Fake Wedgewood Polls Raise Questions About Kalshi Contracts

Wedgewood admitted fabricating most election surveys carried by VoteHub, a data source for Kalshi contracts whose rules exclude post-expiration revisions.

Wedgewood admitted on Oct. 20, 2025, that it fabricated most of its election surveys, including polls carried by VoteHub. VoteHub is the source for recurring Kalshi contracts, raising questions about settlements based on polling data later withdrawn or found to be false.

Researchers at FiftyPlusOne first questioned Wedgewood’s poll of the Mississippi Senate race after requesting the firm’s raw data. They found inconsistencies in respondent identifiers and transaction records. A comparison survey commissioned through Pollfish produced records that differed materially from Wedgewood’s data.

Wedgewood later acknowledged that most of its election surveys had been fabricated while identifying one Nebraska poll as legitimate. Zachary Donnini, VoteHub’s head of data science, said he had spoken with a Wedgewood representative before the results were added. VoteHub gave Wedgewood a D rating but included its polling in its database.

A Sept. 12, 2025, Kalshi filing for its VOTEHUBAPPROVE contracts identifies VoteHub’s time-adjusted polling average as the underlying data. The filing states that “revisions to the Underlying made after Expiration will not be accounted for in determining the Expiration Value.”

The rule fixes the settlement value using the data available at a specified time. It also states that if the relevant VoteHub average is unavailable, Kalshi will use the most recent eligible average before the specified date. The filing does not explain how that fallback applies when a poll is later withdrawn because it was fabricated.

Kalshi’s Oct. 20 generic congressional ballot contract, identified as KXVHGCB-26OCT20, uses VoteHub’s time-weighted Democratic-minus-Republican polling margin. The contract measures the margin shown at a specified observation time rather than the party that eventually wins an election.

The 2025 filing refers to the first available 10 a.m. Eastern reading after the specified date. The Oct. 20 market page lists 5:05 p.m. Eastern as the observation time. The rulebook for that contract series would determine whether a correction or dispute procedure applies.

VoteHub’s methodology describes two forecasting models. Its Complete model combines polling, political fundamentals, expert assessments and Kalshi prices. Its Legacy model uses polling and fundamentals. VoteHub’s methodology says that market prices can create feedback loops in forecasts. The polling average used for settlement is described as a separate product.

The public record does not show that a Kalshi contract settled using a VoteHub average affected by Wedgewood’s fabricated polls. The identified Wedgewood surveys involved state races, while the example Kalshi contract tracks a national polling average. It is therefore not established whether a Wedgewood result affected a contract’s settlement or a trader’s payout.

A separate August incident involved a Kalshi contract tied to a candidate that moved by about two cents within 15 minutes of a fabricated poll of the Los Angeles mayoral race by Median Strategies. Several Polymarket accounts traded in the same direction afterward. The account owners and any profits were not established.

The public documents do not identify which rulebook governs the October contracts, whether a contract has settled on a figure later corrected or withdrawn, or whether Kalshi can reopen a settlement after a source admits fabrication. VoteHub has not published which Wedgewood surveys entered its averages or when they were removed.

Articles by this author