a statistical method called a sign-randomization test to sort traders. They wanted to see whose profits came from real skill and who was just lucky.
The finding is stark. Only 3.14% of accounts were skilled winners. These people earned persistent profits, traded across an average of 79 markets each, and consistently bet in the direction that matched the final outcome. The other 96% of accounts either broke even by luck or lost money overall.
Polymarket, along with platforms like Kalshi, often describes its accuracy as a product of collective intelligence from a diverse crowd. This study directly contradicts that idea.
Predictive Power
The skilled traders’ order flow predicted both next-period price changes and final market outcomes. A one-percentage-point increase in skilled net buying meant an 8 basis point increase in the probability of the correct final outcome. The lucky winners? They showed no meaningful predictive power at all.
The researchers also tested for skill persistence over time. They split events into training and test sets. Among traders classified as skilled in the training set, 44% kept that classification in the test set. For unskilled losers, 51% remained in that category. To put that in context, skilled mutual funds in a parallel test retained their classification only 10% of the time. The authors say prediction markets show unusually high persistence of both skill and anti-skill.
Insider Trading and Market Makers
Skilled traders also reacted first to scheduled news. In tests around FOMC announcements and corporate earnings releases, only the skilled group shifted its order imbalance in the direction of the news surprise. Other groups did not.
Then there is the insider trading angle. The researchers identified 1,950 accounts that met timing and conviction criteria suggesting they traded on non-public information. These accounts averaged roughly $15,000 in profits each. One documented case involved three accounts betting on a contract tied to Venezuelan President Nicolas Maduro hours before a secret U.S. military operation on Jan. 3, 2026. They collectively earned more than $630,000.
On April 23, 2026, the CFTC filed a complaint alleging that an active-duty U.S. Army service member engaged in insider trading using one of those accounts. Still, the researchers concluded that insider activity was too concentrated in isolated events to explain broad price discovery.
The vast majority of participants, the study found, funded accuracy rather than produced it. Unlucky and unskilled losers made up 67% of all accounts and absorbed the entirety of aggregate losses. Market makers and skilled takers together represented fewer than 3.5% of accounts but captured more than 30% of total gains.
The authors conclude that prediction market accuracy reflects the behavior of a small, identifiable group of informed traders. Whether those traders stick around as platforms grow and fees increase remains an open question for future research.
