Kalshi asks the CFTC for perpetuals, and names a DAO as price source

What happened

KalshiEX LLC submitted two perpetual futures contracts to the Commodity Futures Trading Commission on 18 August 2026 under section 5c(c) of the Commodity Exchange Act and Commission regulation 40.3(a), which is a voluntary submission for Commission review and approval rather than the self-certification route a designated contract market normally uses. Both submissions say the exchange 'intends to list this Contract shortly following approval by the Commission'; the Commission's own industry filings register marks each Approval Pending (45), while Kalshi's other product filings of the same day, a set of sports event contracts, are marked Certified. Each submission is signed by Xavier Sottile, head of markets, and comes with terms and conditions, enhanced trading prohibitions, an account of the source agency, two appendices on the core principles, a confidential appendix containing the exchange's agreement with the source agency, and a request for confidential treatment under freedom of information law. The US500 contract is a perpetual future on the MerQube US Large Cap Index, a float-adjusted, capitalisation-weighted index of the 500 largest companies listed and domiciled in the United States, administered by MerQube and disseminated as MQ5C. Because the underlying is a broad-based security index it is an excluded commodity under section 1a(19) of the Act, so the contract is not a security future and sits inside the Commission's exclusive jurisdiction under section 2(a)(1)(C); the submission records that no component exceeds about 7 per cent of the index weight and the five largest together account for about a quarter. It trades from 6pm Eastern on Sunday to 5pm Eastern on Friday, with funding measured only during the 9:30am to 4pm regular equity session and a single daily funding rate struck at the close, and it carries position accountability at $25m of mark-to-market value rather than a hard limit. The COPPERPERP contract is a perpetual future on the spot price of copper quoted in dollars per pound, referencing the Pyth Network XCU/USD price feed, trading on the same weekday schedule and adopting the federal spot-month position limit that applies to the enumerated COMEX copper contract. Both submissions devote several pages to Chicago Mercantile Exchange Inc. v. Selig, 1:26-cv-02157, filed in the District of Columbia on 18 June 2026, in which CME argues that a perpetual contract is a swap rather than a future. Kalshi's answer quotes CME's own comment letter of 21 May 2025 saying that 'certain perpetual-style derivative contracts may accurately be classified as futures', its 2011 letter on the preservation of customer choice between the two categories, and a Seventh Circuit holding on contracts of indefinite duration in a case, as the filing notes, 'to which CME was the very beneficiary'.

Why it matters

The finding is in the appendix on the source agency, and it is the clearest statement yet of what a regulated market will and will not accept from an oracle. Pyth is described exactly as its designers would wish: first-party institutional publishers rather than scraped quotes, a confidence-weighted median in which each publisher gets three votes and none can determine where the aggregate lands, a new aggregate every 200 milliseconds, and every submission and aggregate signed and time-stamped so that the exchange or the Commission can reconstruct any published value after the event. That last property is offered as better than what offshore venues do, and it is. Then the appendix removes the part that makes the arrangement cryptoeconomic. Oracle Integrity Staking, under which publishers stake PYTH against their own data and can be slashed for failing defined thresholds, had its reward rate set to zero by governance action OP-PIP-103 in April 2026, and the filing tells the Commission that the mechanisms 'are preserved in the protocol but are not currently operative' and that 'The Exchange does not rely on Oracle Integrity Staking as an operative safeguard for the Contract'. What it does rely on is a permissioned roster: contributors admitted only if their data achieves a normalised root-mean-square error below 0.01 against LSEG's Elektron service, re-evaluated daily, added and removed on measured uptime and accuracy. Read plainly, a CFTC-registered exchange has told a federal regulator that the token-slashing layer is not the safeguard and that the safeguard is a vendor benchmark and a gatekeeper. The network does not describe it that way, and both descriptions are primaries. Pyth's own post of 23 April 2026 states three times that staking and slashing remain live, that publishers and delegators who keep stake in the programme remain subject to slashing, and that 'Enforcement does not require ongoing emissions - it requires stake at risk'. The same post also says that with the reward rate at zero 'For most participants, the straightforward path is to unstake', which is an instruction to withdraw the collateral that the sentence before says enforcement requires. And it records the fact that reconciles the two accounts: across the whole life of the programme 'zero slashing proposals were submitted to the Pyth DAO', which the network offers as 'the cleanest statement available about data quality during this phase'. A penalty never once imposed is being read by its designer as evidence that nothing needed penalising and by its largest new customer as evidence that nothing is being penalised. The second thing worth reading is the route. Self-certification under regulation 40.2 lets a designated contract market list a product on its own word; 40.3(a) hands the decision to the Commission and stops the exchange listing until it decides. Kalshi used the second for these two and the first for its sports contracts on the same day. Asking permission is a legal strategy when the last permission is being challenged in court: an approval order issued after full review is a harder thing to unwind than a certification, and the submissions read as briefs as much as filings.

What is not settled

Nothing is approved and no timetable is published. The register marks both Approval Pending (45), and forty-five days from 18 August is 2 October 2026, which is also the date on which CME must oppose the government's forthcoming motion to dismiss under a minute order of 13 August; the defendants' response to that complaint was extended the same day to 2 September, so on the morning Kalshi asked for approval no court had ruled on whether a perpetual is a future at all. The reports do not agree about what the suit challenges. The Block describes a June lawsuit 'challenging the CFTC's approval of perpetual futures contracts for Kalshi and Coinbase'; Finance Magnates describes a challenge to the May approval of Kalshi's bitcoin perpetual and to the accompanying policy statement. The docket names only the Commission and its chairman as defendants and the filings identify the challenged acts as that order and that policy statement; Coinbase appears in neither. The numbers Pyth publishes about itself are not reconciled anywhere. The filing says the network publishes data for more than 1,000 assets and warns that publisher counts and update figures on Pyth's public site 'relate to the prior generation of the network (Pythnet) and may differ from current production figures'; Pyth's own front page this morning states more than 3,000 price feeds and 138 publishers. Pythnet, the chain those feeds have run on since 2021, is itself being retired under OP-PIP-100. The document that would say what happens when the feed fails is the exchange's agreement with the source agency, and it is the confidential appendix, withheld under a request for confidential treatment, so no reader can see who bears the loss if an aggregate is wrong. The audit trail establishes what was published and by whom; it does not establish liability. And nothing in either filing addresses what a copper hedger's recourse is against a network governed by a token, or against the Marshall Islands company organised to hold its treasury.