Four information points. No docket number.
Four information points, and not one of them is a primary source. No Federal Register citation. No CFTC filing reference. No quote from the entity's own documentation. A New Zealand exchange says it intends to offer perpetual futures on individual stocks and describes the purpose as bringing leveraged equity trading "into DeFi." That is the entire payload.
I have spent enough years pulling apart contract layouts to know what an absence of primary sources implies. It implies one of two states: the filing has not yet surfaced in the public record, or the narrative has outrun the artifact. Both are observable. Neither is fatal on its own. What is fatal is reading a headline as though it were a specification.
Read the assembly, not just the documentation. The documentation here is one sentence. The assembly — the actual mechanism by which a perpetual contract tracks a stock through dividends, splits, halts, and the weekend — is absent entirely. That absence is the only genuinely analyzable object in this story, and I intend to analyze it.
What a perpetual actually needs from its underlying
Perpetual futures have no expiry date. A dated future converges to spot at settlement because the delivery obligation forces convergence; a perpetual has no delivery, so it requires a synthetic tether. That tether is the funding rate — a periodic cash flow between longs and shorts, sized by the premium of the contract price over the index price. If the perpetual trades above the index, longs pay shorts. If below, shorts pay longs. Mechanically, it is a proportional controller with a sign flip, sampled on a fixed interval.
The design works in crypto for three reasons that are easy to overlook because they are always true here: the underlying market never closes, the index price is a continuously computed composite across many venues, and the reference asset has no corporate life cycle that can mutate it. Strip away any one of those and the controller is being asked to regulate a plant it was not tuned for.
Now substitute a single equity for BTC. The controller does not change. The plant does.
Single-stock futures are not a new idea. The United States banned them in 1982 under the Shad-Johnson Accord, largely over margin and jurisdictional anxieties between the CFTC and the SEC; the Commodity Futures Modernization Act of 2000 lifted the ban. The products that resulted existed, traded thinly, and largely died. OneChicago — a joint venture of the CBOE, the CME, and the CBOT — was the flagship venue, and its volume never validated the thesis that traders preferred a futures wrapper over listed options on the same names.
That is the base rate. Any new entrant is not proposing a novel instrument; it is proposing a resurrection with a perpetual wrapper attached. And the wrapper is precisely where the engineering becomes interesting, because the perpetual structure was built for assets that do not pay dividends, do not split, do not halt, and do not have a board of directors empowered to decide on a Tuesday that the thing you are holding is now two different things.
Tracing the logic gates back to the genesis block
Start with dividends. A long position in a perpetual has no ex-dividend date, no record date, and no payment date. The holder never appears on the issuer's register. So the index price the contract tracks is either the raw exchange-quoted price — which drops by the dividend amount on the ex-date, mechanically, with zero relation to supply and demand — or a total-return series that includes distributions.
If the contract tracks the raw quoted price, every long loses the dividend value at the ex-date through no action of their own, and the funding rate must be adjusted to compensate. That adjustment makes funding a calendar-driven event with predictable, scheduled jumps. If the contract tracks a total-return series, the exchange must define that series, compute it, and defend it against the awkward fact that nobody else quotes it. The first turns the funding formula into a calendar function. The second reintroduces the exact oracle dependency the perpetual design was supposed to compress.
Splits are worse in a specific way. A two-for-one split halves the price and doubles the quantity. A perpetual contract has no quantity to double. It has a notional and a multiplier, and the multiplier is a term in a legal document, not a variable in a runtime. So the exchange must either amend the contract — a governance event with counterparty consequences — or define an adjustment factor applied at the index layer. The second is cleaner, and the second also means the contract's economic identity is now a function of an exchange-maintained parameter set. That is a trust assumption wearing an engineering costume.
Then there is the market-structure mismatch, and this is where I expect the design to strain hardest. Crypto perpetuals inherit their stability from a 24/7, globally fragmented, continuously arbitraged spot market. US equity spot markets are open roughly six and a half hours on weekdays, with pre- and post-market sessions of varying and venue-dependent liquidity, and full closure on weekends. What does the funding rate do at 4:05 p.m. ET on a Friday?
If the answer is "nothing until Monday's open," then the contract is not perpetual in any economically meaningful sense. It is a sequence of dated segments, and the funding controller is off for roughly two-thirds of the calendar week. Adverse news arriving over the weekend becomes un-arbitrageable until Monday's bell — which means the perpetual can gap, and the funding rate cannot correct until after the gap has already printed. The controller is now sampling more slowly than the disturbance. That is textbook instability, not a corner case.
If the answer is "funding continues, priced off a proxy," then the exchange must define a proxy for price during closure. Last trade. A stale quote. A futures-curve interpolation. An overseas listing, if one exists for that name. Every one of those is an oracle with a manipulable surface. My Synthetix v1 work in 2020 — six weeks simulating flash-loan-driven decoupling between a protocol's price feed and spot reality — taught me a specific lesson. Oracles do not break randomly. They break precisely at the boundaries where the underlying market's own price discovery is thinnest: off-hours, illiquid names, calendar edges. A stock perpetual builds those boundaries directly into its contract terms.
Liquidation engines inherit the same defect. A liquidation engine needs a mark price it can trust at the exact instant it must act. If the mark is stale, the engine either over-liquidates against a phantom price or delays and accumulates bad debt. Both behaviors are already observed in crypto cascades. On an underlying that closes every weekday night and every weekend, every close is a potential cascade window.
Borrow cost is the next unstated term. A short position in a perpetual is not a short position in a stock. In theory there is no locate, no borrow fee, no buy-in risk. In practice, whoever takes the other side of the book does need to hedge, and hedging a short equity exposure requires borrow or a synthetic substitute. That hedge cost has to appear somewhere in the funding formula. If it does not appear, the exchange is warehousing basis risk on its own balance sheet — which is a capital problem before it is a product problem.
Corporate actions beyond dividends and splits close the loop. Mergers convert one contract into a different one. Delistings terminate the reference asset outright. Tender offers dislocate the price relative to the index. Spin-offs create fractional entitlements with no clean futures representation. Each event forces either a term amendment or an index-layer adjustment, and each adjustment is an opportunity for disagreement between the venue and its counterparties.
One more comparison is worth holding alongside all of this. Equity contracts for difference — offered in Europe, the UK, and many offshore venues — are economically close to a stock perpetual: leveraged, funding-adjusted, no expiry. The payoff function is not novel. What is nominally novel here is a US-registered wrapper, and that is a compliance distinction, not a mathematical one. The invention is jurisdictional.
None of the above is unsolvable. Every problem has an engineering answer. All of it is undocumented. That is the finding: a product whose entire value proposition is mechanical innovation has disclosed zero mechanics.
The word that does not compile
The counterintuitive element is not the perpetual. It is the word "DeFi."
A CFTC-facing derivative available to US participants is, by construction, permissioned, intermediated, identity-gated, and cleared through a regulated venue or a registered foreign board of trade. Those are not incidental properties bolted on afterward; they are the compliance requirements. DeFi, used with any precision at all, means permissionless access, non-custodial settlement, and pseudonymous participation. The two sets do not intersect. A product cannot require identity verification and simultaneously be permissionless, because verification is a gate and a gate is a permission.
So "bringing equity leverage into DeFi" is either marketing or an error. It cannot be a description.
There is a second, subtler problem hiding in the phrase "applying for CFTC approval." US derivatives listing does not generally work that way. A designated contract market lists products through self-certification: the exchange certifies compliance with the Commodity Exchange Act and CFTC regulations, the filing sits for a window, and the product goes live. A foreign board of trade can solicit US participants under a different exemption path. "Approval" is a verb that belongs to banking charters and state licensing, not to the standard route for a futures contract. If the filing is real and the word is accurate, there is a structure I have not reconstructed. If the word is loose, the reporting is.
And single-stock futures sit under shared jurisdiction by design. Shad-Johnson split the territory precisely so securities and their derivatives would not slip through a single regulator's floor. A story that mentions only the CFTC may be simplified in retelling, or may be flagging a genuine jurisdictional question that has to be resolved before anything lists.
What to watch
Watch the Federal Register, not the press release. The verifiable signals are narrow: whether a filing appears at all, which regulator's docket it lands on, whether the entity registers as a DCM or claims foreign-board relief, and whether product terms are published with funding formulas and corporate-action procedures attached.
Absent those, this is a narrative position, not a product. The fragility worth flagging is not that one filing fails. It is that the convergence trade has learned to use compliance vocabulary as a pricing mechanism, and the market has learned to read a filing as a launch. Each such announcement compresses the distance between statement and expectation. The gap between expectation and artifact is where retail capital gets consumed.
