Kalshi Settled a Game Before It Ended, Then Un-Settled It. The $50 Credit Is the Whole Argument.

Editor's note: TrueEdge builds odds tools and earns affiliate commissions from licensed sportsbooks. This piece argues a position on prediction-market regulation, and our commercial interest runs toward the sportsbooks. Weigh it accordingly.

At roughly 11 p.m. Eastern on Saturday, the clock in Ann Arbor read zero, Western Michigan players were on the field celebrating a 12-7 win over Michigan, and Kalshi paid them. Not the players — the traders. The exchange settled its Western Michigan–Michigan contract as a Western Michigan win and moved the money.

Then the officials came back from replay, put one second on the clock, and Bryce Underwood threw a 47-yard touchdown to JJ Buchanan. Michigan won 13-12.

The market had traded $18.6 million. Kalshi clawed back what it had just paid, reimbursed the people it had just marked as losers, paid the traders who had actually been right, and gave the users whose winnings evaporated a $50 credit. CNN's Marshall Cohen, who got the confirmation, laid the sequence out in three steps in that order. One Kalshi account had a $2,265 Western Michigan position that briefly showed a payout of $93,201.

The dollar figure is not the story. Kalshi did $2.29 billion in volume on Saturday, an all-time single-day record, with $2.02 billion of it in sports and combination contracts. Against that, $18.6 million is a rounding error and the correction cost the company very little.

The story is that it had to decide.

What an exchange is supposed to be

Kalshi's argument to the Third Circuit, to the Ninth, to the Massachusetts Supreme Judicial Court and eventually to the Supreme Court is not that sports contracts are harmless. It is structural: Kalshi is a designated contract market under the Commodity Exchange Act, its rules are filed with the Commodity Futures Trading Commission, its settlement terms are published in advance, and a nationwide financial exchange cannot be supervised by fifty separate gambling regulators applying fifty different standards. The rulebook is the thing that makes it a market. A bookmaker decides; an exchange executes.

On Saturday night the rulebook produced an answer and Kalshi went the other way.

Traders noticed within minutes. The reading circulating in the replies — and it is a trader's reading, not a court's — is that Kalshi's filed contract terms bar the use of a post-expiration revision to change a contract's Expiration Value. If that is right, the market expired on the exchange's own determination, the Western Michigan holders were paid under the rules as written, and the remedy for Kalshi's error was to pay the Michigan holders out of its own pocket rather than to reach into settled accounts.

We cannot tell you whether that reading is correct. The KalshiEX rulebook is on file at cftc.gov, we downloaded it, and the text is not extractable from the version served there. We are not going to characterise a federal filing we could not read. What we can say is that Kalshi has not published the provision it relied on, has not published a rationale, and resolved a disputed settlement question overnight by fiat and a credit.

The Michigan part is not a coincidence, but it is a coincidence

Five days before all this, on September 1, Ingham County Circuit Court Judge Rosemarie E. Aquilina signed a preliminary injunction barring Kalshi from offering, listing, executing or settling sports contracts to anyone in Michigan, including anything "functionally similar" to sports betting — parlays, moneylines, in-game markets, props. Kalshi must geofence the state using a third-party provider licensed by the Michigan Gaming Control Board. Violations carry $500,000 a day.

Aquilina wrote that Michigan and its most vulnerable citizens face immediate and irreparable harm from what she called a sports betting operation "masquerading as an investment opportunity." Attorney General Dana Nessel sued in March on behalf of the state and the Gaming Control Board under the Lawful Sports Betting Act. A temporary injunction came first, in late June, at $120,000 a day. September's order converted it and raised the number more than fourfold. Michigan now sits alongside Nevada and Washington as states where Kalshi's sports contracts are dark.

So the game Kalshi mis-settled was played in a state where a judge had just ordered it to stop settling games. That is a coincidence of geography and nothing more, and we are not going to pretend otherwise. But it does mean the exchange's worst operational night of the season landed on the desk of every attorney general currently drafting a brief about whether these are markets or bets — and it handed them a fact pattern rather than a theory.

The strongest case against us

Here is the other side, and it is good.

Michigan won the game. Anyone holding Western Michigan at expiration was wrong about the world. The money they briefly received came from a clock-reading error inside a trading system, not from a correct forecast, and a market that pays out on its own glitch and then refuses to fix it is not a more principled market — it is a worse one. Every regulated sportsbook in America operates under palpable-error provisions that let it void obviously wrong prices, and nobody calls that lawlessness. Correcting toward the true outcome is the conservative choice, not the radical one. The $50 was a courtesy for the inconvenience of a reversal, and the alternative — Kalshi eating an eight-figure loss to reward people who backed a team that lost — would have been a windfall, not justice.

That is right about the outcome. It is not responsive to the objection.

The point of a filed rulebook is that it binds in the cases where it produces an answer you dislike. That is the only case in which having one matters. A sportsbook's palpable-error clause is defensible precisely because a sportsbook does not claim to be something other than a firm exercising commercial judgment, and because a state regulator with subpoena power reviews the calls. Kalshi's whole position is that it is not that, that it does not need that, and that the CFTC's exclusive jurisdiction is adequate because the terms are set in advance and applied mechanically. An overnight reversal, with no published provision and no route to appeal, is a commercial judgment. It may well have been the right one. It was still discretion, exercised by an interested party, on its own money, with no reviewer.

We spent last week arguing that the CFTC clearing product categories by telephone was a worse arrangement for the exchanges than the one they are describing to the courts. This is the same defect one level down. Regulation by request at the agency; settlement by request at the venue.

What would change our mind

If Kalshi publishes the rule it applied — the provision, the internal timeline, who authorised the reversal and when — the transparency objection largely goes away, and this becomes an ordinary operational failure handled competently under terms that were on file the whole time. Exchanges break. NYSE has busted trades. The question is never whether an error happened but whether the fix was traceable to a rule.

And if the CFTC's event-contracts rulemaking, proposed June 10 and still not final, ends up specifying settlement-correction procedure for sports contracts, then the gap that let this happen closes on its own and the criticism expires with it.

Neither has happened yet. What happened is that an exchange settled a football game while the officials were still reviewing it, and then decided — reasonably, unilaterally, and without showing anyone the rule — to settle it again.