The Final Ledger: The Four Contracts That Outlived the Exchange

On 23 July, BitMEX told the world it was going to kill itself, and it gave the date in the same sentence: 23 September at 04:00:00 UTC. Not a restructuring, not a pause while a strategic review runs its course. Shut down. The exchange I co-founded in 2014, the venue where I built the perpetual swap in 2016, announced its own demolition and then, with the bureaucratic courtesy of a well-run exchange, posted the timeline for how its own corpse would be processed.

I have spent eleven years thinking about this company's plumbing — how its margins, funding and settlement hold together. So let me do what I always do with a balance sheet story: walk the entries, find the forced actors, and tell you who loses what. Because the last move BitMEX makes is a beautiful one. It involves four futures contracts whose settlement dates outlive the exchange itself.

The freeze

Here is the sequence, straight from the wind-down FAQ. Between 26 August and 23 September, risk limits go on and the book flips to reduce-only: traders can close positions, but they cannot open new ones. Think about what that does. The entire remaining book becomes one-sided. Every order that is still resting on the day the freeze hits is a pill nobody can take back, and the only allowed action is selling (or buying) to get flat. Liquidity in the thin corners of that book — and there are thin corners — is about to get very honest.

Then, on 2 September at 12:00 UTC, BitMEX delists eleven perpetual swap contracts, settles them early, cancels all open orders. These are the perps with insufficient trading interest, the ones the market had already stopped caring about. Notice the order of operations: the exchange is shedding instruments in ascending order of liquidity. The emptiest contracts go in August. The four contracts that matter — the four that outlive the venue — are the ones nobody can shed, because their settlement dates are on the far side of the funeral.

And at 04:00 UTC on 23 September, with the exchange now less than an afternoon old in death, the system force-closes every position still open and settles it at the relevant Settlement Price or index, standard procedures. Deposits that arrive after the closure time are non-recoverable. Balances left sitting on the corpse are charged a monthly fee of 1% per annum, or fifty dollars, whichever is greater. BitMEX is going to keep billing you from beyond the grave, just like a landlord who hasn't noticed the building burned down.

The four that outlived it

Now the part I actually care about. BitMEX was, and in its dying weeks still is, running dated quarterly futures. Under normal circumstances each one walks to a scheduled settlement date — the contract converges to the index, the long and the short settle, done, dusted, on with life. BitMEX's closure breaks that assumption for exactly four contracts.

The September 2026 batch — XBTU26 and ETHUSDU26, the BTC/USD and ETH/USD quarterlies — are scheduled to settle on 25 September 2026. That is two days after the exchange dies. The December batch, XBTZ26, settles on 25 December 2026. The March 2027 batch, XBTH27, settles on 26 March 2027. Four contracts, four settlement dates, every one of them on the far side of 04:00 UTC on 23 September. Every one of them run past the closure date.

BitMEX already dealt with this once. On 10 August it delisted and settled the XRPU26, ADAU26, ETHU26 and XBTUSDTU26 futures — four quarterlies originally scheduled to settle 25 September — because the exchange closes before they mature. "Delist and settle early," the notice said, in the cheerful passive voice of an exchange guide. Fine. But that was the easy cohort. The four that remain cannot be early-delisted in the same tidy way, because two of them — XBTZ26 and XBTH27 — are six months to a year short of natural maturity. Force-settling those was never about "delisting"; it is about confiscating time.

Here is the accounting. A March 2027 bitcoin future embeds the carry — the time value that separates a forward price from the spot market. The whole trade in dated futures is that carry: roll it, harvest it, hedge it. When the venue force-settles XBTH27 on 23 September against the spot index instead of letting it walk to natural settlement in March, that remaining carry is crystallized on the venue's schedule, at the venue's chosen reference price, with six months of term left unearned. Longs who were carrying the contract forward lose the right to hold it further. Shorts get their hedge terminated by someone else's clock. The market participants who chose a March contract because they wanted March exposure are getting September exposure and being told to like it. That is a transfer, executed by the exchange's own funeral home.

And the reduce-only freeze makes it worse. From 26 August, the only way out of these contracts is to sell into a market where every other participant is also only allowed to sell. The marginal buyer does not exist in a close-only book; he has been legislated out of existence. Whatever open interest is left in those four contracts on 23 September gets matched at the settlement index, period, and the people holding it eat whatever the mark says it is worth. The exchange is doing, at the very end, exactly what I argued for years a derivatives exchange should always do: mark to spot, not to the price of your own derivative. BitMEX is force-marking its own orphans to the index because there will be no venue left to run the auction that would have priced them fairly later.

Why the perp made these contracts obsolete

Now let me tell you why this is funny, and it requires no hindsight.

I invented the perpetual swap in May 2016 specifically because dated futures have this disease. A quarterly has a maturity, and a maturity is a rendezvous: at some point every single one of you bastards has to show up and settle, whether you want to or not. The perpetual has no maturity. It settles continuously — funding every eight hours, pinned to the index forever — so it never runs past anything. It cannot be orphaned. It cannot be force-settled early for the simple reason that it is always already settled. When BitMEX seals the doors on 23 September, the perps die cleanly at the single click of a force-close, because there is no future date on which they were supposed to be refined. The dated futures, the old technology my own instrument was built to kill, are the only things leaving debris on the floor.

The perp market, incidentally, is now estimated at $85 trillion a year in traded volume, and the establishment has spent the past two years eating its own words: listing houses and exchanges that once treated crypto derivatives as casino products have started shipping perps themselves, which is exactly what I argued they would have to do eight months ago when I wrote that traditional exchanges must adapt or die. The product outlives the venue. BitMEX — the exchange that popularized the 100x perpetual swap, that crossed $3 trillion in cumulative volume at its peak, that went eleven years without losing a cent of customer money to a hack — is being buried while its invention takes over the world. The company lost; the contract won; the index never even noticed.

Which brings me to the autopsy, because I do not do sentiment, I do causes. BitMEX did not die of a decline in bitcoin, and anyone reading this as crypto's death certificate is reading it wrong. Bitcoin is trading around $76,000 as I write, up roughly 20% over the past month, up about a quarter over three months, with the fear-and-greed dial at 66, firmly in greedy territory, and bitcoin dominance near 59%. Total crypto market cap sits around $2.6 trillion. This is not a dying asset class; this is a risk-on regime. BitMEX died of a self-inflicted wound dressed as a strategic review: years of accumulated regulatory settlements north of $200 million, a sale that fell through, and a token that plunged toward a 98% drawdown. The state and the balance sheet killed the exchange, not the market it invented. The marginal price-setter in bitcoin moved off BitMEX years ago and never noticed the obituary sheet was out.

I will be honest about what I could not find. I could not pull the exact open interest left in XBTU26, ETHUSDU26, XBTZ26 or XBTH27 at the moment of writing, and it may well be trivial — this is a footnote inside a funeral on a venue that stopped mattering for price discovery a long time ago. The magnitude of the dollars at stake is not the point, and I refuse to pretend otherwise. The point is structural: these four contracts are the only instruments that cannot be closed cleanly at a timestamp, and their existence is the final exhibit in the case for what replaced them. If you have any size in them, the correct move is not to leave it to the forced mark on 23 September. Close before the freeze on 26 August, into a two-sided market, at a price you chose. And get your cash off the corpse — a 1% annual fee, minimum fifty dollars a month, on money sitting with a company that just fired its own exchange is not idle money, it is a slow-motion haircut.

Here is the epitaph, and it is a good one. On 23 September at 04:00:00 UTC, an exchange that built the most durable trading instrument of its generation will die at a single moment, cleanly, marking the entire book to spot — and the only contracts to outlive the exchange will be the dated futures, the very technology the perpetual was invented to replace, kept alive just long enough to prove their fatal flaw one last time. The building comes down on schedule. The perpetual that I built inside it does not even feel a draft.