Vanderbilt researchers flag poll-timed election bets on Polymarket and Kalshi

Vanderbilt University researchers have identified a cluster of large election wagers on Polymarket and Kalshi that were placed shortly before two New York Times/Siena polls were released, raising questions about whether traders had nonpublic information or simply made unusually well-timed bets.

The New York Times reported that one Polymarket user wagered about $191,000 on Oct. 31 that Nebraska’s Republican senator, Pete Ricketts, would win re-election. About an hour later, a Times/Siena poll put Ricketts six points ahead, one of his largest leads of the contest. Polymarket’s implied probability of a Ricketts victory had moved from roughly 67% to 73% after the bets, then reached 79% after the poll’s publication. The trader made more than $77,000 on the Nebraska market.

The same user made two early-morning Oct. 31 wagers totaling about $622,600 on Democrat Abdul El-Sayed winning Michigan’s Senate race. A Times/Siena poll released Nov. 1 showed El-Sayed with an 11-point lead, and he went on to win the race two days later. The account made more than $180,000 on Michigan and more than $250,000 across 25 Michigan and Nebraska trades placed over two hours.

The activity began after more than $2 million was deposited into an anonymous cryptocurrency wallet on Oct. 30, then routed to Kalshi and Polymarket over the following minutes. Before October, the Polymarket account had traded in only three markets. Large bets were subsequently made on similar Kalshi contracts, although Kalshi declined to comment on the activity.

Brett Goldstein, one of the Vanderbilt researchers, said the circumstances were difficult to dismiss as random and required further investigation. The researchers said several explanations remained possible, including knowledge of the polls, an ability to anticipate when they would be published, personal election views or trades designed to offset losses elsewhere.

Nansen and the Anti-Corruption Data Collective reviewed the Vanderbilt research and confirmed its overall findings, but said the evidence did not resolve the trader’s motive. The collective noted that the account held its positions through Election Day, whereas someone trading on confidential polling data might be expected to sell when prices peaked after a poll release.

The Times and Siena Research Institute said they took the allegations seriously and would review the research. The Times prohibits staff from using unpublished coverage for investments, while ReconMR employees are subject to confidentiality and nondisclosure agreements barring any use of unpublished data.

Polymarket said it had detected the trades immediately and was investigating them. As we reported last month, Polymarket had already introduced an election-market surveillance system that uses automated alerts and human investigators to examine potential insider trading and manipulation.

The episode adds to scrutiny of prediction markets, whose odds can move rapidly as users trade contracts on individual races. Kalshi has said it opened nearly 60 election-trading investigations since Jan. 1 and referred about half to law enforcement, though roughly 40 to 45 inquiries ended without a finding of wrongdoing.