Columbia University Study Finds One-Quarter of Polymarket Trading Volume Potentially Artificial

Researchers uncover systemic wash trading across prediction markets, peaking at 60% of volume in late 2024

TL;DR

• Columbia University researchers estimate roughly 25% of Polymarket’s total trading volume over the past three years was likely wash trading.
• Suspicious activity peaked near 60% of weekly volume in December 2024, driven by wallet clusters executing repetitive trades.
• Lack of KYC rules, zero trading fees, and potential airdrop incentives may have enabled widespread artificial volume.

A new academic paper from Columbia University has raised red flags over trading integrity on Polymarket, the leading blockchain-based prediction platform. The study, “Network-Based Detection of Wash Trading,” authored by Allen Sirolly, Hongyao Ma, Yash Kanoria, and Rajiv Sethi, was published on November 6 and estimates that about one-quarter of all trades executed on Polymarket between 2022 and 2025 were likely wash trades. Researchers describe the behavior as “artificial trading volume with no real market risk,” often resulting from coordinated self-trading between related accounts.

Polymarket, which lets users wager on real-world outcomes using USDC stablecoins, has reportedly handled more than $18 billion in total volume and drawn around 1.3 million users since launch. The Columbia team analyzed years of on-chain activity, mapping wallet interactions through a network-clustering algorithm that identifies groups of addresses repeatedly trading against one another. Their analysis found roughly 25% of aggregate trading activity across the period displayed characteristics consistent with wash trading—wallets exchanging contracts back and forth with negligible gains or losses.

Suspicious activity was not evenly distributed. Researchers observed a steady uptick from mid-2024, climbing sharply to nearly 60% of total weekly volume by December that year. The elevated levels persisted into early 2025 before falling below 5% by May, then rising again to about 20% by October 2025. One network cluster comprising more than 43,000 wallets accounted for nearly $1 million in trades, often producing per-trade outcomes of less than a cent in profit or loss.

The paper breaks down the distortion by market category. Sports prediction markets showed the highest concentration of suspicious activity, with around 45% of all-time volume flagged as likely wash trades. Election markets followed at 17%, politics at 12%, and crypto-related markets at roughly 3%. Some individual weeks saw far more extreme figures: election markets reached 95% of flagged activity during the week of March 24 2025, while sports markets hit 90% during the week of October 21 2024.

Researchers cite several structural factors that may have made the behavior easier. Polymarket does not require users to complete Know-Your-Customer verification, allowing traders to generate multiple anonymous wallets. The platform also operated without trading fees during the study period, removing a financial disincentive for repetitive transactions. Additionally, speculation around a potential future token launch or “airdrop” may have encouraged traders to inflate their on-chain activity to secure eligibility for potential rewards.

Lead author Allen Sirolly emphasized that the findings show no evidence of Polymarket’s direct involvement but highlight how wash trading can “recycle capital across multiple trades without real exposure.” Co-author Yash Kanoria added that he hoped Polymarket would “welcome the analysis” and use the methodology to identify and exclude suspicious accounts from future reward programs. The researchers suggest that their network-based detection framework could serve as a self-regulatory tool for decentralized exchanges and prediction platforms facing similar challenges.

Wash trading, the study concludes, undermines the predictive power of such markets by distorting real liquidity and crowd intelligence. It also risks misleading investors, analysts, and regulators about genuine user participation. The authors argue that transparent detection and mitigation are crucial for the credibility of decentralized prediction platforms operating at scale—especially as the line between speculative betting and data-driven forecasting continues to blur.

Polymarket previously faced regulatory scrutiny when the U.S. Commodity Futures Trading Commission fined it in 2022 for offering unregistered binary options markets. While the new report does not accuse the platform of complicity, it underscores persistent risks in decentralized finance ecosystems where anonymity, incentives, and zero-fee structures can amplify non-economic trading. For a market designed to crowdsource truth, Columbia’s findings expose a paradox: too much fake activity may drown out the very signal prediction markets are meant to reveal.

This article has been refined and enhanced by ChatGPT.