One Roof, Four Registrations, Seventy‑Seven Questions

The proposal — styled “Conflicts and Affiliations,” announced July 30 and published August 6 at 91 FR 50926 — amends the rulebooks for all four registrant categories at once: Part 38 for futures exchanges (DCMs), Part 37 for swap execution facilities (SEFs), Part 39 for clearinghouses (DCOs), and Regulations 1.52 and 1.55 for futures commission merchants (FCMs). That scope is the tell. The CFTC is not regulating a firm; it is regulating a shape — the corporate family in which one owner holds the exchange, the clearinghouse, the retail broker, and sometimes a trading desk that quotes on its own market.

Chairman Michael Selig’s framing was deliberate: “principles-based regulations for vertically integrated market structures” that “support responsible innovation.” The preamble goes further, describing the exercise as codifying “sound practices responsible registered entities have already developed” into “a consistent regulatory baseline.” Read that carefully. The agency that in 2022 treated FTX’s direct-clearing application as an existential structural question is now writing the model into the CFR — which is simultaneously a legitimation and, for the first time, an enforceable set of terms.

The proposal has four load-bearing pillars:

Two things are conspicuously absent. There is no ownership cap — nothing prevents one company from holding all four registrations, and no equity-divestment or structural-separation requirement appears anywhere in the operative text. And there is no prohibition on self-clearing: the DCM-plus-DCO combination that defines every prediction-market vertical is accepted as given, and regulated at the level of procedures. The proposal’s 77 numbered questions do ask whether the Commission should go further — including toward outright prohibitions — but the proposed text itself is a building code, not a wrecking order.

The preamble counts roughly 27 DCMs, 24 DCOs, and 20 SEFs, and then gives the two numbers that matter: about eight DCMs have affiliated market makers — a structure the Commission says is “particularly prominent in prediction markets” — and about five DCOs have affiliated clearing members. This is a rule written for a population you can list by name. So let’s list it.

Notice what the table already tells you before any analysis: the “crypto-native venue” and the “prediction-market platform” are no longer distinguishable categories. Kalshi lists bitcoin perpetuals; Crypto.com’s Nadex-descended stack white-labels sports contracts; Robinhood’s prediction JV bought FTX’s old derivatives licenses; Coinbase distributes Kalshi’s markets while acquiring clearing infrastructure of its own. The July 30 proposal treats them as one population because they are one population — the same four-license shape, assembled between 2020 and 2026, mostly by acquisition.

Most of the proposal codifies things well-run venues already do — information barriers, disclosures, compliance reporting lines. One pillar does not. Proposed Reg 38.852 changes the economics of the most common liquidity arrangement in prediction markets, and it is worth being precise about how.

The baseline rule is a prohibition: a DCM may not have an affiliated principal trading firm active on its own market. The exception — the only door left open — is an affiliated market maker, on four conditions:

1. Filled last, always. The matching engine must not favor the affiliate, and the affiliate’s orders must be filled last at every price level regardless of time priority. This is not neutrality; it is mandated disadvantage. The affiliate loses the queue position that is the entire microstructural reward for quoting early.
2. A real market-making agreement. Continuous two-sided quotes, minimum trading hours, bid-ask spread limits, and — the clause that will get the most comment-letter ink — a prohibition on directional proprietary trading, on terms no more favorable than any unaffiliated market maker gets.
3. Someone else watches the books. An independent third party conducts the affiliate’s financial surveillance and certifies compliance annually to the CFTC and the exchange’s board.
4. The customer is told, every session. Notice of the affiliate’s existence and relationship before orders are entered, each trading session — point-of-transaction disclosure, not a page-40 rulebook footnote.

Why this lands on Kalshi first

Kalshi Trading LLC has provided liquidity on KalshiEX since June 2021 and was disclosed to the CFTC in September 2021. Per Sportico’s reporting, it has shared board members with the exchange itself. It is, functionally, the industry’s argument for why affiliated liquidity exists: a new event contract has no natural market maker, so the operator seeds it. The CFTC’s preamble concedes the point almost verbatim — “the operators of such markets believe that an affiliated market maker can be especially important in the creation and maintenance of new markets.”

The proposal accepts that argument and then makes it expensive. Under 38.852, Kalshi Trading survives only as a formal, obligated market maker: continuous two-sided quotes during specified hours, spread caps, no directional book, orders subordinated at every price level, third-party surveillance, per-session disclosure to every customer. Three of those six terms are straightforwardly costly:

• Subordination is a tax precisely where affiliated liquidity claims its value. In a thin, newly listed event market, the affiliate is often most of the book. Filled-last means every unaffiliated order at the same price jumps it — adverse selection concentrates on the affiliate exactly when it is providing the service the exception exists to preserve. The venue’s answer will be that seeding new markets was never about queue priority; the CFTC’s 77 questions effectively invite that data.
• “No directional proprietary trading” is a genuinely awkward fit for binary event contracts. In futures, a market maker’s hedged book is well-defined. In a yes/no contract, any net inventory is directional, and there is frequently no offsetting instrument. Two-sided quoting discipline is the only workable reading, and the rule text will need to say so — expect this to be a major comment theme from every prediction-market operator.
• The governance overlap has to end. Shared directors between exchange and trading affiliate is the exact pattern 38.853’s independence requirements and 38.852’s information barriers are written against. Whatever else survives, that does not.

Rothera: a venue designed around the thing the rule now regulates

The sharper case is the newest one. The Robinhood-Susquehanna joint venture that acquired MIAXdx (the old LedgerX licenses) and relaunched it as Rothera in January 2026 has the affiliation built into its capital structure: Susquehanna is simultaneously co-owner and the venue’s day-one designated liquidity provider, and Robinhood — the controlling owner — operates the affiliated FCM that routes retail flow to it. Susquehanna spent 2024–25 as the canonical unaffiliated institutional market maker on Kalshi; at Rothera the same firm is an affiliate, and everything it does on the venue runs through 38.852’s gate. A structure announced in November 2025 as its selling point — deep, committed, house-adjacent liquidity from day one — becomes, eight months later, the most heavily conditioned arrangement in the rulebook. Add Robinhood Derivatives as the affiliated FCM of the venue’s SRO and Rothera manages to trigger all four pillars simultaneously — the only group in the table that does.

The DCO pillar looks underpowered next to the market-maker rules, and the mismatch is informative. For an affiliated clearing member, the proposal requires conflicts procedures, information barriers, non-preferential treatment, and independent reporting lines for the chief risk and compliance officers — process, not structure. Davis Polk notes what is missing: nothing specific on margin methodology, skin-in-the-game, or default management where an affiliate is involved. The Commission knows it; the preamble’s alternatives section floats segmented skin-in-the-game and restrictions on mutualizing affiliate losses, and the questions ask whether DCO-specific guidance should be “more specifically tailored.”

Why so light? Because the prediction-market verticals that dominate the affiliated-clearing-member population — Kalshi Klear, QC Clearing, CDNA, Aristotle, ProphetX, Gemini — mostly run fully collateralized models. Every position is prefunded to its maximum loss; there is no margin call to shade in an affiliate’s favor, no default fund to mutualize, no procyclical haircut discretion. The classic clearinghouse conflicts simply have less surface area. A conflicts regime written as procedures fits a clearing model with no risk decisions to bias.

But that description is going stale in real time, and the venues themselves are the reason:

• Bitnomial has run margined perpetual futures since April 2025, accepts digital-asset margin collateral since September 2025, and clears through its own DCO for its own affiliated FCM. That is the full incumbent risk profile — margin methodology, collateral haircuts, default waterfall — inside a three-registration family. The floated skin-in-the-game and no-mutualization provisions read like they were drafted with exactly this structure in mind.
• Kalshi won approval for bitcoin perpetual futures in 2026 — the product that pushed CME into federal court against the CFTC. Leverage on a self-cleared venue converts Kalshi Klear from a prefunded settlement utility into a real clearinghouse, and pulls it from the light end of this pillar to the heavy end.
• Crypto.com’s Foris DAX FCM exists precisely to intermediate margined products on its affiliated DCM/DCO.

So read the DCO pillar as a placeholder with a fuse. The proposal regulates the self-clearing vertical as it looked in 2024 — fully collateralized event contracts — while the comment questions sketch the regime for what it is becoming in 2026: leveraged crypto derivatives cleared in-house. The gap between those two is where the second rulemaking lives.

The FCM pillar has the best origin story, and the preamble names it. In October 2024, NFA approved the FCM application of F&O Financial LLC — jointly owned by CME and an unaffiliated firm — which put CME in the position of serving as designated self-regulatory organization for FCMs that compete with its own affiliate. FIA objected loudly at the time. CME’s answer was to route surveillance of the affiliate to NFA voluntarily. The proposal takes that arrangement and makes it mandatory for everyone: an SRO with an affiliated FCM must hand the affiliate’s financial surveillance to an independent third party, wall off the affiliate from non-public information about competing FCMs, and give any FCM the right to elect NFA as its DSRO instead of the exchange it competes with.

Who inherits the chaperone: Coinbase (Coinbase Derivatives is an SRO; Coinbase Financial Markets is its affiliated FCM), Crypto.com (CDNA + Foris DAX), Bitnomial (Exchange + Bitnomial Clearing), Interactive Brokers (ForecastEx + IBKR LLC), Kraken once The Small Exchange relaunches against NinjaTrader, and Rothera (with Robinhood Derivatives as the controlling owner’s FCM). For most of them this is the easiest pillar — NFA oversight is a cost line, not a redesign. The interesting effects are second-order:

• It neutralizes the “we regulate our competitors” attack in both directions. Crypto venues complained for years that CME’s SRO status let an incumbent supervise its rivals; FIA made the same complaint about CME’s FCM. Mandatory third-party surveillance dissolves the argument for everyone at once — which, notably, protects the vertically integrated model from its most politically potent criticism.
• The information walls run against the grain of the pitch. The retail crypto full stack is sold as one seamless product: one app, one margin view, one balance. Reg 1.52’s walls don’t prohibit that product, but they do prohibit the exchange side from sharing non-public market data with the affiliated broker — the exact data advantage a cynic would say the stack exists to capture. Compliance here is invisible to customers and expensive to demonstrate, which is the worst combination.
• The disclosure pillar has a blind spot the direct-clearing venues walk through. Reg 1.55 disclosures travel through FCMs — but Kalshi, Polymarket, Gemini, Underdog, and ProphetX onboard retail directly, with no FCM in the chain. For them, the customer-facing transparency obligation arrives instead through 38.852’s per-session notice, which only fires if there is an affiliated market maker. A disintermediated venue with no house desk — Polymarket, again — ends up with the lightest disclosure load in the entire framework. Whether that is elegant calibration or an accident of plumbing is question-77 material.

The matrix makes the asymmetry legible. The proposal’s costs concentrate on the two groups whose liquidity is affiliated (Kalshi, Rothera), land moderately on the groups whose clearing and brokerage are affiliated (Bitnomial, Crypto.com, IBKR), and barely graze the group that outsourced clearing (Coinbase) and the group that outsourced liquidity (Polymarket). For CME, the incumbent the crypto venues define themselves against, the rule is close to costless — it codifies arrangements CME already runs. A regulation aimed at the newest structures ends up cheapest for the oldest one. That is not necessarily capture; it is what codifying incumbent best practice mechanically does. But it means the “maturation tax” falls on startups at exactly the moment they were converting regulatory arbitrage into regulatory legitimacy.

1 · The SEF asymmetry

The affiliated-trading-firm prohibition applies to DCMs only. SEFs get parallel conflicts procedures (new Reg 37.1201) but no prohibition — the Commission merely asks whether to extend it. Only one group in our population holds all three venue licenses: Rothera, whose stack includes the old LedgerX SEF. Nothing about event contracts moves to a SEF (they are DCM products), so this is not today’s loophole — but as tokenized swaps and institutional crypto derivatives grow, a family that can host affiliated liquidity on its SEF leg but not its DCM leg has been handed a routing incentive the rule text creates and does not resolve.

2 · Ownership without control

The framework triggers on affiliation — control-based definitions in Reg 1.52. It has little to say about influence below control: ICE’s stake in Polymarket (up to $2B, announced at a ~$9B valuation) makes the NYSE’s owner a major minority holder of a vertically integrated competitor-adjacent venue, and FanDuel’s prediction product runs on CME’s DCM in a partnership whose ownership mechanics remain partly opaque. If the conflicts logic is that economic alignment distorts venue behavior, minority stakes by the two largest exchange groups in America are squarely inside the logic and outside the rule.

3 · Distribution is out of scope — and distribution is the business

The 2026 prediction-market land grab is mostly a distribution war: Kalshi through Robinhood, Webull, Coinbase, and moomoo; CDNA white-labeled by Fanatics and Truth Predict; CME through FanDuel. None of these are affiliations in the rule’s sense — they are commercial contracts — so none of them are touched. The proposal polices the conflict between a venue and its house desk while the order-flow economics between venues and their mega-distributors, complete with exclusivity and revenue-sharing, remain governed by nothing more specific than general DCM impartial-access principles. The payment-for-order-flow debate of the 2030s is being seeded in these contracts now, unregulated by this rulemaking.

4 · One commissioner, one lawsuit, one comment period

All of this is being proposed by a Commission with a single confirmed member. Chairman Selig — confirmed in December 2025 after Brian Quintenz’s nomination collapsed and Caroline Pham departed — is running the agency with four vacant seats, a situation the House Agriculture Committee formally asked the White House to fix in May. Meanwhile CME is in federal court challenging the CFTC’s approval of Kalshi’s and Coinbase’s perpetual futures. A structurally significant rulemaking, proposed by a one-member Commission, with the industry’s largest incumbent simultaneously suing the agency, and 77 open questions inviting everything from prohibition to deregulation — that is not a stable equilibrium; it is an opening bid. Whatever finalizes will be shaped less by the proposed text than by who shows up by October 5, and every name in Figure 2 has both the incentive and the sophistication to show up.

Load-bearing primary sources

• The proposal: Conflicts and Affiliations, 91 FR 50926 (published Aug 6, 2026; RIN 3038-AF76; comments due Oct 5, 2026) and CFTC Release 9274-26 (July 30, 2026) — pillar structure, preamble counts (~8 DCMs with affiliated MMs, 5 of 24 DCOs with affiliated clearing members), the prediction-markets passage, the F&O Financial/CME DSRO passage, Selig quote.
• Rule mechanics detail: Davis Polk, “Vertical integrity” — the four market-maker exception conditions (incl. filled-last-regardless-of-time-priority), 35% board independence / fully independent ROC, the absence of DCO skin-in-the-game provisions, the SEF non-prohibition, and the 77-question count.
• Registrations: CFTC orders and releases — Kalshi Klear DCO (Aug 2024); Bitnomial Clearinghouse (Dec 2023); Polymarket–QCEX ($112M, July 2025); Crypto.com full stack (Sept 2025); Railbird DCM + QC Clearing arrangement (June 2025); Gemini DCO (Apr 2026); Aristotle DCM/DCO (Sept 2025) and Underdog acquisition (Mar 2026); ProphetX DCM/DCO (June 2026); Kraken–Small Exchange (Oct 2025).
• Kalshi Trading LLC: Sportico (active since June 2021, disclosed Sept 2021, shared directors). Rothera: Robinhood newsroom (Nov 2025) — SIG as co-owner and day-one liquidity provider; closed Jan 2026. ICE–Polymarket: ICE IR (Oct 7, 2025).
• Context: CoinDesk on CME v. CFTC (July 2026); FinanceFeeds naming Coinbase, Kraken, Polymarket in NPRM coverage and Coinbase’s opposition to forced separation; F&O Financial controversy per Risk.net and the FR preamble.

Confidence notes and deliberate hedges

Three points are stated with attribution rather than as settled fact. The four market-maker conditions and the 35% board figure come from Davis Polk’s reading of the proposed rule text (the Federal Register HTML truncates the operative provisions); their characterization is consistent with the preamble sections we could verify directly. The Kalshi Trading board overlap is Sportico’s reporting, not a CFTC finding. The preamble’s ~8 affiliated-market-maker DCMs are not itemized — our Figure 2 “none disclosed” cells mean no public disclosure was found, not that none exists; ForecastEx in particular we mark unclear. No formal comment letters from Kalshi, Polymarket, or DraftKings existed as of August 9; characterizations of their likely positions are our analysis.

Excluded as unconfirmable

A reported Kraken–Bitnomial arrangement and the ownership split of the FanDuel–CME brokerage were not confirmable against primary sources and carry no analytical weight above. The Clearing Company’s own license status post-acquisition is likewise unverified; we treat Coinbase’s DCO ambitions as signaled, not established.