A product announcement lands. It names two companies — Crypto.com and OG.com. It names three product features: perpetual futures, a 5x maximum leverage, and no expiry date. It names a market: the United States. Four factual points. No fee schedule. No list of tradeable assets. No regulatory registration number. No named contracting entity. No state-by-state availability map. That is the entire disclosure.
It is being described, in the first coverage cycle, as a compliance breakthrough.
I pulled this announcement apart the way I pulled apart Uniswap V2's router contracts in 2019 — three weeks of decompiled opcodes, tracing the token transfer logic by hand until an edge case in the reserve math fell out. Back then, the contracts answered questions if you were willing to read them. Here, there is nothing to read. The bytecode didn't ship. Only the press release did.
That asymmetry is the story.
Perpetual futures are not a new category. BitMEX shipped the modern form in 2016, and the engineering has been stable for years. A well-built perpetual has five moving parts: a mark price derived from an external index, a funding rate that pays one side of the order book to keep the contract anchored to spot, a tiered liquidation engine, an insurance fund that absorbs the first layer of bankruptcy loss, and auto-deleveraging as the last resort when the fund cannot cover the tail. Every venue that runs perps runs these same modules. Crypto.com has run them in offshore markets for years.
The question is never whether a firm can build the engine. The question is what legal wrapper the engine sits inside.
In the United States the wrapper matters because a perpetual has no delivery. The CFTC's retail commodity framework has a delivery carve-out — roughly, if actual delivery of the underlying occurs within 28 days of the trade, the transaction can avoid being treated as a regulated futures contract. A perpetual has no delivery date by construction. It cannot ride the carve-out. It has to route through a designated contract market, or through a futures commission merchant that intermediates for the retail client. Those are the two doors. Everything else is an offshore structure wearing a domestic address.
This matters because it determines the legal character of the exposure. A margin spot position is a loan collateralized by assets the user holds. A CFD is a bilateral contract where the dealer is the counterparty. A regulated future is a cleared exchange contract where the clearinghouse stands between the parties. The three structures look identical on a trading screen and are completely different in a bankruptcy. The user's recovery depends on which one they are actually in, and the announcement does not say.
The 5x ceiling suddenly makes sense as a routing artifact, not a product decision. 100x is what an offshore order book prices because there is no supervisor asking why. 5x is what a regulated venue's risk parameters look like after a margin model has been reviewed by someone upstream. The number tells you which door was used.
It does not tell you who holds the deposit.
The Funding Rate Is the Real Cost
Retail users anchor on leverage. It is the visible number, printed in every advertisement, comparable across every venue. What does not get printed is the payment that accrues every funding interval for as long as a position stays open.
Funding on perpetuals has a direction. When the contract trades above spot, longs pay shorts. In a bull market, that skew is persistent — the crowd is long, the funding is positive, and the long side pays rent on every interval. On major pairs in hot conditions the annualized rate can run well into double digits, occasionally much higher during cascades.
A quarterly future has a fixed basis. The user pays one roll cost at expiration and it is legible. A perpetual has a variable cost that resets every interval and is deducted from the margin balance in small invisible increments. Over six months of holding a directional position, the difference can exceed the entire slippage budget of the trade.
That is the whole meaning of "no expiry." It isn't the absence of a cost. It is the replacement of a one-time cost with an open-ended one. The marketing line and the cost structure are the same sentence, and they are being sold separately.
The direction of that cost is not neutral. In a bear market funding can flip negative and shorts pay longs. But the customer base for a newly launched US retail perp skews directionally long in a bull cycle. The base case for the average user is paying the funding, not receiving it — and paying it in a stream that never terminates.
I watched this pattern from the inside during the stETH drawdown. Six months spent mapping exit latency in Lido's withdrawal process taught me that the design flaw in stress was never the fee itself — it was the user's mental model of the fee. A fixed cost that becomes variable in a crisis is the single most expensive surprise in DeFi. Perp funding is that structure applied to directional bias, and it is not a flaw. It is the product.
The 5x Ceiling Is a Bandwidth Constraint, Not a Safety Feature
A 5x position goes to zero on a 20% adverse move. Twenty percent moves happen. They happened in May 2021, in November 2022, and in the first correction of every bull cycle since. The difference between 5x and 100x is not the probability of the wipeout. It is the speed. At 100x, a 1% move takes the margin. At 5x, a 20% move takes the margin. The position still dies. It just dies slower.
What 5x actually does is cap the notional of a single position relative to posted collateral. That is a real constraint on the size of any one trade. It is not a constraint on the cumulative loss of a strategy. A trader who takes the same directional position three times after three stop-outs at 5x has taken the same total loss as a trader who took one position at 15x and stayed in it. The ceiling applies per position, not per account, not per strategy, and not per year.
It also does not prevent a cascading liquidation inside the venue. It shrinks the notional per position, which means more positions are needed to build the same notional in the book. A large number of small, similarly margined positions is not obviously safer than a small number of large ones. It is more granular, which is good for the liquidation engine's matching. It is also more correlated, which is not.
The honest reading of the ceiling is that it is not tuned to reduce a retail user's worst-case outcome. It is tuned to make the product legible to a risk committee. A family office can explain 5x to an investment policy statement. It cannot explain 100x. That is a real function. It is just a different function than "protects the user," and the copy writes the second one.
The Mark Price Is the Number That Decides Who Gets Liquidated
Every perpetual is built on an index. The index produces a mark price, and the mark price decides when a position is liquidated. This is the single most consequential parameter in the entire system. It is also the parameter most frequently left out of public descriptions.
A well-constructed mark price falls back to a composite: multiple spot venues, weighted by volume or liquidity, passed through a median or a trimmed mean, with a deviation band that freezes the index temporarily when any one source prints a tick outside the band. The freeze exists to prevent a single bad print from triggering a mass liquidation event across every open position in the venue. Without it, one illiquid minute on one exchange becomes a chain reaction in every perpetual market quoting that index.
A poorly constructed mark price is a single source's mid. A wick on that source is a wick on every position. I have audited these. The difference between a two-second index refresh and a one-hundred-millisecond refresh is five figures on a large position in a fast tape. The difference between a three-venue composite and an eight-venue composite is a material shift in the probability of a spurious liquidation during a flash move. Neither setting is a marketing term. Both are load-bearing.
The announcement says nothing about which index is used, how it is weighted, which venues contribute, what the deviation band is set to, or how frequently it refreshes.
The most consequential parameter in the product — the number that determines whether a user's position survives a bad minute — is the one the launch omits.
That omission is not unusual. Very few venues publish index composition at launch. But "not unusual" and "acceptable" are different categories. In a product where the index determines liquidation, the composition belongs in the contract terms, not the marketing.
The Counterparty Is Missing From the Launch Announcement
Here is the part that made me put the laptop down.
The announcement names two companies. It names no legal entity a user would actually contract with. No jurisdiction of incorporation for the contracting party. No regulatory registration number. No dispute resolution forum. No named custodian of client funds. In a properly licensed product, those are not supplementary disclosures — they are the first disclosures, because the license is the differentiator. A firm with a designated contract market registration prints the registration. A firm operating as a futures commission merchant prints the registration and the capital requirement. The registration is the marketing.
Two names, no defined roles. Who holds the license? Who clears the trade? Who custodies the margin? Who appears as counterparty if the client has a dispute? The announcement does not answer any of these.
A dual-brand launch with no named contracting counterparty is not a product disclosure. It is a logo placement.
In 2024 I reviewed over 200 smart contract functions for a Layer 2 project against MiCA requirements, checking whether KYC/AML logic was embedded at the protocol layer or merely bolted to the front end. Three privacy-layer gaps turned up, and the final report changed a $2 million grant structure. The first question in that engagement — the one every reviewer asks before anything else — was who the licensed entity was. That question came before the code review, because it determined whether the code review mattered.
OG.com is the sharper edge. The announcement does not state its business category. There is no public exchange infrastructure background I can find. No matching engine history. No regulatory footprint. If OG.com is a distribution partner — a front end with reach — then the integration is a conversion funnel, and the 5x ceiling sits in tension with the messaging an affiliate channel would normally run. If OG.com is the licensed venue, then it is a regulated entity I cannot verify, which is a different category of problem.
I am not accusing anyone of anything. I am noting that the field is blank and that the blank is load-bearing. A due diligence process cannot proceed until it is filled. We didn't get a fee schedule. We didn't get a state map. We didn't get a client agreement. We got four sentences and two brand names, and the legal structure of a leveraged financial product cannot be inferred from brand names.
Liquidity Is the 90-Day Number That Will Actually Define This
A perpetual with no disclosed depth and no named counterparty is a promise. The verification is the open interest curve after ninety days of live trading. That curve has a shape.
Days 1 through 14: a zero-fee or reduced-fee incentive is in effect. Volume is high because volume is subsidized. The number looks excellent and it is meaningless — it is measuring the incentive, not the demand.
Days 30 through 60: fees normalize. The market makers who supported the launch with inventory begin pricing continuous inventory risk. Spreads widen at the wings of the book. Depth concentrates around the touch. This is not a failure. This is the market discovering the real cost of making a two-sided market in a low-volume product.
Days 90 through 180: the number that matters is open interest, not volume. Open interest is the stock of live positions. It is the number that determines whether the liquidation engine has enough counterparties to absorb a cascade without drawing on the insurance fund. Volume can be manufactured with rebates. Open interest cannot — it requires someone to hold a position through a funding cycle, take the cost, and stay in it.
I watched the same curve in Balancer V2 vaults during DeFi Summer 2020. I ran a Python script against the vault contracts to monitor rebalancing costs in real time, and the pattern held across every pool: the winner was never the pool with the highest headline yield. It was the pool whose swap parameters held up after the incentive emissions went to zero. The metric that predicted survival was the one that persisted after the subsidy ended.
In both cases, the metric that matters is the one that survives the subsidy.
The volume print at day 14 is the subsidy. The open interest print at day 120 is the product.
The Insurance Fund and ADL Are Mechanisms, Not Promises
Every venue with a perpetual has an insurance fund. It exists to absorb the loss when a liquidated position's bankruptcy price is worse than its liquidation price — the case where a user's loss exceeds their posted margin. When the fund is depleted, the venue reaches for auto-deleveraging: it forcibly closes the winning side of the book at the bankruptcy price to cover the deficit.
ADL is not a bug. It is the last resort that keeps the exchange solvent. But it operates without consent and without a market price. A user whose winning position is ADL'd gets closed at the venue's chosen number, not the market's.
I have followed the ADL debate for years, and the single most important thing about it is that it is a disclosure problem before it is a design problem. A venue that publishes its insurance fund size, its historical depletion events, and its ADL trigger thresholds is a venue where users can price the risk. A venue that publishes none of those things is a venue where the risk is socialized silently. On the current disclosure, this product is in the second category.
The Compliance Wrapper Reclassifies Risk, It Does Not Reduce It
The consensus read on this launch is that it is a step forward for the onshoring thesis: Crypto.com is doing the boring hard thing, the 5x ceiling is a maturity signal, and US retail finally gets a regulated perpetual.
I want to argue against one dimension of that read. The compliance wrapper does not reduce user risk. It reclassifies it.
A user trading 50x at an offshore venue is exposed to market risk and platform credit risk. A user trading 5x at a regulated venue is exposed to market risk, platform credit risk, and a third thing that nobody prices — regulatory event risk. If a regulator changes a rule, restricts a product category, or brings an enforcement action, the product can stop functioning while positions remain open. Withdrawals can acquire new documentation requirements. The dispute forum has moved. The offshore user at least knows they have no regulator. The regulated user may discover that the regulator is a variable that changed mid-position.
There is also a visibility shift. A regulated venue cannot quietly absorb a bankruptcy. That is a benefit. It is also a disclosure obligation — once the engine is inside a regulatory perimeter, any failure to disclose a change in funding parameters, index composition, or ADL events becomes a supervisory matter rather than a customer-service one. The 5x ceiling means a smaller share of the retail base will ever experience ADL. It does not mean ADL will not run. The mechanism fires when the fund is depleted. A pool of 5x positions exhausts a small fund as efficiently as a pool of 100x positions exhausts a large one. The ratio matters more than the leverage number.
Then there is the question of audience. The real market for this product is not the retail directional trader. It is the registered investment adviser, the family office, and the small self-directed fund that cannot access an offshore venue, cannot justify a high-leverage position to a risk committee, and needs a hedge that does not require rolling. 5x fits the risk committee parameters. That is what the ceiling is for. It is not primarily a retail protection — it is a format an allocator can read. And if that is the intent, then the metric that matters is institutional open interest, which the announcement does not mention because the product does not have it yet.
That is the contrarian read: the ceiling is an interface, not a guardrail. It was designed for the desk that will eventually allocate, not the trader who will open the first account.
The version of this product that succeeds is completely defensible. A named contracting entity, a published regulatory registration, a mark price index with disclosed composition, a fee schedule, a state availability map, and a custody statement — that is a solid product, and it would earn the compliance label honestly. The version that fails is the one currently live on the page: two brand names, four facts, and a counterparty nobody can name.
What I am watching is not volume. Volume is a marketing number and it is subsidized for a window. I am watching three disclosures that should appear in the first thirty days if the product is what it claims to be: the named contracting entity and its registration, the composition of the mark price index, and the identity and role of OG.com. Each of them is trivial to publish. Each costs nothing if the answer is clean. Silence on any one of them is itself the answer.
Volatility is noise. Architecture is the signal. On this launch, the architecture is four sentences long, and two of them are brand names.