Study Connects Polymarket Accuracy to Skilled Minority

(AsiaGameHub) –   A recent SSRN working paper indicates that Polymarket’s predictive accuracy stems less from collective crowd wisdom and more from a small group of traders with a distinct advantage. Researchers also flagged 1,950 accounts whose trading patterns may suggest the use of non-public information.

Good to Know

• The study examined 1.72 million Polymarket accounts.
• Only 3.14% were classified as skilled winners.
• Flagged accounts earned an average of about $15,000 each.

Insider Trading Concerns Are at the Core

The paper, “Prediction Market Accuracy: Crowd Wisdom or Informed Minority?” was published on April 20, 2026, and revised on April 25. It was co-authored by Roberto Gomez-Cram, Yunhan Guo, and Howard Kung of London Business School, along with Theis Ingerslev Jensen of Yale University.

One segment of the research focused on potential insider trading. The authors identified 1,950 accounts with timing and conviction patterns that hinted at possible use of non-public information. These accounts also had significant price impacts when they traded.

One case involved three accounts that bought contracts tied to Venezuelan President Nicolas Maduro hours before a secret U.S S. military operation on January 3, 2026. Together, they made more than $630,000 in profits.

This finding alters the framing of the Polymarket accuracy debate. Prediction markets like Polymarket and Kalshi often present their price accuracy as a result of many participants pooling their views. The paper argues that the useful forecast signal came from a much smaller informed minority.

The authors reached this conclusion after analyzing Polymarket’s full transaction history, including 98,906 events, 210,322 markets, and $13.76 billion in total trading volume. They used a sign-randomization test to separate real skill from luck.

Only 3.14% of accounts qualified as skilled winners. These traders averaged 79 markets each, maintained profits outside the initial sample, and typically traded in the direction of final results. The other 96% either lost money or broke even due to chance.

Order flow data clearly showed this gap. A one percentage point increase in skilled net buying correlated with an 8 basis point rise in the probability of the correct final outcome. Lucky winners had positive balances, but their trades did not provide useful predictions of prices or outcomes.

Polymarket also grew rapidly during the study period. Monthly volume rose from $3.3 million in December 2023 to $1.98 billion in December 2025. Active accounts increased from about 1,600 to more than 519,000. Even so, skill remained highly concentrated.

The researchers found that skill was also persistent. In a random split test, 44% of traders labeled skilled in training data retained that label in test data. Unskilled losers stayed unskilled 51% of the time. Skilled mutual funds kept their label only 10% of the time in a parallel test.

Scheduled news tests produced the same result. Around FOMC announcements and corporate earnings releases, only skilled traders adjusted their order imbalance in the direction of the news surprise. Other account groups showed no consistent reaction.