A Leg Nobody Played Settles at Whatever the Exchange Decides

On a four-leg combination contract, three legs can resolve exactly the way the buyer picked and the fourth can resolve to nothing at all. A player is a late scratch, a game is called off before the whistle, a governing body declares a no-contest. The leg did not lose; it never got the chance to be measured. Something still has to enter the product that decides the payout, and as of 13 November 2026 both American exchanges listing these contracts answer the same way: the leg is assigned a value between zero and a dollar, set by the exchange.

That is defensible, because forcing an unmeasurable leg to zero turns an unmeasured event into a loss. The problem is narrower: of all the moving parts in a combination contract, the number that decides the payout is the only one neither exchange has written a method for.

Five months ago the arithmetic itself was still in dispute

Kalshi's help page on combos, read on 13 November 2026, sets the rule out plainly. The payout is the product of each component's value; a component for a player who never enters the game resolves to its last fair price as determined by the exchange, not to zero or a dollar; and the worked example is one leg at seventy cents times two legs at a dollar, paying seventy cents. A non-appearance, it adds, does not by itself refund or cancel the position, as set out in Kalshi's combos help article.

Polymarket US did not start there. Its original self-certification for the Combinatorial Outcome Contract, filed with the Commodity Futures Trading Commission on 20 May 2026 under 17 CFR 40.2(a), carried one payout bullet. Under the heading Joint Probability it said the contract resolves to a dollar if and only if every leg is satisfied, and that a single unsatisfied leg sends it to zero regardless of legs that have not settled yet. No provision covered a leg landing in between. The filing is public as the exchange's 20 May 2026 product self-certification on the Commission's filings page.

The gap closed on 14 July 2026. In an amendment filed under 17 CFR 40.6 and published by Polymarket US, the exchange added a bullet called Constituent Cancellation: where a constituent contract settles to fifty cents or to last fair market prices under its own terms, the matching leg is treated as having settled at that value. On 14 September 2026 the bullet was renamed Non-Standard Contract Outcome, given a worked example and told to round the product down to the nearest minimum tick. The version in force was filed on 22 September 2026.

So the divergence this site described in September 2026, when one exchange multiplied scalar legs and the other forced them to zero, has gone. Both now multiply, which is an improvement, and it moves the open question one layer down: from the formula to the number going into it.

Each of the three change notices reads identically

All three CAOC amendments carry the same single line under Summary of Amendments: clarifying the terms and conditions of the contracts as well as the explanation and analysis, including clarification on source agencies, payout criterion, and settlement procedures. Word for word the same on 14 July, 14 September and 22 September 2026, as the exchange notices index published by Polymarket US shows. From that summary you would not know that one of the three invented the rule for what a combination pays when a leg never trades out.

Attachment A does go up clean and redlined, so a reader who opens it sees the insertions. Attachment B, the analysis of why the amended product complies with the Commodity Exchange Act, is confidential under 17 CFR 145.9, so the public file holds the new rule and not the reasoning behind it.

The payout number rests on a phrase neither rulebook defines

Kalshi's contract terms are explicit both about where the value comes from and about its being a judgment. In the terms for football statistic markets, a player who is active but takes no part in a single play has all contracts resolve to the last fair price as determined by the exchange in its sole discretion. That phrase appears ten times in a document of roughly 2,430 words, covering non-appearance, cancellation before kickoff, abandonment before fifty-five minutes of play, a delay past forty-eight hours, a pre-kickoff forfeit, a venue change and a format change. It is in the contract terms Kalshi publishes for football entity statistic markets.

How widely that wording is used is measurable. Our own count on 13 November 2026, built from the public series endpoint of Kalshi's trading API, found 16,340 series pointing at 3,179 distinct contract terms documents. We drew a random sample of 300, read 298 and found the phrase last fair price in 60, with 178 occurrences in total: roughly one sampled document in five routes some contingency to a price the exchange sets at its own discretion. That is a sample, not a census, describing documents rather than how often the clause fires.

Polymarket US does the same thing under a different noun. Its Athletic Statistic Contracts, self-certified on 31 March 2026, settle to last fair market prices in eight separate situations: non-participation, an event not rescheduled within two weeks, cancellation, season truncation, a suspended or abandoned event, a forfeit, a format change and a declared no-contest. Non-participation is the only one stated flatly; five of the others say the exchange may, in its sole discretion, settle at last fair market prices or initiate its Contract Outcome Review Process.

Neither exchange defines the term in the document defining everything else. Kalshi's DCM Rulebook version 1.31, which its public document store reports as last modified on 4 November 2026 and which runs to about 50,134 words, contains fair price zero times and fair value zero times. The Polymarket US Rulebook dated 30 September 2026, about 35,451 words, contains fair market price, fair price and fair value zero times each, and minimum tick zero times too, although minimum tick is the unit the combination terms round down to.

A written cascade does exist, in the settlement chapter

Kalshi's Rule 6.3(c) is the closest thing the category has to a method. Where it cannot be determined from the payout criterion what proportion of the settlement value should go to long and short positions, the exchange may use the contract's last traded price; its worked example is a contract last traded at ten cents paying ten cents long and ninety cents short. If no last traded price is available, or if the exchange decides at its sole discretion that the most recent one does not represent a fair settlement, the rule hands the question to an Outcome Review Committee for a binding determination of fair allocation.

That committee is the most interesting structural object here. Under Rule 2.7(e) it has three members appointed by the Regulatory Oversight Committee, at least two of them Public Directors, on one-year terms, and Rule 2.7(c), which otherwise preserves the board's authority over its committees, carves this one out expressly. Rule 7.1 gives it a clock: a final outcome within a twenty-four hour period, with the exchange posting on its website that the outcome is under review. That is the thirty-five percent public director floor doing concrete work, and a reminder that some protections can only be built out of people.

The same rulebook shows that a bounded price is writable. Rule 5.11 on trade cancellations defines a fair market value in advance, from the last price at the time of the questioned trade plus other market information, then draws a twenty cent band either side inside which a trade stands. Same problem, written method, in the same document.

One caveat. Rule 6.3(c) addresses the case where the payout criterion yields no determinable proportion, and for a player who does not play the contract terms do address the contingency; they simply name a price instead of a proportion. Whether that routes into 6.3(c) is not stated in either document: the terms do not cite the rule and the rule does not mention the clause.

On the other exchange the same decision sits with four job titles

Polymarket US routes its equivalent through Rule 10.4. Before settlement, the Chief Executive Officer, Chief Compliance Officer, Chief Operating Officer, Head of Markets or their designate may at their sole discretion review circumstances that may materially affect the reliability or transparency of the underlying event, after which the company may determine the final outcome and may reverse one for obvious error. Participants are notified afterwards, including by a posting to the website.

There is no deadline, no standing committee, no independence requirement and no published method for the number. Nothing in the public record suggests abuse, so this is a design choice rather than misconduct, but it is thinner than the committee next door for a decision of the same kind. It also sits beside the clause that lets either exchange change a contract's terms after the first day of trading, so exposure to exchange judgment is cumulative.

Appendix C already asks for the settlement price to be documented and published

The Commission has written down what it expects from a cash settlement price, just not for this case. Appendix C to Part 38, its guidance on showing that a contract is not readily susceptible to manipulation, says at paragraph (c)(4)(i)(C) that the cash settlement price should be reliable, acceptable, publicly available and reported in a timely manner, and at (c)(3)(iv) that the registered entity should fully document that the price is accurate and reliable. Paragraph (c)(3)(v) covers prices drawn from non-public sources: the exchange should make available to the public, as soon as possible after expiration, the final cash settlement price together with any other supporting information that is appropriate or feasible to release. All of it is in Appendix C to Part 38 in the Electronic Code of Federal Regulations.

A price read off the exchange's own book with no published method is a non-public source in the sense that paragraph means. The guidance is guidance, written for commodity indices rather than for a scratched quarterback, so nobody is in breach of a rule. The measuring stick exists, and asks for what this clause does not supply: a documented method and a published number. Our own count on 13 November 2026 across the four prediction market documents of 2026, from the full texts of the June 2026 proposal on prediction markets and public interest determinations and its three companions, 195,723 words in all, found fair price, last traded price and sole discretion zero times each.

Massachusetts turned the dead leg rule into pre-approved, published text

One regime has answered the question already, in a narrower setting. Under the Massachusetts Gaming Commission's rule 205 CMR 247.02(3)(o), an operator's house rules must address, at a minimum, its policy for when an event or any component of one is cancelled or suspended, including the handling of wagers with multiple selections such as parlays where one or more selections is cancelled. The operator may not take a wager until the Commission has approved those rules, may not change them without prior written approval, and must post them prominently, state their effective date and give previous versions to any patron who asks in writing. The text is in 205 CMR 247.02 as reproduced in Cornell Law School's regulation archive, read on 13 November 2026.

The point is not that one regulator is stricter, since a designated contract market carries obligations a sportsbook never will, and this site has looked at what state reporting does and does not capture about combinations. The point is where this one rule lives: in Massachusetts, a mandatory item of a dated document approved before the first bet; on a federally regulated exchange, a bullet in a self-certified filing effective after filing, with the analysis withheld, naming a price rather than a method.

The bounded wording is one product family away

Kalshi's covered player markets do the most useful thing here, which is to name a moment. In the terms for its protected NFL statistic markets, a player who takes part in at least one play in the first half and none after it has the contract resolve to the last fair price prior to the opening kickoff, unless the statistic already satisfies the payout criterion. The discretion survives; the timestamp removes most of what a reader could not check. Before trading, a buyer knows the price comes from the pre-game market rather than a thin in-game one. The clause is in the contract terms Kalshi publishes for covered NFL entity statistic markets.

Polymarket US has written the other missing half for a different kind of change. Its Consumer Price Index contracts, self-certified on 6 April 2026, allow an alternative source agency only where the original source suffers a material disruption, discontinuation or change, and promise that any such change will be communicated to market participants prior to settlement. A named trigger and notice before the money moves. The fair market price clause three bullets away has neither.

Four things a venue could publish before the next late scratch

None of this needs a rule change. All four could be self-certified under 17 CFR 40.6, the way the payout rule itself was.

  • The moment the price is taken from. Pre-game close, last trade before the official inactive list, a stated clock time. The covered player clause shows it fits in one phrase.
  • The fallback when there is no trade to read. Last traded price, then book midpoint, then a committee, in a stated order, as Rule 6.3(c) does for its own two steps.
  • Who decides, and by when. A named body with a deadline, as Rule 7.1 has, rather than a list of officers and no clock.
  • The number afterwards, and how often it happened. The value used and the supporting information, which is what paragraph (c)(3)(v) asks for anyway, plus a count of contracts settled this way in the period.

A late scratch belongs on the schedule, not in force majeure

Both exchanges file the non-appearance clause alongside cancellations, forfeits, abandonments and no-contests, which reads as a list of disasters. It is not. A player ruled out is an ordinary weeknight across every league, and on combination contracts it is the likeliest way for a position to be priced by judgment instead of measurement. Grouping the routine case with the rare ones is part of why it has gone this long without a method: a clause that looks like force majeure does not invite anyone to ask how the number is computed. That is why being right about the event is not the same as being paid for it, and why the version of this rule that survives will be the one written before the scratch rather than decided after it.

Share X LinkedIn Email