The argument about vertical integration in prediction markets has been about the trading side. We wrote about that side in August, when the same proposal turned out to restrict an affiliated firm from trading on the venue that owns it. Section IV of the same document goes one floor down, to the clearing house, and almost nobody has read it. It is worth reading: prediction markets are the clearest live example of the structure it describes, and the comment window closes on 5 October 2026.
Five of the twenty four clearing houses have a member they own
The proposal is Conflicts and Affiliations, RIN 3038-AF76, issued by the Commodity Futures Trading Commission on 30 July 2026 (release 9274-26) and published at 91 FR 50926 on 6 August 2026. Section IV proposes three changes to the rules governing derivatives clearing organisations, the law's name for a clearing house.
The finding underneath it is one sentence long. Part 39 contains no definition of an affiliate clearing member, and of the twenty four clearing houses registered with the Commission, the proposal states that five have one. A fifth of United States clearing infrastructure runs a structure the rulebook has no name for.
The statutory hook is DCO Core Principle P, at 7 U.S.C. 7a-1(c)(2)(P), which tells every clearing house to minimise conflicts of interest in its own decision making. The Commission implements it through Commission Regulation 39.25, which covers conflicts generally and board member conflicts specifically, and says nothing about the case where the clearing house and one of its members share an owner.
The category the proposal does not name is the one that fits best
The proposal never mentions prediction markets in Section IV. It does not have to. Both large US venues are built exactly this way.
Kalshi Klear LLC has been a registered clearing house since 28 August 2024 (CFTC release 8957-24), and its affiliate KalshiEX LLC is the exchange whose contracts it clears. QCX LLC trading as Polymarket US clears through QC Clearing LLC trading as Polymarket Clearing, which states on its own site that it is a registered clearing house overseen by the Commission. In each case the exchange, the clearing house and the group that owns them are one commercial project.
A self clearing member is the trader, not a broker in between
The word member does hidden work here. In the futures market a clearing member is usually a broker, an FCM standing between the clearing house and the person holding the position. In prediction markets that layer is often absent by design. The Kalshi Klear rulebook filed with the Commission in July 2025 defines a member as an FCM or a self clearing member, the latter being a person other than an FCM that signed the member agreement. Polymarket Clearing uses the same shape in its rulebook filed on 31 December 2025, with direct clearing members alongside FCM clearing members.
Read the proposed definition against that. An affiliate clearing member is a person that is a clearing member of a clearing house and is under common control with it. A proprietary trading firm owned by the venue's parent, clearing its own positions directly, meets both halves with no intermediation at all. The entity Section III calls an affiliate market participant on the exchange is, at these venues, the entity Section IV calls an affiliate clearing member at the clearing house. One company, two rules, two bodies of the same document.
Four separations, and the fourth one is about people
Proposed Regulation 39.25(d) would require a clearing house to have procedures for identifying, addressing and managing conflicts involving an affiliate clearing member, covering four things at a minimum. Systems, so that applications and data do not leak non public information to the affiliate. Personnel, so that the clearing house does not share staff with it except for administrative functions. Office space, so the two do not sit in the same rooms. Documentation, so every conflict is written down with how it was resolved.
Three of those four are things you can buy, build or lease. The fourth is not, in the same way that board composition was the one control in this proposal that cannot be shipped. Separate servers are a procurement decision. Separate staff is an organisational decision about who answers to whom, and at a company of forty people it is an expensive one.
A drafting note worth a comment: the preamble on that fourth item says conflicts "with respect to an affiliate DCO" where the codified text says affiliate clearing member, and Appendix D to part 39 carries a heading naming the DCM inside an appendix that governs clearing houses. Both read as text carried over from the exchange half, the second time this rulemaking has shown a mismatch between preamble and rule text.
Both clearing houses already wrote most of this down themselves
Here is the part that deserves credit rather than criticism. Neither venue waited for the rule. Rule 2.18 of the Kalshi Klear rulebook permits an affiliated entity onto the platform and states flatly that "The Affiliate is a Self-Clearing Member of the Platform". It then imposes operational independence: separate servers, databases, accounts and tooling, no access to clearing house operations, no access to material non public information including order flow, trading, settlement and compliance, and a requirement to "Have and maintain physical separation from the Company". Rule 2.18(B) adds that capital behind the affiliate's trading will not originate from the clearing house.
Rule 3.1(i) of the Polymarket Clearing rulebook does the same job in eight numbered conditions, opening with the statement that affiliates "are permitted to become Clearing Members on the same terms and conditions as other Clearing Members". Its list covers non preferential access, controls against material non public information, physical separation of operations, no capital from the clearing house, no access to servers, databases or source code beyond what other members get, and representations about market integrity. Measured against proposed 39.25(d), that is most of the rule already in force, voluntarily, at both venues.
Neither affiliate rule says anything about shared staff
Compare the two lists item by item. Systems, covered at both. Physical separation, covered at both. Capital, covered at both and not even in the proposed rule. Documentation, partly covered by the board conflict minutes both rulebooks require. Personnel, absent from both.
This is a gap in a text, not an accusation. Neither Rule 2.18 nor Rule 3.1(i) contains a sentence about who the affiliate's employees are or whether any of them also work for the clearing house. Both rulebooks say elsewhere that staff overlap inside the group is contemplated. Kalshi Klear discloses it in Rule 2.13, telling participants that the company clears for the Kalshi Exchange, that "Company employees also work for the Kalshi Exchange", and that this "may present conflicts of interest". Polymarket Clearing's Rule 2.4(a) provides that "Any officer of the Clearinghouse may also be a director, officer, or employee of an Affiliate of the Clearinghouse".
Both disclosures are about the exchange, and the proposed requirement reaches the clearing member instead, so neither venue is out of compliance with a rule that does not exist yet. The point is that personnel is the one item of the four nobody has had a reason to write down, and the one that takes longest to change. A comment filed in October is cheaper than a reorganisation ordered in a final rule.
Full collateralisation is why none of this has bitten yet
The harms Section IV describes are harms of discretion. A clearing house sets margin, decides when a member is in distress, decides when to declare a default, and runs the default waterfall. Where it owns a member, the worry is that it sets margin generously, waits too long to declare a default, and leaves unaffiliated members' guaranty fund contributions to absorb the loss under Commission Regulation 39.16. The Futures Industry Association, quoted in the proposal, pointed to Alameda Research's exemption from FTX's auto liquidation rules as the cautionary case.
None of that machinery has existed at a prediction market clearing house. Polymarket Clearing says it clears only fully collateralised positions and "does not employ a margin-setting methodology or maintain a financial resource package", and Kalshi Klear was registered in 2024 for fully collateralised swaps. Where every position is prepaid in full there is no margin to set generously, no waterfall to sequence and no mutualised fund to favour anyone out of. The conflict was structurally absent because there was nothing to decide.
April 2026 built the machinery that makes the conflict real
That stopped being true this year, at the venue's own request.
On 15 April 2026 the Commission issued an Amended Order of Registration to Kalshi Klear, expanding the products it may clear beyond fully collateralised swaps. The order records representations that a margin methodology will be approved under Regulation 40.5 before use, and that "only clearing members that are either futures commission merchants or eligible contract participants are permitted to clear margined contracts".
Five days later the clearing house filed the rulebook to match. Its self certification of 20 April 2026 moves the rulebook from version 1.3 to 1.4 "to become effective after close-of-business on April 27, 2026", and lists the changes: definitions for products with different collateralisation requirements, a new Rule 3.2C on eligibility, "a comprehensive framework for future products that are not fully cash collateralized" in Chapter 7, and a new Chapter 12 establishing a default management framework. It does not name the affiliate provisions, so on the public record those carry over.
The timeline is unambiguous. In the same year the Commission asks whether clearing houses with affiliated members need safeguards around margin and default, one prediction market clearing house acquired the authority to set margin and wrote its first default management chapter. The conflict Section IV describes is no longer hypothetical there. It is scheduled.
The section carries twelve numbered questions, 60 to 71, and question 64 is the one to read twice. It asks whether the rule should address discretion when an affiliate clearing member is in distress, and gives three examples: declaring the affiliate in default, granting "any exemption from auto-liquidation rules", and departing from the default management playbook. It asks whether such decisions should be documented, reviewed by an independent committee and reported to the Commission.
Those three examples are the FTX failure written as a compliance question with the name removed. A venue that answers question 64 well has already decided, in writing and in advance, who signs off when the affiliate is the one in trouble. A venue that has not will decide it in the hour it happens, which is the hour when nobody's judgement is at its best.
Twelve alternatives went on the table and twelve came off it
Section IV.F walks through twelve alternatives and declines all twelve, inviting comment on each: banning the structure, supplemental capital, subordinating the affiliate's guaranty fund contribution, volume caps, pre approval of default decisions, a codified non preference principle, an independent reporting line for the chief risk officer and six more. The reasoning is about speed, because clearing houses need discretion to manage a default in real time. On volume caps the Commission agreed with the CME Group comment that "volume is not representative of risk", the same distinction between activity and exposure we drew when open interest fell in a record volume month.
What that leaves is a rule made almost entirely of procedure and disclosure. Public Citizen and Better Markets, quoted in the proposal, argued these conflicts cannot be mitigated at all. We do not agree, for the reason the Commission gives, but the answer to them has to be a demonstration, and a procedures rule with nothing published underneath it demonstrates nothing.
The disclosure half costs almost nothing and is still missing
The second proposed change is smaller and more useful. Proposed Regulation 39.21(c)(9) would require a clearing house to disclose the existence of, and its relationship with, any affiliate clearing member, moving the existing catch all category to (c)(10).
Regulation 39.21 already lists nine categories a clearing house must publish, from contract terms and clearing fees to margin methodology, the default resource package, daily settlement data, the rulebook and the list of clearing members. Affiliation is not among them. Someone deciding where to hold a position can look up the fees and cannot look up whether the firm on the other side of the book shares a parent with the venue holding the collateral. That is venue risk in its purest form.
It is also the cheapest item in the proposal. Both venues have the substance in their rulebooks already, and lifting one paragraph onto the public disclosures page, where Regulation 39.12 and Core Principle L point people, is a morning's work that does not wait for a final rule.
The comment window closes on 5 October and this file is short
Question 71 asks clearing houses with an affiliate clearing member to describe four things: the measures in place today, the disclosures made today and where they sit, the cost of complying as drafted, and the cost of each declined alternative.
For a prediction market clearing house that file is short, because most of it exists. The policies are Rule 2.18 or Rule 3.1(i), the disclosures are in the rulebook, and the gap is personnel. The Commission asked for costs so that it can calibrate, and a venue that stays quiet gets calibrated by firms that look nothing like it, which is how rules written for one shape of market end up applied to another. The same file answers question 64: if the affiliate is a member and the clearing house has just built a default management chapter, the sentence naming who decides, and who reviews that decision, either exists or it does not.
A structure you chose is a structure you have to be able to defend
Nothing in Section IV says a venue should not own its clearing house. The Commission looked at a prohibition and declined it, and we think it was right to. There is nothing dishonourable about a structure the statute permits and the Commission has registered twice.
What it does say is that the structure moves risk instead of removing it. The counterparty risk a reader assumes has been mutualised away is, at a venue with an affiliated member and a default waterfall, partly a question about how one company treats another it owns. That question has a good answer at both venues today, in rulebooks that go further than the regulation in force, and it has that answer because someone wrote it down before anybody asked.
The rule asks for the rest in writing by 5 October 2026. As with the incentive advisory in August, the comment quoted in the final rule is the one written by somebody who had already done the work.