Unusual Bets Emerge on Prediction Markets Ahead of Midterm Polls
Suspiciously large wagers appeared on prediction markets just before the release of two New York Times-Siena College polls, prompting an investigation by Vanderbilt University researchers. The team traced an anonymous trader who executed 25 trades within two hours, netting over $250,000. Polymarket has launched its own inquiry.
Research Uncovers Suspicious Trading Patterns
Less than a week after the U.S. midterm elections, a transaction traceback by a Vanderbilt University research team has reignited concerns about insider trading on prediction markets. The team’s analysis of on-chain fund flows and platform transaction records revealed that an anonymous trader established significant positions on both Polymarket and Kalshi simultaneously, just prior to the release of two polls by The New York Times and Siena College. The trader made 25 transactions in two hours, accumulating profits exceeding $250,000.
Nebraska Senate Race: A Case Study
The first instance occurred on October 31. According to the research team, the account in question placed a bet of $191,277 on Republican Senator Pete Ricketts of Nebraska to win reelection. Approximately one hour later, at 5:05 AM, Siena College released poll results showing Ricketts leading independent candidate Dan Osborn by six percentage points, 49% to 43% – one of his largest leads in the race. Before the bet, Polymarket had assigned Ricketts a 67% chance of winning; this figure rose to approximately 73% after the bet and climbed further to 79% following the poll release. This single position ultimately yielded a profit of over $77,000.
Michigan Senate Race: Larger Stakes
The same account placed even larger bets in Michigan. At 3:27 AM on October 31, the account wagered $349,135 on Democrat Abdul El-Sayed to win the Michigan Senate seat. An additional $273,473 was added at 5:20 AM that morning, bringing the total investment to $622,608. On the evening of November 1, The New York Times published poll results for multiple states, including Michigan, showing El-Sayed with an 11-percentage-point lead. He went on to win the election two days later, and this set of positions generated over $180,000 in profit for the account.
Fund Flows Raise Red Flags
The trajectory of the funds proved more noteworthy than the bets themselves. The research team’s reconstructed fund timeline indicates that around 5:30 PM on October 30, over $2 million was deposited into an anonymous cryptocurrency wallet. Within minutes, these funds were transferred in batches to Kalshi and Polymarket. Prior to October, this account had only traded in three other markets. The research team also observed similar large-scale betting activity on Kalshi.
Calls for Further Investigation
Brett Goldstein, co-host of the Vanderbilt University National Security Institute’s prediction market research team, stated that this series of events is “difficult to dismiss as a random occurrence” and warrants further investigation. A Polymarket spokesperson confirmed that the platform detected the transactions and has initiated an investigation. Kalshi declined to comment.
Independent Verification and Alternative Explanations
The research was further reviewed by on-chain analysis firm Nansen and the Washington-based research group Anti-Corruption Data Collective. Both confirmed the overall findings but were unable to determine the trader’s motivations. The Anti-Corruption Data Collective offered a counterpoint: the account held its positions until Election Day. Typically, individuals with access to non-public polling data would exit their positions after the poll is released and prices surge.
Media and Polling Firms Respond
Charlie Stadtlander, a spokesperson for The New York Times, stated that the newspaper prohibits employees from investing based on unpublished reporting. Lyle Durbin, CEO of ReconMR, the parent company of Siena College and a data collection firm, said that employees are bound by confidentiality and non-disclosure agreements and are prohibited from using unpublished data for any purpose. Both The New York Times and Siena College indicated they would review the research to see if any conclusions could be drawn.
Precedent of Prediction Market Scrutiny
This is not the first time prediction markets have faced scrutiny. In April, the U.S. Attorney’s Office for the Southern District of New York indicted Army Sergeant Gannon Ken Van Dyke, accusing him of using confidential information gained from his participation in Operation “Absolute Resolve” to bet on contracts on Polymarket, such as “Maduro resigns by January 31,” netting over $400,000. The operation ultimately led to the arrest of then-President Maduro in Caracas, Venezuela.
The Information Advantage Dilemma
The inherent logic of prediction market products is to monetize information advantages: those who know the answer before the market can profit. The true challenge for regulators lies in the difficulty of distinguishing between “knowing the content of a poll” and “guessing when a poll will be released,” as both appear identical in transaction records. The Vanderbilt team’s alternative explanations also point to this ambiguity, suggesting the trader might have simply capitalized on the timing of poll releases, possessed strong personal convictions about the election, or engaged in cross-platform hedging strategies without any special information.
Platform Investigations Yield Limited Results
Platform reporting data indicates that this defense mechanism is operational but has limited efficiency. Kalshi stated that it has initiated nearly 60 election-related transaction investigations since January 1, with approximately half referred to law enforcement. However, 40 to 45 cases were closed without findings of wrongdoing. In other words, most flagged abnormal trades ultimately proved to be false positives, and genuine insider trading cases often require external intervention from entities like the Department of Justice.
Prediction Markets in Taiwan
Turning the lens to Taiwan, prediction markets have previously featured local political events. Contracts included “Will Cheng Li-wen meet with Xi Jinping before the end of June 2026” and “Will Lai Ching-te step down before the end of 2026.” The market assigned probabilities of 34% and 12%, respectively.
The information sources for such contracts often stem from internal Taiwanese polls, party confidences, and media schedules. Taiwan’s current legal framework offers virtually no avenues for cooperation with offshore platforms, making the difficulty of investigating precise bets placed before poll releases even higher than in the United States.
Leave a Reply