Onchain data casts doubt on insider-trading theory in NYT's 'suspicious' election betting story

The New York Times reported Friday that Vanderbilt University researchers had flagged large, profitable bets on Polymarket and Kalshi that were placed before certain polls on the Nebraska and Michigan Senate races were released. The polls favored the candidates the trader had backed, raising questions about whether the trader had advance access to the results.

However, The Block's own analysis of onchain records and trading data found that the trader has a long history of both large, risky trades and complex DeFi yield-farming that was unreported by the Times. Matching onchain deposits to Kalshi trades also suggests that the trader may have placed two large bets 10 minutes after one poll in question was published. 

The trader held on to the bets all the way to Election Day rather than taking profits from short-term moves in the markets in the wake of the polls, and placed a bet on the race in Michigan while the Times/Siena polling was still underway, facts that the Times acknowledged in its report.

Those facts complicate the theory that the trader's profits resulted from advance access to unpublished Times/Siena polls. Two experts who reviewed The Block's findings said that while foul play can't be definitively ruled out, other explanations, like the trader being a crypto whale with a proclivity for large directional bets, fit the evidence better.

The Times report acknowledged it was unclear whether the trader had inside information. The researchers offered other possible explanations, and Polymarket told the paper it was conducting its own investigation.

Rajiv Sethi, an economics professor at Barnard College who reviewed The Block's findings, said the activity looked more like bets on the eventual election outcomes than an attempt to profit from the publication of the Times/Siena polls.

"I think these were just directional bets. Maybe they were based on non-public information such as privately commissioned surveys, but I don't think the Times/Siena polls were the information source," Sethi said.

What drew scrutiny

The Vanderbilt researchers identified a Polymarket user who bet $191,277 that Sen. Pete Ricketts, a Nebraska Republican, would win re-election. Most of that money was wagered in the early hours of Oct. 31, roughly an hour before a Times/Siena poll published at 5:05 a.m. ET showed Ricketts ahead by six points. The user ultimately made more than $77,000 on the race, according to the Times.

The same user bet $622,608 on a Democratic victory in Michigan's Senate race. At around 7:53 p.m. ET on Nov. 1, the Times published polls of Michigan and six other states showing Michigan Democrat Abdul El-Sayed ahead by 11 points. The user made more than $180,000 on that race when El-Sayed won the seat. 

The Block's analysis of Polymarket's public records matched those bets to an account named dubdubdub2. On Oct. 31, it bought $349,135 of "Yes" shares on a Democratic win in Michigan at 3:27 a.m. and $273,473 of "No" shares on a Republican win at 5:20 a.m., which are essentially bets in the same direction. Together, the two bets match the Michigan total reported by the Times.

The bets were unusually large for those markets. The account's Michigan purchases were the largest daily total by one account in any of the 16 Senate races The Block analyzed back to mid-September, and were nearly three times the next largest.

"There's a series of circumstances here that it's hard to write it off as a random occurrence," Brett Goldstein, who co-leads the prediction markets research team at Vanderbilt's Institute of National Security, told the Times. He said the activity "requires further investigation."

But while the Polymarket trades stood out on the platform, The Block's review of the trader's onchain history suggests the size of the bets was not unusual for the trader, and the specific details of some trades cut against the theory that they were exploiting advance knowledge of the polls. 

Michigan also wasn't the race the Nov. 1 polls moved most. That release covered seven states, and the biggest market reaction came in Kansas, where Polymarket's odds of a Democratic win jumped from about 52.5% to 60.5% within half an hour. In Michigan, where Democrats were already the strongest favorites at about 74.5%, the odds rose to about 79.5% over two hours.

The trader did not bet on Kansas. Its market was much smaller, however, with about $1.9 million traded on a Democratic win through the election, compared with about $7.9 million in Michigan.

Big Nebraska bets came after the poll's release

The researchers told the Times that the trader sent large sums to Kalshi and that large bets on similar markets followed, but the report didn't identify those bets.

The Block traced funds from wallets linked to the Polymarket account to a crypto payment provider used by Kalshi, then compared those deposits with Kalshi's public trade records. The records suggest the trader was betting on the same races on both platforms in the same directions, including possible bets placed after the Nebraska poll became public.

The account's Senate bets were funded by a separate wallet that also traded on Hyperliquid. The Block linked that wallet to the one controlling dubdubdub2 through a history of transfers between connected addresses. Its five transfers to dubdubdub2's Polymarket deposit address on Oct. 30 and 31, totaling $850,000, also match the account's deposit records to the dollar.

At 5:29 p.m. ET on Oct. 30, the betting wallet withdrew its remaining $2.2 million from Hyperliquid, where it had been trading over the previous day. About a minute later, it began sending funds to a deposit address that forwards money to a wallet labeled by blockchain analytics firm Arkham and Arbiscan as belonging to Zero Hash.

Kalshi lists Zero Hash among its crypto deposit providers. The betting wallet sent about $1.39 million to the deposit address through Oct. 31. The Block found no earlier transfers to Zero Hash from the trader's known wallets.

Kalshi's data does not link specific trades to specific accounts. However, since the transfers were consistent with Kalshi deposits and six-figure bets were rare in Kalshi's Nebraska and Michigan Senate markets, The Block was able to match certain deposits to trades by timing and size.

Several of the trader's Zero Hash deposits were followed within roughly a minute by large purchases in Kalshi's Nebraska Senate markets. On Oct. 30, a $175,000 deposit at 5:30:08 p.m. was followed 31 seconds later by about $159,000 in purchases backing Ricketts at 66 cents. A further $157,000 deposit at 5:31:58 p.m. was followed at 5:33:29 p.m. by about $169,000 in bets against Osborn at 67 cents. Both positions backed the same election outcome.

At about 5:15 a.m. on Oct. 31, 10 minutes after the Times/Siena poll was published, Kalshi recorded about $153,000 in purchases backing Ricketts to win and independent Dan Osborn to lose. The two groups of trades came eight seconds apart at prices around 75 cents, echoing the paired purchases seen earlier.

These trades are harder to attribute to the trader, however. They came about 50 minutes after a $250,000 deposit, while several earlier purchases followed deposits within roughly a minute. The deposit could also have funded bets in other markets.

If the trader did make those bets, continuing to buy in after the poll was published would suggest it was betting on the election outcome generally rather than trying to profit from the poll's short-term effect on the market, though late buys do not completely rule out the trader having advance knowledge of the poll.

Onchain transfers suggest the trader may have made more from its Kalshi activity than the roughly $258,000 its Polymarket Senate positions earned. On Nov. 4 and 5, its wallet received about $1.98 million from Zero Hash-labeled wallets, roughly $590,000 more than it had sent to Zero Hash in the first place, presumably as Kalshi deposits.

The transfers alone don't show exactly how much the trader made on Kalshi. The difference between the funds deposited and the funds received could reflect winning bets in other markets, trades that are missing from Kalshi's public records or money the trader originally deposited via a bank account.

A history of large bets and overnight trading

The Times reported that the user had traded in just three Polymarket markets before October, but its story did not detail what those trades were. The Block's review found those positions were also substantial, and that holding through settlement was a recurring strategy for the trader. 

The trader's dubdubdub2 account bet about $1.05 million on Zohran Mamdani winning the 2025 New York City mayoral election, about $102,000 against Andrew Cuomo in the same race and about $131,000 against the San Francisco 49ers winning the 2026 Super Bowl. It held all three positions until the markets resolved. The Block did not identify any bets that were placed shortly before mayoral polls were published.

Vincent Grégoire, a finance professor at HEC Montréal whose research includes insider trading, described the trader's history as "consistent with a wealthy individual (or small trading firm) who likes to take large bets on various platforms."

While the Times story notes that the trades were made early in the morning U.S. time, The Block's review of roughly two years of activity beginning in October 2024 found that 3 a.m. ET was the busiest hour for wallets linked to the trader. The analysis excluded Oct. 29 through Nov. 5, 2026, so the Senate bets themselves did not influence the result. It covered activity on Ethereum, Arbitrum, Base, Polygon, Hyperliquid and Polymarket, plus Solana bridge transfers.

The trader also has extensive experience within decentralized finance outside prediction markets, per The Block's analysis. The linked wallets have repeatedly farmed lending and yield protocols such as Aave and Pendle, adding up to tens of millions of dollars in total transfer volume. They have withdrawn more than $13 million in stablecoins from Kraken since October 2025. A linked Solana wallet also traded memecoins heavily in early 2025, including the official TRUMP token on its launch weekend.

The trader has also seen its share of trades gone wrong. Immediately before the election bets, the betting wallet deposited about $2.99 million in USDC on Hyperliquid and lost about $783,000 over 26 hours, including fees and funding costs, according to Hypurrscan's records for the account. The losses included about $217,000 on HYPE spot trading, $234,000 on a leveraged HYPE long and $311,000 on leveraged Intel perpetual futures positions opened after Intel reported third-quarter results.

The wallet behind dubdubdub2 also bought 60.43 million WLFI tokens for about $906,500 on the first day of the October 2024 public token sale by World Liberty Financial, the crypto venture owned by President Donald Trump's family, a purchase the Times also noted. The purchase made the wallet the largest buyer in the sale that month.

The Times said the purchase accounted for nearly one-third of all tokens sold that October. Onchain records reviewed by The Block instead put its share at about 6% of the roughly 970 million WLFI sold by the project that month, or about 9% of first-day sales. The Block has asked the Times about the discrepancy.

What the records can — and cannot — tell us

Grégoire said in an interview with The Block that large election bets can reflect a trader's forecasting models, private polls or other sources of information.

"To me, the simplest explanation is that there's no foul play," he said. "It's just someone that has other sources or other models."

Grégoire questioned why a polling employee would place a bet worth hundreds of thousands of dollars ahead of publication and then leave the position open rather than taking the immediate profit. He also noted that some of Polymarket's biggest historical winners made their money through a small number of large presidential election bets.

Sethi also said the use of Kalshi, which verifies traders' identities, suggests "they were not too concerned with concealing their identity."

Though The Block's analysis casts doubt on the idea that the trader had advance access to the Times/Siena polls, the reasons for the trader's bets cannot be established from the data alone. 

"It's almost impossible to tell from the data; you have to know who was trading and on what information," Grégoire said, noting that it's reasonable for the institutions involved to investigate whether their processes had been compromised.

Investigations and responses

A Polymarket spokeswoman told the Times that the company spotted the trades immediately and is investigating. Kalshi declined to comment to the Times.

The Times also reported that spokespeople for the Times and the Siena Research Institute said they took the allegations seriously and planned to review the research. Times spokesperson Charlie Stadtlander said the newspaper prohibits employees from using unpublished coverage to make investments.

Lyle Durbin, CEO of ReconMR, the data collection firm that owns the Siena Research Institute, said employees are bound by confidentiality and nondisclosure agreements prohibiting the use of unpublished data for any purpose, according to the Times.

World Liberty Financial declined to comment through a representative. Kathleen Clark, a spokesperson for Vanderbilt's Institute of National Security, told The Block that Goldstein would not be able to comment. The Block also sought comment from Polymarket, Kalshi, the Times and Siena but has not heard back as of publication time.

The Times' scrutiny comes as prediction markets face broader questions about their level of oversight over their own platforms. Both Kalshi and Polymarket tightened their insider trading rules earlier this year amid scrutiny from the Senate. 

In April, the Justice Department charged U.S. Army soldier Gannon Ken Van Dyke with using classified information about the operation to capture former Venezuelan President Nicolás Maduro to make about $409,881 on Polymarket. A former White House teleprompter operator agreed in August to pay $172,000 to settle CFTC charges that he used advance knowledge of Trump's speeches to bet on Kalshi's "mention markets."