A comprehensive analysis of Polymarket trading activity spanning 2023 to early 2025 reveals a stark concentration of skill and profitability within the platform. The findings, presented in a working paper examining millions of trades, show that roughly 3% of active accounts are responsible for the majority of price discovery—the mechanism by which markets converge on accurate probabilities. This pattern mirrors observations in traditional financial markets, but the degree of concentration on Polymarket appears particularly pronounced, suggesting that prediction markets may attract either exceptionally skilled operators or reveal fundamental imbalances in how retail participants engage with probabilistic information.
The mechanics underlying this inequality reflect both structural and behavioral factors. Prediction markets like Polymarket require participants to assess complex, multi-variable outcomes and convert those assessments into capital allocation decisions. This process demands not only domain expertise but also an understanding of market microstructure, liquidity dynamics, and behavioral psychology. The elite 3% likely possess combinations of these competencies—whether through professional backgrounds in quantitative trading, academic training in forecasting, or simply ruthless operational discipline. Meanwhile, the remaining 97% may engage with Polymarket as a form of entertainment, seeking alpha but lacking the systematic approach or informational edge that separates consistent winners from those who leak value session after session.
What makes this finding particularly significant is its implication for market efficiency. Price discovery—the process by which markets aggregate dispersed information into consensus valuations—typically benefits from broad participation and diverse perspectives. When a small minority dominates the mechanism, it raises questions about whether marginal traders are pricing in genuine new information or simply exploiting inefficiencies created by less-informed participants. The data suggests that Polymarket may function less like a pure information aggregation mechanism and more like a venue where sophisticated traders extract rents from casual speculators. This dynamic isn't unique to crypto-native prediction markets, but the relative youth of Polymarket and the heterogeneity of its user base make such disparities more visible and potentially more acute.
The broader implication extends to how we should interpret prediction market outcomes. If outcomes are weighted heavily by a narrow set of well-capitalized, sophisticated traders, those predictions may reflect their particular blind spots, biases, or incentive structures rather than a truly decentralized consensus. As prediction markets expand their role in financial and political forecasting, understanding who actually drives price movements becomes essential for evaluating their reliability as information sources.