Kalshi suspends three congressional candidates for self-betting

Three congressional candidates have been suspended from Kalshi after the platform determined they placed bets on their own races — a straightforward conflict of interest, and one that Kalshi's rulebook prohibits. The company made the suspensions public without naming the candidates.

The mechanics of why this matters are worth slowing down on. Prediction markets price probability through aggregated information. When a candidate bets on themselves, they are not contributing information to the pool — they are positioning on information the market cannot see. If the candidate knows something about their own ground operation, their internal polling, or a story about to break, their bet distorts the price in a direction that harms every other participant in that contract. This is not a gray area. It is the same asymmetry problem that makes insider trading illegal in equity markets.

The difference, legally, is that no statute currently governs this specific act in prediction markets the way securities law governs trading on material non-public information. Kalshi is acting on its own terms of service, not on regulatory mandate. That is both the right call and a problem: it means the deterrent is only as strong as Kalshi's enforcement, and enforcement requires Kalshi to know the account belongs to the candidate in the first place.

The consensus reading of this story is that Kalshi moved fast and did the right thing. I think that framing understates the structural exposure. Kalshi has spent eighteen months arguing to federal courts and state regulators that its markets are not gambling — that they are information-aggregation instruments with genuine social utility. That argument rests entirely on the quality of the information being aggregated. Three candidates betting on their own races is direct evidence that the market can be polluted by participants with private information and no obligation to disclose it. The suspended accounts are the known cases. The question of how many similar positions went undetected is one Kalshi cannot answer publicly without undermining the same integrity argument it needs for the preemption fights currently running in five states.

The model integrity problem in political prediction markets has always been distinct from other contract types. A trader with an opinion on copper prices does not have privileged access to the copper market. A congressional candidate has privileged access to the single variable — their own campaign — that the contract is resolving on. The two situations are not analogous, and pricing them under the same framework was always going to produce this.