A new working paper argues that the accuracy of fast-growing prediction markets is driven not by broad crowd wisdom but by a small cadre of skilled participants who consistently move prices toward the correct outcomes. The paper—titled “Prediction Market Accuracy: Crowd Wisdom or Informed Minority?” by Roberto Gómez-Cram, Yunhan Guo, Theis Ingerslev Jensen, and Howard Kung—analyzes Polymarket trades from 2023 to 2025 and finds that roughly 3% of traders account for most price discovery, tend to be first to react to new information, and regularly push prices in the right direction, according to CoinDesk.
The researchers examined approximately $13.76 billion in trading volume spanning 1.72 million accounts and 98,906 events over the 2023–2025 period. Their analysis shows that the informed minority—comprising market makers and skilled traders—accounts for over 30% of total profits while representing less than 3.5% of accounts. By contrast, 67% of accounts are categorized as “unlucky or unskilled losers,” a group whose losses largely fund the gains of better-informed participants. The study emphasizes that the majority contributes substantial trading volume without adding much information and that the order flow of the informed minority predicts both next-period price changes and final outcomes at statistically significant levels. When these skilled participants make up a larger share of market activity, prices move closer to the eventual result.
The researchers also report strong evidence of persistence in trader skill. Forty-four percent of accounts classified as skilled in an initial learning sample remained skilled in a separate sample, a rate far higher than the roughly 10% persistence observed in a parallel test of active mutual funds. Yet raw profitability alone often masked luck. Among the biggest winners by profit, only 12% outperformed a benchmark generated by running 10,000 random outcome simulations of their trading, and about 60% of apparent “lucky winners” turned into losers when their results were checked against an independent set of events.
These findings arrive as prediction markets surge in popularity and scale. On Polymarket, monthly trading volume jumped from $3.3 million in December 2023 to $1.98 billion in December 2025, while active accounts expanded from about 1,600 to more than 519,000 over the same period. More broadly, prediction markets have emerged as one of the fastest-growing use cases in the crypto ecosystem, with monthly trading volume exceeding $15 billion. A separate study indicates that only 0.015% of Polymarket users have achieved consistent profits at a level that might prompt them to consider leaving their primary jobs, underscoring how rare durable, market-beating performance appears to be even amid headline growth. The concentration of returns in a tiny slice of the user base is consistent with the new paper’s core conclusion that it is an informed minority, rather than the average participant, that endows these markets with their forecasting power.
Contrarian, information-driven strategies
Behavioral dynamics appear to reinforce the edge of contrarian, information-driven strategies. Prediction markets are often lauded for harnessing financial incentives to produce rational forecasts of real-world outcomes, but emotional reactions, hype, and confirmation bias can tilt probabilities away from fundamentals. News cycles featuring dramatic geopolitical or economic themes tend to inflate perceived likelihoods, luring traders into low-probability scenarios that nevertheless attract millions in volume; examples include speculative contracts on the United States acquiring Greenland or winning a Nobel Prize, a pattern behavioral economics links to narrative and confirmation biases. Data show that more than 73% of prediction market outcomes resolve as “No,” a long-run tendency that supports strategies focused on systematically betting against unlikely outcomes, according to TokenPost.
Some high-profile participants have leaned into this contrarian approach. Vitalik Buterin reported earning $70,000 on Polymarket by targeting what he characterized as irrational predictions, allocating $440,000 across contracts on unlikely events and posting a 16% return by opposing exaggerated market sentiment. The approach highlights how strategic traders may profit from price distortions created by public enthusiasm and attention—distortions that the new paper attributes to the trading activity of a large, less-informed majority. When the informed minority is most active, prices tend to move faster toward accurate probabilities, and when exuberance pushes odds away from reality, disciplined betting against the crowd can capture the gap.