Perpetual futures, explained

Programming note: Bits about Money is supported by our readers. I generally forecast about one issue a month, and haven't kept that pace that this year. As a result, I'm working on about 3-4 for December.

Much financial innovation is in the ultimate service of the real economy. Then, we have our friends in crypto, who occasionally do intellectually interesting things which do not have a locus in the real economy. One of those things is perpetual futures (hereafter, perps), which I find fascinating and worthy of study, the same way that a virologist just loves geeking out about furin cleavage sites.

You may have read a lot about stablecoins recently. I may write about them (again; see past BAM issue) in the future, as there has in recent years been some uptake of them for payments. But it is useful to understand that a plurality of stablecoins collateralize perps. Some observers are occasionally strategic in whether they acknowledge this, but for payments use cases, it does not require a lot of stock to facilitate massive flows. And so of the $300 billion or so in stablecoins presently outstanding, about a quarter sit on exchanges. The majority of that is collateralizing perp positions.

Perps are the dominant way crypto trades, in terms of volume. (It bounces around but is typically 6-8 times larger than spot.) This is similar to most traditional markets: where derivatives are available, derivative volume swamps spot volume. The degree to which depends on the market, Schelling points, user culture, and similar. For example, in India, most retail investing in equity is actually through derivatives; this is not true of the U.S. In the U.S., most retail equity exposure is through the spot market, directly holding stocks or indirectly through ETFs or mutual funds. Most trading volume of the stock indexes, however, is via derivatives. 

Beginning with the problem

The large crypto exchanges are primarily casinos, who use the crypto markets as a source of numbers, in the same way a traditional casino might use a roulette wheel or set of dice. The function of a casino is for a patron to enter it with money and, statistically speaking, exit it with less. Physical casinos are often huge capital investments with large ongoing costs, including the return on that speculative capital. If they could choose to be less capital intensive, they would do so, but they are partially constrained by market forces and partially by regulation.

A crypto exchange is also capital intensive, not because the website or API took much investment (relatively low, by the standards of financial software) and not because they have a physical plant, but because trust is expensive. Bettors, and the more sophisticated market makers, who are the primary source of action for bettors, need to trust that the casino will actually be able to pay out winnings. That means the casino needs to keep assets (generally, mostly crypto, but including a smattering of cash for those casinos which are anomalously well-regarded by the financial industry) on hand exceeding customer account balances.

Those assets are… sitting there, doing nothing productive. And there is an implicit cost of capital associated with them, whether nominal (and borne by a gambler) or material (and borne by a sophisticated market making firm, crypto exchange, or the crypto exchange’s affiliate which trades against customers [0]).

Perpetual futures exist to provide the risk gamblers seek while decreasing the total capital requirement (shared by the exchange and market makers) to profitably run the enterprise.

Perps predate crypto but found a home there

In the commodities futures markets, you can contract to either buy or sell some standardized, valuable thing at a defined time in the future. The overwhelming majority of contracts do not result in taking delivery; they’re cancelled by an offsetting contract before that specified date.

Given that speculation and hedging are such core use cases for futures, the financial industry introduced a refinement: cash-settled futures. Now there is a reference price for the valuable thing, with a great deal of intellectual effort put into making that reference price robust and fair (not always successfully). Instead of someone notionally taking physical delivery of pork bellies or barrels of oil, people who are net short the future pay people who are net long the future on delivery day. (The mechanisms of this clearing are fascinating but outside today’s scope.)

Back in the early nineties economist Robert Shiller proposed a refinement to cash settled futures: if you don’t actually want pork bellies or oil barrels for consumption in April, and we accept that almost no futures participants actually do, why bother closing out the contracts in April? Why fragment the liquidity for contracts between April, May, June, etc? Just keep the market going perpetually.

This achieved its first widespread popular use in crypto (Bitmex is generally credited as being the popularizer), and hereafter we’ll describe the standard crypto implementation. There are, of course, variations available.

Multiple settlements a day

Instead of all of a particular futures vintage settling on the same day, perps settle multiple times a day for a particular market on a particular exchange. The mechanism for this is the funding rate. At a high level: winners get paid by losers every e.g. 4 hours and then the game continues, unless you’ve been blown out due to becoming overleveraged or for other reasons (discussed in a moment).

Consider a toy example: a retail user buys 0.1 Bitcoin via a perp. The price on their screen, which they understand to be for Bitcoin, might be $86,000 each, and so they might pay $8,600 cash. Should the price rise to $90,000 before the next settlement, they will get +/- $400 of winnings credited to their account, and their account will continue to reflect exposure to 0.1 units of Bitcoin via the perp. They might choose to sell their future at this point (or any other). They’ll have paid one commission (and a spread) to buy, one (of each) to sell, and perhaps they’ll leave the casino with their winnings, or perhaps they’ll play another game.

Where did the money come from? Someone else was symmetrically short exposure to Bitcoin via a perp. It is, with some very important caveats incoming, a closed system: since no good or service is being produced except the speculation, winning money means someone else lost.

One fun wrinkle for funding rates: some exchanges cap the amount the rate can be for a single settlement period. This is similar in intent to traditional markets’ usage of circuit breakers: designed to automatically blunt out-of-control feedback loops. It is dissimilar in that it cannot actually break circuits: changes to funding rate can delay realization of losses but can’t prevent them, since they don’t prevent the realization of symmetrical gains.

Perp funding rates also embed an interest rate component. This might get quoted as 3 bps a day, or 1 bps every eight hours, or similar. However, because of the impact of leverage, gamblers are paying more than you might expect: at 10X leverage that’s 30 bps a day. Consumer finance legislation standardizes borrowing costs as APR rather than basis points per day so that an unscrupulous lender can’t bury a 200% APR in the fine print.

Convergence in prices via the basis trade

Prices for perps do not, as a fact of nature, exactly match the underlying. That is a feature for some users.

In general, when the market is exuberant, the perp will trade above spot (the underlying market). To close the gap, a sophisticated market participant should do the basis trade: make offsetting trades in perps and spot (short the perp and buy spot, here, in equal size). Because the funding rate is set against a reference price for the underlying, longs will be paying shorts more (as a percentage of the perp’s current market price). For some of them, that’s fine: the price of gambling went up, oh well. For others, that’s a market incentive to close out the long position, which involves selling it, which will decrease the price at the margin (in the direction of spot).

The market maker can wait for price convergence; if it happens, they can close the trade at a profit, while having been paid to maintain the trade. If the perp continues to trade rich, they can just continue getting the increased funding cost. To the extent this is higher than their own cost of capital, this can be extremely lucrative.

Flip the polarities of these to understand the other direction.

The basis trade, classically executed, is delta neutral: one isn’t exposed to the underlying itself. You don’t need any belief in Bitcoin’s future adoption story, fundamentals, market sentiment, halvings, none of that. You’re getting paid to provide the gambling environment, including a really important feature: the perp price needs to stay reasonably close to the spot price, close enough to continue attracting people who want to gamble. You are also renting access to your capital for leverage.

You are also underwriting the exchange: if they blow up, your collateral becoming a claim against the bankruptcy estate is the happy scenario. (As one motivating example: Galois Capital, a crypto hedge fund doing basis trades, had ~40% of its assets on FTX when it went down. They then wound down the fund, selling the bankruptcy claim for 16 cents on the dollar.)

Recall that the market can’t function without a system of trust saying that someone is good for it if a bettor wins. Here, the market maker is good for it, via the collateral it kept on the exchange.

Many market makers function across many different crypto exchanges. This is one reason they’re so interested in capital efficiency: fully collateralizing all potential positions they could take across the universe of venues they trade on would be prohibitively capital intensive, and if they do not pre-deploy capital, they miss profitable trading opportunities. [1]

Leverage and liquidations

Gamblers like risk; it amps up the fun. Since one has many casinos to choose from in crypto, the ones which only “regular” exposure to Bitcoin (via spot or perps) would be offering a less-fun product for many users than the ones which offer leverage. How much leverage? More leverage is always the answer to that question, until predictable consequences start happening.

In a standard U.S. brokerage account, Regulation T has, for almost 100 years now, set maximum leverage limits (by setting minimums for margins). These are 2X at position opening time and 4X “maintenance” (before one closes out the position). Your brokerage would be obligated to forcibly close your position if volatility causes you to exceed those limits.

As a simplified example, if you have $50k of cash, you’d be allowed to buy $100k of stock. You now have $50k of equity and a $50k loan: 2x leverage. Should the value of that stock decline to about $67k, you still owe the $50k loan, and so only have $17k remaining equity. You’re now on the precipice of being 4X leveraged, and should expect a margin call very soon, if your broker hasn’t “blown you out of the trade” already.

What part of that is relevant to crypto? For the moment, just focus on that number: 4X.

Perps are offered at 1X (non-levered exposure). But they’re routinely offered at 20X, 50X, and 100X. SBF, during his press tour / regulatory blitz about being a responsible financial magnate fleecing the customers in an orderly fashion, voluntarily self-limited FTX to 20X.

One reason perps are structurally better for exchanges and market makers is that they simplify the business of blowing out leveraged traders. The exact mechanics depend on the exchange, the amount, etc, but generally speaking you can either force the customer to enter a closing trade or you can assign their position to someone willing to bear the risk in return for a discount.

Blowing out losing traders is lucrative for exchanges except when it catastrophically isn’t. It is a priced service in many places. The price is quoted to be low (“a nominal fee of 0.5%” is one way Binance describes it) but, since it is calculated from the amount at risk, it can be a large portion of the money lost. If the account’s negative balance is less than the liquidation fee, wonderful, thanks for playing and the exchange / “the insurance fund” keeps the rest, as a tip.

In the case where the amount an account is negative by is more than the fee, that “insurance fund” can choose to pay the winners on behalf of the liquidated user, at management’s discretion. Management will usually decide to do this, because a casino with a reputation for not paying winners will not long remain a casino.

But tail risk is a real thing. The capital efficiency has a price: there physically does not exist enough money in the system to pay all winners given sufficiently dramatic price moves. Forced liquidations happen. Sophisticated participants withdraw liquidity (for reasons we’ll soon discuss) or the exchange becomes overwhelmed technically / operationally. The forced liquidations eat through the diminished / unreplenished liquidity in the book, and the magnitude of the move increases.

Then crypto gets reminded about automatic deleveraging (ADL), a detail to perp contracts that few participants understand.

We have altered the terms of your unregulated futures investment contract.

(Pray we do not alter them further.)

Risk in perps has to be symmetric: if (accounting for leverage) there are 100,000 units of Somecoin exposure long, then there are 100,000 units of Somecoin exposure short. This does not imply that the shorts or longs are sufficiently capitalized to actually pay for all the exposure in all instances.

In cases where management deems paying winners from the insurance fund would be too costly and/or impossible, they automatically deleverage some winners. In theory, there is a published process for doing this, because it would be confidence-costing to ADL non-affiliated accounts but pay out affiliated accounts, one’s friends or particularly important counterparties, etc. In theory.

In theory, one likely ADLs accounts which were quite levered before ones which were less levered, and one ADLs accounts which had high profits before ones with lower profits. In theory. [2]

So perhaps you understood, prior to a 20% move, that you were 4X leveraged. You just earned 80%, right? Ah, except you were only 2X leveraged, so you earned 40%. Why were you retroactively only 2X? That’s what automatic deleveraging means. Why couldn’t you get the other 40% you feel entitled to? Because the collective group of losers doesn’t have enough to pay you your winnings and the insurance fund was insufficient or deemed insufficient by management.

ADL is particularly painful for sophisticated market participants doing e.g. a basis trade, because they thought e.g. they were 100 units short via perps and 100 units long somewhere else via spot. If it turns out they were actually 50 units short via perps, but 100 units long, their net exposure is +50 units, and they have very possibly just gotten absolutely shellacked.

In theory, this can happen to the upside or the downside. In practice in crypto, this seems to usually happen after sharp decreases in prices, not sharp increases. For example, October 2025 saw widespread ADLing as (more than) $19 billion of liquidations happened, across a variety of assets. Alameda’s CEO Caroline Ellison testified that they lost over $100 million during the collapse of Terra’s stablecoin in 2022, but since FTX’s insurance fund was made up; when leveraged traders lost money, their positions were frequently taken up by Alameda. That was quite lucrative much of the time, but catastrophically expensive during e.g. the Terra blowup. Alameda was a good loser and paid the winners, though: with other customers’ assets that they “borrowed.”

An aside about liquidations

In the traditional markets, if one’s brokerage deems one’s assets are unlikely to be able to cover the margin loan from the brokerage one has used, one’s brokerage will issue a margin call. Historically that gave one a relatively short period (typically, a few days) to post additional collateral, either by moving in cash, by transferring assets from another brokerage, or by experiencing appreciation in the value of one’s assets. Brokerages have the option, and in some cases the requirement, to manage risk after or during a margin call by forcing trades on behalf of the customer to close positions.

It sometimes surprises crypto natives that, in the case where one’s brokerage account goes negative and all assets are sold, with a negative remaining balance, the traditional markets largely still expect you to pay that balance. This contrasts with crypto, where the market expectation for many years was that the customer was Daffy Duck with a gmail address and a pseudonymous set of numbered accounts recorded on a blockchain, and dunning them was a waste of time. Crypto exchanges have mostly, in the intervening years, either stepped up their game regarding KYC or pretended to do so, but the market expectation is still that a defaulting user will basically never successfully recover. (Note that the legal obligation to pay is not coextensive with users actually paying. The retail speculators with $25,000 of capital that the pattern day trade rules are worried about will often not have $5,000 to cover a deficiency. On the other end of the scale, when a hedge fund blows up, the fund entity is wiped out, but its limited partners—pension funds, endowments, family offices—are not on the hook to the prime broker, and nobody expects the general partner to start selling their house to make up the difference.) 

So who bears the loss when the customer doesn’t, can’t, or won’t? The waterfall depends on market, product type, and geography, but as a sketch: brokerages bear the loss first, out of their own capital. They’re generally required to keep a reserve for this purpose. 

A brokerage will, in the ordinary course of business, have obligations to other parties which would be endangered if they were catastrophically mismanaged and could not successfully manage risk during a downturn. (It’s been known to happen, and even can be associated with assets rather than liabilities.) In this case, most of those counterparties are partially insulated by structures designed to insure the peer group. These include e.g. clearing pools, guaranty funds capitalized by the member firms of a clearinghouse, the clearinghouse’s own capital, and perhaps mutualized insurance pools. That is the rough ordering of the waterfall, which varies depending geography/product/market.

One can imagine a true catastrophe which burns through each of those layers of protection, and in that case, the clearinghouse might be forced to assess members or allocate losses across survivors. That would be a very, very bad day, but contracts exist to be followed on very bad days.

One commonality with crypto, though: this system is also not fully capitalized against all possible events at all times. Unlike crypto, which for contingent reasons pays some lip service to being averse to credit even as it embraces leveraged trading, the traditional industry relies extensively on underwriting risk of various participants.

Will crypto successfully “export” perps?

Many crypto advocates believe that they have something which the traditional finance industry desperately needs. Perps are crypto’s most popular and lucrative product, but they probably won’t be adopted materially in traditional markets.

Existing derivatives products already work reasonably well at solving the cost of capital issue. Liquidations are not the business model of traditional brokerages. And learning, on a day when markets are 20% down, that you might be hedged or you might be bankrupt, is not a prospect which fills traditional finance professionals with the warm fuzzies.

And now you understand the crypto markets a bit better.

[0] Brokers trading with their own customers can happen in the ordinary course of business, but has been progressively discouraged in traditional finance, as it enables frontrunning. 

Frontrunning, while it is understood in the popular parlance to mean “trading before someone else can trade” and often brought up in discussions of high frequency trading using very fast computers, does not historically mean that. It historically describes a single abusive practice: a broker could basically use the slowness of traditional financial IT systems to give conditional post-facto treatment to customer orders, taking the other side of them (if profitable) or not (if not). Frontrunning basically disappeared because customers now get order confirms almost instantly by computer not at end of day via a phone call. The confirm has the price the trade executed at on it. 

In classic frontrunning, you sent the customer’s order to the market (at some price X), waited a bit, and then observed a later price Y. If Y was worse for the customer than X, well, them’s the breaks on Wall Street. If Y was better, you congratulated the customer on their investing acumen, and informed them that they had successfully transacted at Z, a price of your choosing between X and Y. You then fraudulently inserted a recorded transaction between the customer and yourself earlier in the day, at price Z, and assigned the transaction which happened at X to your own account, not to the customer’s account.

Frontrunning was a lucrative scam while it lasted, because (effectively) the customer takes 100% of the risk of the trade but the broker gets any percentage they want of the first day’s profits. This is potentially so lucrative that smart money (and some investors in his funds!) thought Madoff was doing it, thus generating the better-than-market stable returns for over a decade through malfeasance. Of frontrunning Madoff was entirely innocent.

Some more principled crypto participants have attempted to discourage exchanges from trading with their own customers. They have mostly been unsuccessful: Merit Peak Limited is Binance’s captive entity which does this. It also is occasionally described by U.S. federal agencies as running a sideline in money laundering, Alameda Research was FTX’s affiliated trading fund. Their management was criminally convicted of money laundering. etc, etc.

One of the reasons this behavior is so adaptive is because the billions of dollars sloshing around can be described to banks as “proprietary trading” and “running an OTC desk”, and an inattentive bank (like, say, Silvergate, as recounted here) might miss the customer fund flows they would have been formally unwilling to facilitate. This is a useful feature for sophisticated crypto participants, and so some of them do not draw attention to the elephant in the room, even though it is averse to their interests.

[1] Not all crypto trades are pre-funded. Crypto OTC transactions sometimes settle on T+1, with the OTC desk essentially extending credit in the fashion that a prime broker would in traditional markets. But most transactions on exchanges have to be paid immediately in cash already at the venue. This is very different from traditional equity market structure, where venues don’t typically receive funds flow at all, and settling/clearing happens after the fact, generally by a day or two.

[2] I note, for the benefit of readers of footnote 0, that there is often a substantial gap between the time when market dislocation happens and when a trader is informed they were ADLed. The implications of this are left as an exercise to the reader.