The congressman who bet against himself
Stock markets settled this question a century ago. Prediction markets are working out what happens when the person whose behaviour decides the outcome is also allowed to trade on it.
Prediction markets have a structural problem that stock markets solved a century ago. What happens when the person whose behaviour determines the outcome is also allowed to trade on it? George Santos has just given regulators a textbook answer, and it cost him about $35,000.
The setup
Santos was expelled from Congress in December 2023, convicted of wire fraud and identity theft, and released in October 2025 after President Trump commuted his seven-year sentence. Ahead of February’s State of the Union he spent weeks publicly insisting he would be in the chamber, posting about his plans repeatedly.
Kalshi was running a market on who would attend the address, and Santos’s own confirmations helped move the odds. What the other traders did not know was that he held a position on that same market betting he would not show up.
The trade
Minutes into the speech, Santos posted on X that a cancelled flight had left him stranded at the airport. The market resolved his way. He walked away with roughly $17,570.
The catch
Kalshi caught it first. Robert DeNault, the exchange’s head of enforcement, confirmed the company flagged Santos’s trading activity to regulators itself. Given the scrutiny prediction-market platforms are under, that detail carries weight: the operator policed its own highest-profile political trader before anyone made it.
The CFTC’s case was not built on inside information in the classic sense. It centred on the public posts Santos made while holding the position, which the agency determined were misleading. That is a meaningfully different theory from a standard insider-trading charge. The violation was not what Santos knew. It was what he said in public while his own position depended on people believing the opposite.
The settlement
Announced July 31: Santos forfeits the $17,570 in winnings, pays a $17,500 civil penalty, and accepts a three-year trading ban. About $35,000 in total. Kalshi says it is separately pursuing its own exchange-rule enforcement action, and will look to reimburse affected traders if it recovers anything.
Why it matters beyond the headline
Every domain RadarPulse tracks (options flow, congressional stock disclosures, prediction markets) runs on the same premise: information asymmetry is supposed to be visible rather than quietly exploited. STOCK Act disclosures exist because members of Congress trading on non-public knowledge is a solved problem in theory, even where enforcement lags in practice.
Prediction markets are the newest venue for the same tension, and this case previews how it gets litigated. The question is shifting from “did you know something early” to “did you say something false while a market was pricing you as the source of truth.” That is a harder standard to argue against, because the evidence is public by definition.
Kalshi catching this in-house is the more interesting long-term story. A prediction-market operator that self-polices its highest-profile political traders is clearing a bar most exchanges have not had to. Kalshi separately banned a MrBeast video editor for insider trading in February, so this is not a one-off. Two cases are still not a policy, and whether it becomes the industry norm is the open question.
Settlement terms and figures as announced. Educational content, not financial advice.