> **Source:** https://permissionless.fi/en/02-perpetual-futures
> From *Permissionless Finance* (Permissionless Finance: From Perpetual Futures to the On-Chain Global Market) by Eric Cheung. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

# Chapter 2: Derivatives Markets and the Rise of Perpetual Futures

By CoinGecko's top-10 perpetual-exchange measure, crypto perpetual futures trading volume reached $92.9 trillion in 2025, of which centralized exchanges accounted for roughly $86.2 trillion and decentralized exchanges (DEXs) for roughly $6.7 trillion. Decentralized perpetual volume grew by roughly 3.5 times over the prior year's roughly $1.5 trillion, reaching approximately 4.5 times that level (about +347%) [1] [2]. A product that BitMEX did not formally launch until May 13, 2016 grew from nothing, in less than a decade, into one of the most important pricing and leverage infrastructures in the digital-asset market. Early BitMEX data show that derivatives markets lead BTC price discovery [3], and later research indicates that the regulated CME bitcoin futures market performs an important price-discovery function [4]. Together, this evidence shows that digital-asset price discovery has migrated substantially toward highly liquid derivatives markets, but it does not reduce to the claim that all spot prices mechanically track all perpetual futures (see Section 2.4.1).

This raises two questions: what problem does the design of perpetual futures solve, and what new cost does it introduce? Chapter 1 argued that the institutional costs of traditional finance can be unified as a trust tax, and that the six distinctive properties of digital assets compress this tax through multiplicative interaction. It closed by identifying perpetual futures as the composability product of these six properties in the domain of derivatives: the combination of 24/7 continuous trading, programmable rules, and atomic settlement creates a structural conflict between the fixed-expiry design of traditional futures and the continuous nature of digital assets. By abolishing the expiry date and introducing a funding rate mechanism, perpetual futures became the most natural form of derivative in this new environment. But being the *most natural* form does not make it *costless*. This chapter's central position is that perpetual futures are not an upgraded version of traditional futures but a costly design trade-off; only by understanding what they gain and what they lose can one understand this market.

This chapter retells the 400-year history of derivatives as three leaps in trust technology: personalized trust, institutionalized trust, and algorithmic trust. On that basis, it identifies three structural conflicts between traditional futures and digital assets—temporal mismatch, liquidity fragmentation, and trust mismatch—and shows how the two core design innovations of perpetual futures, together with their market outcome, address each conflict directly. It then analyzes the structural costs that accompany these innovations and uses a leverage-cycle model to reveal the market's endogenous fragility. These frameworks are prerequisites for the book's later analysis: the multiple economic natures of the funding rate, the endogenous logic of the liquidation cascade, and the migration of the trust paradigm in decentralized perpetual futures.

## 2.1 The trust evolution of derivatives

The 400-year evolution of derivatives can be retold as three leaps in trust technology. Forward contracts rely on bilateral, personalized trust; futures contracts achieve institutionalized trust through standardization and the central counterparty (CCP); and perpetual futures achieve algorithmic trust through predefined rules and economic incentives. This section examines the trust logic and limitations of forwards, futures, and options in turn, distills a unified functional framework, and shows how perpetual futures complete the third leap in derivatives design.

### 2.1.1 Personalized trust and forward contracts

The forward contract is the logical starting point for all financial derivatives. Its definition is straightforward: a customized over-the-counter (OTC) agreement that binds two parties to buy or sell an underlying asset on a specified future date at a price agreed today. The wheat-forward trades between farmers and flour mills in the nineteenth-century American Midwest illustrate the basic form of intertemporal risk management: the farmer locks in the autumn selling price, the mill locks in its raw-material cost, and each sheds its exposure to future price uncertainty. Hull (2022) [5] notes that this risk-transfer function is the economic foundation of all derivatives.

The core dilemma of the forward contract is counterparty credit risk. If the autumn market price of wheat is far above the agreed contract price, the farmer has a strong incentive to default; if it is far below, the same is true of the mill. Without a neutral and powerful guarantee of performance, the forward market's size is limited by its radius of trust: it can operate only within a circle of acquaintances bound by social ties [5]. Illiquidity compounds the problem. Every forward contract is customized and can barely be transferred to a third party; once a position is established, the only exits are performance and default. The forward market is essentially a network of countless isolated bilateral credit relationships rather than a single, liquid market.

Viewed through the trust-layer framework of Chapter 1, the fundamental problem of the forward contract is that it has no independent trust-function layer: no independent custody, no independent clearing, and no independent data verification. All trust functions are folded into a single bilateral relationship, so if one party's credit collapses, the entire contract collapses. The forward contract therefore pays a very high tax for trust, in the form of a constrained market size and counterparty risk; every potential transaction beyond the radius of trust goes unconsummated because the cost of trust is too high.

### 2.1.2 Institutionalized trust and the futures revolution

Futures contracts answered the trust bottleneck of the forward through two core institutional innovations. Standardization converted the customization of the forward into uniform specifications—contract size, grade of the underlying, delivery date, and delivery location were all fixed—creating fungibility: any contract for a given month is identical to every other contract for that month. This fungibility lays the foundation for free transfer and markedly improves liquidity. The founding of the Chicago Board of Trade in 1848 marked the institutional starting point of this model [5]. The CCP interposes itself in every trade through novation, becoming the seller to every buyer and the buyer to every seller; it consolidates countless dispersed bilateral credit relationships into concentrated trust in a single central institution, fundamentally resolving counterparty risk.

The CCP's risk-management architecture centers on a loss-sharing waterfall: strict membership admission ensures that only well-capitalized institutions can participate; the margin system requires clearing members to post initial and maintenance margin and to submit to daily mark-to-market; and the default waterfall draws in turn on the defaulter's margin, the CCP's own capital, and the guarantee fund contributed by non-defaulting members. This architecture converts the problem of *trusting a person* into the problem of *trusting an institution*. The CCP's credit rating far exceeds that of any individual trader, but the CCP itself becomes a systemic node. *Concentrating risk in order to manage risk* is the inherent contradiction of the CCP model—a point Chapter 1 developed in its analysis of the chain reaction triggered by Lehman Brothers' collapse.

Viewed through Chapter 1's five-layer trust-function lens, futures contracts realize, through specialized institutions, the trust functions that forward contracts lacked: the exchange provides fair execution, the CCP provides clearing, custodian banks hold margin, and regulators impose rule-based constraints. Institutionalizing trust extended the market's reach from a circle of acquaintances to strangers worldwide. This extension raises the trust tax, because each independent trust-function layer needs a specialized institution to run it, with its own operating costs and compliance requirements; yet the liquidity and risk-management efficiency it releases far exceed that incremental tax. The inherent contradiction of the CCP model—whether technological means can inherit the CCP's risk-management function while avoiding its centralization risk—sets up the third leap.

### 2.1.3 Options and nonlinear payoffs

Options moved beyond the linear payoff of futures, granting the holder a right rather than an obligation and creating an asymmetric risk structure. The buyer of a call profits when the underlying rises and loses only the premium when it falls, combining limited downside with unlimited upside. The buyer of a put, conversely, gains protection when the underlying falls, retaining upside while hedging downside risk.

Black and Scholes (1973) [6] solved the central problem of option pricing, showing that an option's price is determined by five factors: the price of the underlying, the strike price, the risk-free rate, the time to expiry, and volatility. Their core insight is that the higher the volatility, the greater the option's value. In the highly volatile digital-asset market, this relationship implies that options should, in theory, command a relatively high premium. The empirical outcome is the opposite: options have not become the dominant derivative in the digital-asset market, and crypto option trading volume in 2025 was far below that of perpetual futures [1]. This *thing that did not happen* demands explanation just as much as the pronounced dominance of perpetual futures.

From the standpoint of trust technology, the innovation of options lies in the payoff structure, not in the trust mechanism. Options rest on the same institutionalized-trust foundation as futures: exchange matching, CCP clearing, and margin management. The options clearing corporation—epitomized by the Options Clearing Corporation (OCC) in the United States—is structurally isomorphic to the futures CCP. Options therefore extend derivatives within the payoff space rather than mark a new leap in the trust paradigm. This distinction helps explain why options have failed to displace perpetual futures in the crypto market: options solve a payoff-structure problem, whereas perpetual futures solve a trust-paradigm problem, and the latter is the more fundamental challenge facing the digital-asset market.

Preliminary clues point in three directions. Although the nonlinear payoff of options is theoretically superior, for most crypto-market participants, who engage primarily in directional speculation, the linear payoff of futures is more intuitive and easier to grasp. Behavioral finance offers a deeper explanation: mental accounting leads traders to perceive an option premium as a *certain loss*, while leveraged margin is filed as a *temporary loss that may yet be recovered*, and this framing systematically favors futures over options. Market-making in options is also far more complex than in linear contracts, because market makers must manage five risk dimensions at once (delta, gamma, theta, vega, and rho); under the crypto market's limited liquidity, this complexity widens quoted spreads and thins depth. Most fundamentally, one core input to option pricing—the time to expiry—runs into structural difficulty in a product designed without an expiry date, a point analyzed in Section 2.2.4.

### 2.1.4 A unified functional framework

All derivatives, whether the structurally simple forward or the intricately designed option, serve four core economic functions. Risk transfer moves risk from those unwilling to bear it to those willing to bear it, allocating risk more efficiently across society. Price discovery, through traders' buying and selling, incorporates information and expectations into derivative prices, giving the whole economy forward-looking signals. The pioneering study by Garbade and Silber (1983) [7] provided a classic empirical framework for price discovery in futures markets, and its methodology has strongly shaped later research on price discovery in digital-asset markets. Capital efficiency, through the margin system, lets traders control positions of relatively large value with little capital. Liquidity provision, through standardization and centralized trading, attracts many traders and market makers, so participants can enter and exit at any time and at relatively low cost.

Through institutionalized trust, traditional derivatives have served global capital markets for more than a century, with exchanges such as CME, ICE, and Eurex processing tens of trillions of dollars in contract volume each year. But this system rests on a set of basic assumptions about trading hours, jurisdiction, and trusted intermediaries. When those assumptions collide with the native characteristics of digital assets, structural conflicts become unavoidable. Section 2.2 develops these conflicts systematically.

### 2.1.5 Algorithmic trust and perpetual futures

The path from theory to practice passed through three milestones. Shiller (1993) [8] proposed a perpetual, delivery-free futures contract that would adjust the deviation between the contract price and the underlying index through an overnight financing rate. In traditional finance, the idea sat unused, overlooked by academics for more than two decades for lack of an application. In 2011, developers implemented a preliminary inverse bitcoin futures contract on the ICBIT exchange, demonstrating technical feasibility; limited in scale and influence, it nonetheless carried the idea from theory to a working prototype. On May 13, 2016, BitMEX formally launched the XBTUSD perpetual futures contract [9], turning the concept into market reality. Two core design innovations—abolishing the expiry-and-delivery date and anchoring the price through a funding rate—together with their market outcome, the concentration of liquidity in a single contract, established BitMEX as the paradigm for digital-asset derivatives.

The trust mechanism of perpetual futures no longer rests on the personalized trust of a circle of acquaintances, and it also reduces reliance on the institutionalized trust of the traditional CCP; instead, it hands part of risk control to predefined algorithmic rules and economic incentives. Through internal transfer payments among traders, the funding rate achieves price anchoring through a market-based negative feedback loop. The two settings differ. On a centralized exchange (CEX), anchoring still relies on platform matching, the mark price, the margin system, the insurance fund, and auto-deleveraging (ADL) rules; only in the DEX setting does it approach an architectural trust jointly constituted by code, oracles, sequencing, and governance. How these two innovations and their market outcome address the structural conflicts between traditional futures and digital assets is argued in Sections 2.2 and 2.3.

## 2.2 The three structural conflicts

The institutional design of traditional futures embeds three assumptions: discrete trading hours, geographically defined jurisdiction, and centralized trusted intermediaries. The native characteristics of digital assets (continuous global trading 24/7, the absence of jurisdictional boundaries, and code as the basis of trust) conflict with these assumptions not as technical friction but as a fundamental incompatibility between two market paradigms. Transplanting traditional futures directly into the digital-asset world therefore entails structural friction, and the market needs a derivative that fits digital assets more closely. Perpetual futures are one of the principal responses to this need; they are not an incremental improvement on traditional futures but a fundamental redesign built for the native character of digital assets—round-the-clock, globalized flow. The argument proceeds along three dimensions: temporal mismatch, liquidity fragmentation, and trust mismatch.

### 2.2.1 Temporal mismatch

One of the core assumptions of traditional futures is a fixed, discrete expiry date: the contract has a defined life cycle and is delivered or rolled over before expiry. The digital-asset market runs around the clock and year-round, with no open and no close, no weekends, and no holidays, and information must be priced continuously. This conflict generates friction along three dimensions.

The first friction concerns the expiry date. Long-term holders must roll over repeatedly—closing the contract about to expire and opening a new, more distant one—and so bear double the transaction costs and basis risk. In the highly volatile crypto market, basis fluctuations during the roll window can wipe out weeks of trading profit. The second concerns trading hours. Even when traditional futures extend trading through electronic sessions, they remain constrained by weekends, holidays, daily settlement, and the contract life cycle, whereas the digital-asset market generates information continuously, day and night. When information accumulates during a market closure or a low-liquidity window but cannot be fully priced, the result can be price gaps and interruptions in risk management. The third concerns informational continuity. Traditional futures struggle to provide fully continuous price discovery for digital assets, which generate information around the clock.

The essence of this conflict is that the temporal structure of traditional futures was designed for traditional assets. The information flow and trading activity of traditional assets follow a natural daily and weekly cycle, and digital assets break that cycle. Forcing a continuous information flow into a discrete contract structure necessarily produces friction. Roll costs deserve careful economic accounting here. The concentration of trading volume in the front-month contract is a general phenomenon in futures markets—whether for crude oil or Treasury futures, the front-month contract typically accounts for the vast majority of volume—so the concentration of volume in CME bitcoin front-month futures [10] is not itself specific to digital assets. The more direct mechanism is this: perpetual futures remove the choice of maturity from the product structure, so traders need not choose among front-month, quarterly, and back-month contracts, nor roll over repeatedly to hold a continuous position. This is why a single contract that never has to be rolled naturally attracts more concentrated liquidity.

### 2.2.2 Liquidity fragmentation

Traditional futures meet demand across different maturities through multiple monthly contracts (front-month, back-month, quarterly), each month forming an independent liquidity pool. A digital asset is globally unified: BTC, unlike oil, has no delivery location, and, unlike Treasuries, no differing maturities; in essence it calls for a single, maximally deep liquidity pool. The multi-month design scatters across several pools the liquidity that ought to be concentrated in one, leaving each pool too shallow and raising the impact cost of large trades. In the digital-asset market, where liquidity is already thin relative to traditional markets, this fragmentation is especially costly.

Liquidity fragmentation is closely tied to price discovery. Precisely because multi-month futures cannot provide enough depth, price discovery for digital assets cannot proceed effectively within traditional futures. The market needs a product that gathers all liquidity in one place. By abolishing the expiry date, perpetual futures concentrate all trading depth in a single contract, achieving a high degree of liquidity aggregation. This is not a marginal improvement but a paradigm shift from *liquidity dispersed across multiple temporal dimensions* to *liquidity concentrated in a single contract with no temporal dimension*.

Yet while perpetual futures eliminate fragmentation in time, they do not eliminate it in space. Perpetual futures on the same underlying are spread across dozens of exchanges—Binance, OKX, Bybit, dYdX, and Hyperliquid each run independent order books—and their depth is not shared, so fragmentation shifts from the temporal to the spatial. Under normal conditions, cross-exchange arbitrageurs effectively close this spatial gap, hedging across platforms whenever a price gap appears and keeping prices consistent among exchanges. In extreme markets, however, arbitrageurs face capital constraints (margin is locked on multiple exchanges and cannot be moved between them) and execution constraints (an exchange may restrict new position-opening or suspend withdrawals), so spatial fragmentation reemerges precisely when unified liquidity is most needed. Section 2.3.1 develops the analysis of arbitrageurs' fragility in extreme environments.

### 2.2.3 Trust mismatch

The trust foundation of traditional futures is the CCP—a centralized, regulated, well-capitalized institution. Participants trust the CCP, and the CCP trusts its members, forming layer upon layer of nested institutional trust. The core value of the digital-asset community is verifiability: it seeks permissionless trust guaranteed by code. This philosophical incompatibility generates conflict at three levels.

The CCP model requires high entry barriers: membership, capital, and compliance review are prerequisites for participation, at odds with the permissionless access of digital assets. The CCP's operations are opaque to outside participants—its core processes, such as risk-management models, margin calculations, and shortfall-handling procedures, are not public—conflicting with the radical transparency of on-chain systems. And the CCP is a single point of failure whose collapse can trigger a systemic crisis, contrary to the distributed resilience that digital assets pursue. This conflict is not a technical obstacle but a fundamental tension between two trust paradigms; Chapter 1's analysis of the three trust paradigms maps here onto the derivatives market. Perpetual futures provide, through the funding rate mechanism, an algorithmic-trust solution that does not depend on a CCP; but the solution is imperfect, and its costs are developed in Sections 2.3 and 2.4.

### 2.2.4 Ruling out the alternatives

If abolishing the expiry date and adapting to a round-the-clock market are the core requirements, at least two other products could serve similar functions: the perpetual option and the contract for difference (CFD). Understanding why perpetual futures won out helps pinpoint their distinctive value.

The perpetual option faces a structural market obstacle. One core input to option pricing is the time to expiry, and time-value decay is a fundamental feature of options. Abolishing the expiry date sends the time to expiry to infinity, and the standard Black-Scholes framework for European options no longer applies. In theory, a perpetual American option does have a closed-form solution—Merton (1973) [11] derived the analytical pricing formula for the perpetual American put, and Hull (2022) [5] sets it out systematically—and academics and practitioners have proposed innovations such as everlasting options. The two must be kept distinct. Merton's perpetual American option is a mathematical object whose pricing depends on solving for the optimal exercise boundary and assumes the holder can exercise at any moment; everlasting options, by contrast, are a product design that rolls a series of short-dated options into long-term exposure through a funding-rate mechanism similar to that of perpetual futures, so the holder makes no exercise decision but pays a continuous financing cost. The core obstacle to the perpetual option is therefore not theoretical feasibility but the practical constraints of liquidity and complexity. As Section 2.1.3 showed, the nonlinear payoff of options and the five-dimensional complexity of market-making make it hard to build competitive market-making depth under crypto-market liquidity conditions. Options thus have distinctive value in the crypto market, particularly for institutional tail-risk hedging, but for these structural reasons they cannot become the dominant derivative.

The CFD has existed in traditional finance for decades. It, too, is a synthetic derivative (no expiry date, high leverage, no need to hold the underlying) and looks very much like a perpetual future. Yet the two differ fundamentally in structure. Under the typical B-book model, the counterparty to a CFD is the broker or platform: the user bets against the platform, the platform profits from the user's losses, and a structural conflict of interest exists. Some CFD brokers adopt the A-book model, hedging client order flow to external liquidity providers; the broker then does not profit directly from client losses, but the A-book model is not the norm in the retail CFD market. Perpetual futures, by contrast, match a trader's order against other traders' orders through an order book, so the market discovers the price, the platform earns fees, and users' gains and losses are not directly tied to the platform. A CFD's price-anchoring mechanism varies by model: under the B-book model, anchoring depends on the broker's reputation and regulatory constraints and is essentially an unverifiable promise; under the A-book model, the price tracks the underlying market's liquidity but has no explicit algorithmic anchor. Either way, the anchor is unverifiable: the user cannot independently audit whether the price is fair. The funding rate of a perpetual future supplies a mathematically verifiable algorithmic anchor instead. The distinctive position of perpetual futures is thus clear: they combine the convenience of a CFD (no expiry date, synthetic exposure, high leverage) with the market-based price discovery of futures (order-book matching eliminates the platform-user conflict of interest), and, through the algorithmic anchor of the funding rate, they strike an effective balance among degree of innovation, simplicity, and market-makability. In 2021, the United Kingdom's Financial Conduct Authority (FCA) banned the sale of crypto derivatives to retail consumers, and the European Securities and Markets Authority (ESMA) has held that crypto assets should be classified by reference to the boundary of financial instruments [12] [13]. This suggests that, as a matter of legal classification, perpetual futures may be brought within the CFD or other derivatives frameworks and subjected to the associated leverage limits, negative-balance protection, and retail-investor protection rules—a risk discussed further in Chapter 26 (for regulatory evolution, see Section 2.4.5).

In sum, the two core design innovations of perpetual futures map onto the three structural conflicts above: abolishing the expiry date addresses temporal mismatch, the funding rate addresses trust mismatch, and the concentration of liquidity in a single contract—the market outcome of these two innovations—addresses liquidity fragmentation. The innovations and the outcome are mutually dependent: abolishing the expiry date lets liquidity concentrate, but it also removes the hard anchor of expiry convergence, which is why the funding rate is needed as a substitute soft anchor. Section 2.3 turns to the internal logic of these mechanism designs.

Figure 2-1 places the four forms—forward, futures, option, and perpetual—on a timeline, corresponding to the three leaps of personalized, institutionalized, and algorithmic trust.

![Figure 2-1](./images/fig-2-1-en.png)

**Figure 2-1.** Timeline of derivatives evolution—the three leaps in trust mechanisms (Data source: Hull 2022 [5], Shiller 1993 [8], Hayes 2016 [9]; the roughly $92.9 trillion figure for 2025 annual perpetual futures volume shown in the figure follows [1][2])

The difference in payoff structure between futures and options is a key premise for understanding why perpetual futures rose to dominance. Figure 2-2 juxtaposes the two payoff curves: the futures payoff is linear and symmetric, while the option payoff is asymmetric (the downside is capped at the premium and the upside is theoretically unbounded). It is precisely this contrast between linear and nonlinear that drives the divergence in market-making complexity, cognitive barriers, and use cases described in Sections 2.1.3 and 2.2.4.

![Figure 2-2](./images/fig-2-2-en.png)

**Figure 2-2.** Comparison of the payoff structures of futures and options—linear versus nonlinear payoffs (Data source: Hull 2022 [5], Black & Scholes 1973 [6])

## 2.3 The core mechanisms of perpetual futures

This section examines the internal logic behind the design of perpetual futures. Their operation relies on two core mechanisms: the funding rate, which achieves price anchoring, and the mark price and liquidation mechanism, which achieve risk control. It also turns on two key design dimensions: the convexity difference between inverse and linear contracts, and the structural costs of abolishing the expiry date.

### 2.3.1 The funding rate

The funding rate is the core design innovation of perpetual futures. It is a fee settled directly between holders of long and short positions at regular intervals—commonly every 8 hours on mainstream exchanges, and every hour or continuously on some—whose direction and magnitude depend on the deviation between the perpetual price and the spot index price. The funding rate is not a fee charged by the exchange but an internal transfer payment among traders. Its common calculation framework is $\text{rate}_{8h} = \text{interest component}_{8h} + \text{premium component}_{8h}$; that is, the two components are summed over a single funding interval (such as 8 hours). Some exchanges set the interest component as a fixed term close to 0.01% per 8 hours, while others use different interest rates, caps and floors, and dynamic-adjustment mechanisms; the premium component is typically computed from a time-weighted average of the perpetual price's deviation from the spot index and is the main source of variation in the rate. Exchanges differ significantly in sampling frequency, weight-decay method, and the setting of caps and floors, and Chapter 10 develops a detailed comparison of the formulas.

Its core logic is a negative feedback loop: when the perpetual price is above spot, the funding rate is positive, longs pay shorts, the cost of being long rises, and longs are encouraged to close while new shorts enter, which pushes the price back down; the reverse holds when the perpetual price is below spot. Spot-futures arbitrageurs are the key force behind the anchoring mechanism: when the rate is positive, they short the perpetual while buying spot to earn the rate, placing selling pressure on the perpetual and buying pressure on spot and so accelerating the close of the gap. In extreme markets, however, arbitrageurs face multiple constraints, so the anchoring mechanism is most likely to fail precisely when it is most needed. The margin on a short perpetual position is continuously eroded as the underlying price rises, and in extreme cases the arbitrageur is exposed to liquidation; cross-exchange arbitrage requires capital to be locked separately on the spot and derivatives exchanges, doubling the total capital committed with none of it usable across venues; and an exchange may restrict new position-opening or suspend withdrawals, directly blocking the arbitrage. These constraints mean that arbitrageurs—the force that holds the anchor—may be squeezed out or retreat voluntarily in extreme environments, which forms the microfoundation of the soft anchor's fragility and echoes the analysis in Section 2.3.4.

Ackerer, Hugonnier, and Jermann (2025) [14] derived an explicit pricing formula for perpetual futures under a framework of no frictions, continuous trading, and no arbitrage, proving mathematically that the funding rate mechanism can guarantee consistency between the futures price and the spot price and giving this negative feedback loop a rigorous theoretical basis. The model's core assumptions—especially frictionlessness and continuous trading—do not necessarily hold under real crypto-market conditions, which is itself a starting point for understanding how the rate deviates in extreme markets.

Figure 2-3 presents the internal logic of this negative feedback loop: when the perpetual price deviates from spot, the funding rate adjusts automatically with the deviation, guiding the price back by changing the cost of holding long and short positions. The figure also shows the two paths of premium and discount and the role of spot-futures arbitrageurs in accelerating convergence.

![Figure 2-3](./images/fig-2-3-en.png)

**Figure 2-3.** The negative feedback loop of the funding rate price-anchoring mechanism—the funding rate is positive when the perpetual price is above spot (longs pay) and negative when it is below spot (shorts pay), so a deviation automatically produces an opposing economic incentive (Data source: Ackerer, Hugonnier & Jermann 2025 [14])

Under extreme market conditions, the funding rate behaves in ways far more complex than this description suggests: it can spike to thousands of percent on an annualized basis, becoming a punitive tax on holding a position and triggering large-scale position adjustments and further volatility. Chapter 10 develops the analysis of the rate's game-theoretic structure and extreme behavior.

The economic nature of the funding rate extends well beyond that of an anchoring tool. Its first nature is that of the continuous form of the traditional cost of carry. Traditional futures pricing follows the cost-of-carry model, in which the futures price equals the spot price plus the cost of carry; the cost of carry is embedded in the basis, and the more distant the expiry, the larger the basis. Perpetual futures abolish the expiry date and eliminate the explicit term basis, but they cannot eliminate the cost of carry: it shifts from being embedded in the price to being paid explicitly and continuously through the funding rate. A positive funding rate equals the cost of being long and can also be read as the implied interest rate on borrowed risk exposure. The Bank for International Settlements (BIS) working paper by Schmeling, Schrimpf, and Todorov shows that crypto carry can reach very high levels in some periods and that its returns relate to trend-chasing demand, limited arbitrage capital, and crowded-trade risk [15]. This is not a free lunch but the market's equilibrium pricing of leverage demand, funding constraints, and risk-bearing. The funding rate is in essence a real-time, explicit, and fragmented version of the basis.

Its second nature is that of a real-time market-sentiment indicator. Traditional sentiment indicators—the volatility index (VIX), based on option-implied volatility, and the put-call ratio, based on option volume—are complex to compute, limited in update frequency, and dependent on the liquidity of the options market. The funding rate provides a more direct, higher-frequency signal: a rate that stays positive and rises indicates that leveraged longs are accumulating and systemic fragility is rising. A positive rate, however, need not signal pure speculative long accumulation, because a large volume of delta-neutral basis trades also pushes the rate positive, and these arbitrage positions do not feed a liquidation cascade, so systemic fragility is far lower than when speculative longs dominate. The rate must therefore be read together with the increment (rather than the stock) of open interest, the distribution of liquidation thresholds, and spot-market capital flows; on its own it is a necessary but insufficient sentiment indicator. A rate that turns negative abruptly signals spreading panic, with longs forced to close or flip short; a rate above 100% annualized signals an extreme market state, in which the cost of holding a position itself begins to force traders to adjust. The rate is thus not only an anchoring tool but a piece of financial-information infrastructure in its own right, supplying real-time signals for volatility forecasting and market-state assessment that do not exist in traditional finance. Chapter 23 uses the rate to construct an early-warning indicator for the *calm trap*, and Chapter 24 incorporates it into a volatility-forecasting framework.

Its third nature is that of a foundation for an on-chain-native carry asset. The funding rate creates an on-chain-native carry trade: shorting the perpetual while buying spot to earn the rate; its risks are a rate reversal and liquidation. Projects such as Ethena have turned this strategy into a protocol; USDe delta-hedges spot against a corresponding short perpetual or dated futures position, packaging the strategy into a composable on-chain financial product [16]. But when a large amount of capital concentrates in the same carry strategy, that crowding fundamentally changes the return profile: it compresses carry, and when the rate reverses, the simultaneous unwinding of large positions becomes a *carry unwind* that is itself an additional source of market impact. The carry strategy is therefore procyclical: it attracts capital and suppresses volatility in calm periods, and it is forced to deleverage and amplifies volatility in turbulent ones. Chapter 1's composability thesis is confirmed here at the level of derivatives: round-the-clock operation lets the rate settle continuously, atomic settlement makes rate payments certain, and programmability lets the strategy be packaged into a smart contract. This innovation is possible only where perpetual futures exist, marking their upgrade from a trading tool into infrastructure for a new asset class.

### 2.3.2 Mark price and the liquidation mechanism

The mark price is a fair-price benchmark used to calculate unrealized profit and loss (unrealized P&L) and margin levels. It synthesizes the price indices of several mainstream spot exchanges (for example, the median or weighted average of Coinbase, Binance, and Kraken) rather than relying on the traded price of a single exchange. It is designed to provide a manipulation-resistant liquidation benchmark, keeping an instantaneous price anomaly on a single exchange—a sharp, short-lived price deviation that quickly reverts, colloquially called a *wick*—from triggering unjust, large-scale liquidations. In the early history of the crypto market, mass liquidations caused by this kind of price manipulation were common. The mark price is typically smoothed with an exponential moving average (EMA) or a time-weighted average price (TWAP) to filter short-term noise. Even so, its price sources can still be manipulated. The 2019 BitMEX event, in which a flash crash on the single exchange Bitstamp caused mass liquidations, is a typical case: a large sell order propagated through a heavily weighted price source and indirectly triggered more than $200 million in liquidations. The governance mechanism for selecting price sources and adjusting weights is itself a point of operational risk; its decisions are typically made unilaterally by the exchange and lack transparency. Chapter 11 analyzes in depth how the smoothing design creates an *accumulation–release* effect in fast markets.

When the margin ratio (calculated on the mark price) falls below the maintenance margin ratio, the forced-liquidation process is triggered. The simplified formula for the margin ratio is:

$$\text{margin ratio} = \frac{\text{margin balance} + \text{unrealized P\&L}}{\text{position notional value}}, \qquad \text{liquidation} \iff \text{margin ratio} < \text{maintenance margin ratio}$$

where both unrealized P&L and notional value are calculated on the mark price. This is a conceptually simplified formula; for the rounding direction, precision alignment, and the margin-aggregation basis for isolated versus cross margin in a fixed-point implementation, see Chapter 11. Under isolated margin, only the margin balance and profit and loss of a single position count; under cross margin, the margin levels of all positions in the account are combined, so a loss on one position can be partly offset by profits on others, at the cost of cross-asset contagion risk. This section focuses on how liquidation-mechanism design differs across exchanges (for the basic framework of the liquidation-execution process, see Section 1.4.3), because those differences directly affect the intensity of the liquidation cascade.

Exchanges' choices in liquidation-mechanism design form a continuous spectrum from aggressive to conservative. Table 2-1 sets out three representative models in the current market—cliff-edge liquidation, progressive liquidation, and vault takeover—comparing them along four dimensions (mechanism logic, market impact, advantages, and disadvantages) and revealing the different trade-offs each makes between the certainty of risk removal and minimal market impact.

| Model | Mechanism | Market impact | Advantages | Disadvantages | Representative |
| :--- | :--- | :--- | :--- | :--- | :--- |
| Cliff-edge liquidation | 100% of the position is liquidated instantly once margin hits the threshold | Greatest | Simple and certain; clears risk thoroughly | The entire liquidation volume is dumped at once, accelerating the cascade | Early CEX designs |
| Progressive liquidation | The position is reduced in batches as margin approaches the warning line | Moderate | Slows the cascade and gives traders time to react | Complex to implement and may delay risk removal | Bybit, OKX |
| Vault takeover | Liquidated positions are absorbed by the protocol's market-making vault | Smallest | Greatly slows the cascade and protects market structure | The vault bears directional risk and may itself become a risk point | Hyperliquid HLP |

**Table 2-1.** Three design models for the liquidation mechanism of perpetual futures (Data source: compiled by the author)

Each design strikes this trade-off differently. Cliff-edge liquidation pursues certainty, but at the cost of dumping the entire liquidation volume into the order book at once, accelerating the liquidation cascade. Progressive liquidation seeks a balance, reducing positions in batches to spread the impact, but it adds implementation complexity and may, in extreme conditions, delay risk removal if the batch pace is too slow. Vault takeover is the newest direction, represented by Hyperliquid's HLP market-making vault: liquidated positions are absorbed within the vault rather than hitting the public order book, so the impact on the market price is minimal, but the vault itself bears concentrated directional risk. When a major directional move persists, the vault may accumulate a large opposing position and become a systemic risk point. Hyperliquid's JELLY incident of March 2025 offers a case study of this risk: public post-mortems show that the abnormal price and position pressure of the low-liquidity token JELLY forced the platform to protect the HLP vault through discretionary intervention at the validator level [17]. This does not mean that every vault-takeover model will fail, but it exposes a key tension: whether, in a crisis, a protocol retains enough governance and validator discretion to overturn market outcomes. Chapter 11 systematically compares these designs, and Chapter 29 designs an optimal solution from an engineering perspective.

### 2.3.3 Inverse contracts and linear contracts

BitMEX's XBTUSD perpetual futures contract is the classic inverse-contract design: it uses BTC as both margin and settlement currency and is quoted in USD. This design has an important economic feature: it superimposes coin-denominated settlement on same-direction movement of the collateral price. For an inverse contract with a fixed U.S.-dollar notional principal, the BTC-denominated profit and loss can typically be written as a reciprocal price difference of the form $\text{PnL}_{\text{coin}} \propto Q\left(\dfrac{1}{P_{\text{entry}}} - \dfrac{1}{P_{\text{exit}}}\right)$ (where $Q$ is the contract quantity, or notional principal); nonlinearity therefore first appears in the coin-denominated settlement outcome rather than in a simple dollar-P&L formula. The trader in fact bears a dual exposure: the position's direction is exposed to the BTC price, and so is the margin asset itself. When the price falls, the position's loss widens while the dollar value of the margin shrinks, the margin buffer thins nonlinearly, and liquidation triggers faster. An inverse contract thus fuses the directional bet and the collateral exposure into one, creating a nonlinear risk structure rarely seen in traditional dollar-margined linear contracts.

The convexity effect of the inverse contract shows up on the payoff curve. Figure 2-4 juxtaposes the long payoff curves of inverse and linear contracts: the linear contract, quoted in a stablecoin, is approximately linear and proportional, whereas the inverse contract superimposes the coin-denominated payoff on the value of the BTC collateral, so the margin buffer is consumed faster in a falling market—the geometric reason it triggers liquidation faster.

![Figure 2-4](./images/fig-2-4-en.png)

**Figure 2-4.** Comparison of the payoff curves of inverse and linear contracts—the convexity effect (Data source: derived by the author)

As stablecoins such as USDT and USDC became widespread, the market shifted toward linear contracts margined in stablecoins. A linear contract's profit and loss is linear, more intuitive and easier to compute for risk, but it loses the convexity of the inverse contract. In a rising market, a linear contract's leverage does not amplify automatically as an inverse contract's does. Linear contracts now dominate, because the spread of stablecoins has made dollar-denominated linear payoffs the default mental model for users. Inverse contracts retain their market, particularly where stablecoins are not used and in strategies that require the convexity effect. The difference between inverse and linear contracts bears directly on the later analysis of liquidation mechanisms and market anomalies: Chapter 11 formalizes the nonlinear liquidation complexity of the inverse contract, and Chapter 18 analyzes the pricing differences between the two.

### 2.3.4 The cost of abolishing the expiry date

Abolishing the expiry date is the fundamental design decision that sets perpetual futures apart from traditional futures, but it is not free. In traditional futures, the expiry date performs three often-overlooked economic functions, and their disappearance imposes a threefold cost that perpetual futures must confront.

The first cost is the disappearance of a mechanism for the forced disclosure of information. The expiry-and-delivery of traditional futures guarantees that the contract price must converge to the spot price on the expiry date—a hard anchor backed by physical delivery or cash settlement. All market participants know this endpoint exists, so on every day before expiry, price discovery revolves around where the price will ultimately converge. Perpetual futures abolish this endpoint; the funding rate is a soft anchor that guides the price back through economic incentives but does not guarantee the return. The strength of the anchor depends on the activity and capital adequacy of arbitrageurs. Under normal conditions, arbitrageurs can usually compress the deviation between the perpetual and the spot index fairly quickly; but in extreme markets, capital constraints, exchange risk, and execution constraints weaken their capacity to correct the deviation, so it persists longer. As a result, the deviation between the perpetual price and the spot price can, in theory, be more persistent and more severe than in futures with an expiry date. One root of the persistent-premium or persistent-discount anomaly is precisely this structural weakening of the anchor once a hard anchor gives way to a soft one, an anomaly analyzed in depth in Chapter 18.

The second cost is the possibility of continuous leverage accumulation. Traditional futures have expiry, rolling, daily mark-to-market, and margin adjustment—mechanisms that together impose periodic repricing and risk checks; a position can be rolled over, so systemic exposure is not automatically reset to zero on an expiry date, but the roll window forces traders to reconfirm their funding cost, the basis, and margin constraints. Perpetual futures have no such window. Leverage can accumulate more continuously during calm periods until, at some trigger point, it sets off a liquidation cascade that forces destructive deleveraging. The leverage-cycle dynamics of Section 2.4.3 formalize this process.

The third cost is the loss of a termination date for market-making risk. In futures with an expiry date, market makers know that their inventory risk has a definite end date and can price and manage it precisely on that basis; in the worst case, everything is settled on the expiry date. In perpetual futures, a market maker's risk exposure is in theory indefinite and terminates only when the position is actively closed or liquidated. This changes their behavior: in a long-lasting trend especially, they may lean toward reducing quote depth or widening spreads, because they cannot be sure when the trend will end. Chapter 19 formalizes this effect in the market maker's profit equation. This cost echoes Chapter 1's analysis of round-the-clock trading at the level of derivatives: the expiry date is another risk-mitigation mechanism for market makers, and perpetual futures abolish it.

Perpetual futures replace hard anchoring, periodic resets, and finite-horizon risk with soft anchoring, the possibility of unlimited accumulation, and indefinite-horizon risk. They are not an upgraded version of traditional futures but a different design trade-off, accepting a new set of structural costs in exchange for convenience. Understanding these costs is the prerequisite for analyzing the many distinctive phenomena in the perpetual futures market.

## 2.4 Market landscape and ecosystem evolution

Mechanism design determines what perpetual futures can do; the market itself reveals the impact they have had. By CoinGecko's top-10 perpetual-exchange measure, perpetual futures grew in less than a decade from nothing into a market with annual trading volume exceeding $92 trillion [1]. Their influence reaches far beyond a trading tool: they have reshaped the price-discovery hierarchy of digital assets, popularized high-leverage trading, given rise to an endogenous Minsky-style fragility, driven the rise of decentralized alternatives, and prompted a reconstruction of the global regulatory framework.

### 2.4.1 Market size and price discovery

In 2025, in the CoinGecko top-10 sample, the annual perpetual futures volume of centralized exchanges reached $86.2 trillion, up 47.4% year over year, while the annual volume of decentralized perpetual futures reached $6.7 trillion, up 346% year over year [1] [2]. By combining high liquidity, low transaction costs, and a broad base of participants, perpetual futures naturally became an important venue for expressing market information.

The evidence on price discovery must be read separately by market type. Alexander et al. (2020) [3] found that early BitMEX derivatives contributed a significant information share to BTC price discovery (this study uses data from 2016 to 2018, after which market structure has changed significantly: competition among exchanges, the degree of institutional participation, and the distribution of market liquidity are all now different). The cross-market analysis of Robertson and Zhang (2025) [4], by contrast, indicates that the regulated CME bitcoin futures market played a leading role in price formation within their sample. Together, the two support a more careful judgment: highly liquid derivatives markets tend to become important venues for expressing digital-asset information, but the conclusion for CME dated futures cannot be extended directly to all offshore or on-chain perpetual futures. This pattern, in which futures guide spot, has its root in the fragmentation of the digital-asset spot market (hundreds of exchanges worldwide, with no central market), which leaves leading derivatives exchanges, with their deeper liquidity pools, as the de facto centers of price coordination. When an important piece of information appears, the first to react may be the derivatives market, with its higher leverage efficiency and more concentrated liquidity; the reaction is then transmitted to the various spot exchanges through arbitrage.

This constitutes a conceptual inversion: the derivatization of the spot market. In traditional finance, the spot market is usually primary and derivatives are built on top of it. In the digital-asset market, this hierarchy weakens, at least in some periods and for some trading pairs: derivatives markets such as perpetual futures and dated futures may become the faster layer of information expression, with spot prices then absorbing that information through arbitrage and market-making. This structural change means that understanding the dynamics of the derivatives market is just as important to understanding spot prices. A deeper systemic risk follows: when price discovery is shaped by leveraged market depth, the price signal may embed the systemic noise of the leverage cycle. This has profound effects on DeFi protocols that rely on spot prices (for collateral valuation and oracle price feeds, for example), because they typically assume the spot price reflects fundamental information rather than the accumulation or liquidation of leveraged positions. Just how strongly perpetual futures lead price discovery, and whether that leadership is stable across market states, calls for microstructure tools such as the Hasbrouck information share, which Chapters 13 through 15 analyze systematically.

### 2.4.2 The democratization of leverage

In traditional finance, high-leverage trading is usually subject to strict account, margin, and investor-suitability constraints. In the U.S. stock market, Regulation T limits initial margin to 50%, equivalent to a maximum of 2x leverage [18]. Perpetual futures sharply lowered the barrier to leverage, turning nominal 10x-to-125x leveraged trading into something anyone can access simply by downloading an app or connecting a wallet. Nominal leverage must be distinguished from effective leverage: 125x is usually available only for very small positions, the tiered-margin system makes the effective maximum for large positions far lower, and actual leverage is further constrained by platform risk controls, liquidity, and liquidation thresholds. This is a democratization of leverage that never appeared in traditional finance. It echoes the broader debate over the democratization of finance in fintech: technology has lowered the barrier to entry for financial services, but whether wider access comes with adequate risk education and investor protection remains a central, ongoing debate among academics and regulators. This book uses *democratization of leverage* as a descriptive term, with no normative connotation.

Understanding it starts with the behavioral drivers on the demand side. The forces that lead retail traders to choose 50x-to-125x leverage include overconfidence bias, gambling utility, and a returns-display bias on social media: profit screenshots are widely shared while losses are hidden, producing a socialized amplification of survivorship bias. The default setting of the leverage slider and one-tap execution also reduce perceived friction, making high leverage cognitively too easy to choose.

The democratization of leverage cuts both ways. On the positive side, it releases capital efficiency: traders with little capital can gain exposure to large positions on limited margin, hedgers can lock in risk with less capital, and overall capital efficiency improves markedly. On the negative side, it amplifies risk systemically: a large volume of dispersed leverage risk accumulates at the system level, and every high-leverage retail account is a potential liquidation trigger. When the market makes a directional move, these dispersed liquidations fire at once, forming a liquidation cascade. Democratizing leverage thus concentrates risk. The concentration unfolds along three dimensions: the dense distribution of liquidation thresholds across the price space produces the cascade effect; tail losses concentrate on the insurance fund and ADL recipients; and systemic risk concentrates on a few leading exchange platforms. At the individual level, 50x-to-100x leverage means that a mere 1%-to-2% adverse move triggers liquidation, leaving the trader's risk-tolerance window extremely narrow. At the system level, when millions of such narrow windows are densely packed along the price axis, any directional shock sweeps across many liquidation thresholds, forming a wave of liquidations. The selling pressure from those liquidations pushes the price further, sweeping across still more thresholds, and a positive feedback loop forms—the microstructure mechanism of the liquidation cascade.

### 2.4.3 Leverage-cycle dynamics

The liquidation cascade is not an occasional event but an endogenous, cyclical dynamic of the perpetual futures market. This cycle maps closely onto Minsky's (1986) [19] financial instability hypothesis, but it runs far faster and lacks the institutional tools to manage the cycle.

Figure 2-5 formalizes this leverage cycle as a four-stage closed loop (accumulation, criticality, deleveraging, and recovery), in which the end of the recovery phase is the start of the next accumulation phase, so the cycle repeats. The figure marks the typical direction of change in each phase's core observables—volatility, open interest, the direction of the funding rate, and effective liquidity depth—providing a signal framework for later empirical identification.

![Figure 2-5](./images/fig-2-5-en.png)

**Figure 2-5.** A four-stage model of the leverage cycle in the perpetual futures market—accumulation, criticality, deleveraging, and recovery (conceptual schematic; the leverage level is on a relative scale and the time axis is illustrative; theoretical source: Minsky 1986 [19])

In the accumulation phase, fragility builds beneath surface calm. Volatility is low, traders gradually add leverage, the rate is mildly positive, the carry trade attracts more capital, open interest grows steadily, and the market appears to run smoothly. The process shows a clear reflexivity loop: low volatility → added leverage → carry attracts capital → volatility suppressed further. Danielsson, Shin, and Zigrand's (2012) [20] analysis of endogenous risk reveals the core paradox of this loop: the risk indicator is itself endogenously suppressed by trading behavior, so risk assessments based on historical volatility systematically underestimate true risk. Effective liquidity—order-book depth divided by the size of positions that can be liquidated—declines steadily. This is precisely the stability-breeds-instability that Minsky described: hedge financing gradually degrades into speculative financing and then into Ponzi financing, and it unfolds faster in the perpetual futures market. At the critical point, the system is extremely fragile. The ratio of open interest to market capitalization climbs to a high level, the rate exceeds 100% annualized, the cost of holding rises but long holders do not cut positions, and the market's tipping point falls ever lower. Here a medium-sized shock can, given the system's fragility, produce consequences far exceeding its own scale, and the system's nonlinear amplification peaks.

In the deleveraging phase, the liquidation cascade unfolds. The market event of October 10–11, 2025 provides a typical empirical case [21] [22]. An external macro shock—the spark, though months of prior leverage accumulation were the dry tinder—set off the first wave of liquidations; the selling pressure from those liquidations pushed the price down further and triggered still more, forming a positive feedback loop. Public data indicate that reported liquidations exceeded $19 billion within about one day [21]; Ali (2025), in an SSRN working paper, uses a measure of open interest erased to describe a contraction of open interest of roughly $19 billion within 36 hours [23]. The two figures are close but mean different things: the former measures reported forced liquidations, the latter the reduction in open interest, and they cannot be used interchangeably. Exchange liquidation data, moreover, generally underestimate the true scale.

Figure 2-6 uses market data from October 10–11, 2025 to depict this positive feedback loop: the phases in which the BTC price accelerated downward coincide closely on the time axis with pulse-like surges in hourly liquidation volume, confirming the positive-feedback amplification between forced-liquidation selling pressure and price decline.

![Figure 2-6](./images/fig-2-6-en.png)

**Figure 2-6.** During the October 2025 liquidation-cascade event, the BTC price trajectory (top) and hourly and cumulative liquidation volume (bottom); the phases of accelerating price decline coincide closely in time with the peaks in liquidation volume, revealing forced liquidation as a positive-feedback amplification mechanism of the market crash (the data form is an illustrative model based on the event's public measures, not a tick-by-tick measurement; the cumulative-liquidation figure of "reported liquidations exceeding $19 billion within about one day" is a reported measure [21] and differs in meaning from the open-interest-contraction measure [23]; source: FTI Consulting 2025 [21], CoinGlass 2026 [22])

Ali (2025) [23] dissects three key risk amplifiers in this event through a market-microstructure lens. First, the procyclicality of unified margin: when multiple assets serve as unified margin, the contagion path unfolds clearly—initial shock → decline in the value of margin assets → cross-asset forced liquidation → decline in the price of the liquidated asset → further margin shrinkage—forming a cross-asset liquidation spiral, for which Brunnermeier and Pedersen's (2009) [24] framework of the margin spiral provides a rigorous analytical basis. Second, exchange infrastructure bottlenecks: during periods of sharp price movement, APIs and interfaces stall, and traders cannot top up margin or close positions manually in time. Third, insurance-fund strain and auto-deleveraging risk: the insurance fund is funded by the surplus when positions are liquidated at prices better than the bankruptcy price. In most deleveraging events the fund is merely strained, but in extreme cases it can be exhausted, triggering ADL. ADL forcibly cuts the positions of the winning side, stripping away their profits—an additional blow that further shakes market confidence. The second cost of abolishing the expiry date becomes concrete here: the expiry, rolling, daily mark-to-market, and margin system of traditional futures provide periodic repricing windows; perpetual futures have no expiry window, so leverage accumulates more easily and continuously and is ultimately released through a catastrophic deleveraging.

Similar leverage accumulation and catastrophic deleveraging have played out repeatedly in the digital-asset market. The chain-liquidation event of May 2021 followed a similar accumulation–criticality–deleveraging–recovery path, providing cross-period empirical support for the model's cyclicality hypothesis. The deleveraging triggered by the collapse of FTX in November 2022 must be classified separately: it was counterparty-risk-type deleveraging (exchange failure → collapse of trust → forced deleveraging), fundamentally different in trigger from the market-endogenous deleveraging of May 2021 and October 2025 (leverage accumulation + price shock → liquidation cascade). This chapter's leverage-cycle model primarily explains market-endogenous deleveraging; counterparty-risk-type deleveraging involves the broader problem of institutional trust. The October 2025 event, larger in scale and faster in pace, further confirmed the systemic nature of this dynamic.

The recovery phase is asymmetric: the crash is measured in hours, recovery in days to weeks. Leverage is cleared out, open interest falls to a low level, market makers reenter, liquidity is restored, the market returns to the accumulation phase, and the cycle begins again.

The perpetual futures leverage cycle is highly isomorphic to Minsky's classic credit cycle in the dynamics of financial-leverage accumulation and collapse, but it lacks the dimensions of capital misallocation and overcapacity found in the real economy. Brunnermeier and Pedersen's (2009) [24] theory of the margin spiral may be the more precise benchmark, since it directly models the evaporation of liquidity and forced deleveraging under margin constraints and fits the microstructure of the perpetual futures market more closely. The two models differ significantly in speed, management tools, and information visibility. Table 2-2 compares them along four core dimensions, showing why the perpetual futures cycle is faster, more severe, and thinner in institutional buffers, yet at the same time more amenable to early warning because of the transparency of on-chain data.

| Dimension | Traditional credit cycle | Perpetual futures leverage cycle |
| :--- | :--- | :--- |
| Speed | Measured in years; a full cycle lasts 5 to 10 years | Measured in weeks to months; a full cycle within a few months |
| Cycle-management tools | Central-bank rate hikes and cuts, macroprudential policy, lender of last resort | The funding rate passively reflects the market state; no active steering tool |
| Deleveraging mode | Policy guidance, orderly institutional bankruptcy, and central-bank liquidity injection | Liquidation cascade; market-crash-style automatic deleveraging; limited insurance-fund capacity |
| Information visibility | Leverage accumulation is usually invisible, attributed after the fact | Aggregate-level open-interest (OI) and rate data are public, but key risk-distribution information (the distribution of liquidation thresholds, whale concentration, cross-platform net exposure) remains opaque on CEXs; on-chain DEX data are more transparent but limited in share. Visible does not mean manageable |

**Table 2-2.** Comparison of the traditional Minsky credit cycle and the perpetual futures leverage cycle (Data source: compiled by the author, drawing on Minsky (1986) [19] and Brunnermeier & Pedersen (2009) [24])

Is the liquidation cascade an endogenous feature of perpetual futures, or can mechanism design mitigate it? Why does liquidity disappear precisely when it is most needed? Of the price crash produced by liquidations, how much is information and how much is noise? Can leverage accumulation during calm periods be monitored and flagged in real time? These questions are taken up, respectively, in Chapter 11 on liquidation-mechanism design, Chapters 19 through 21 on the economics of liquidity, Chapter 22 on mechanism-driven volatility, and Chapters 23 and 24 on volatility forecasting and Minsky-moment early warning.

### 2.4.4 Risks unique to DEXs

The structural risks of centralized exchanges have grown increasingly exposed (the collapse of FTX being the most extreme case), driving significant growth in the market share of decentralized perpetual futures exchanges. Extrapolating from full-year volume ($6.7 trillion / $86.2 trillion), the ratio of DEX perpetual to CEX perpetual volume in 2025 was roughly 7.8%, well above the roughly 2.1% level in early 2023 [1] [2]. Total DEX perpetual volume in 2025 was about $6.7 trillion, up roughly 3.5 times over the prior year (for the data basis, see the opening of this chapter). DEX nominal volume may be inflated by trade-mining incentives and wash trading, so real economic volume may be lower than the nominal figure. With its distinctive architectural choices—a purpose-built Layer-1 app-chain, an on-chain central limit order book, and the HLP market-making vault—Hyperliquid recorded about $2.9 trillion in DEX perpetual volume for the full year of 2025, roughly 43% of the CoinGecko top-10 DEX perpetual sample, and now represents the architectural direction of app-chains, on-chain order books, and protocol-run market-making vaults [1] [25]. Hyperliquid's own Layer-1 chain warrants careful assessment: the size and distribution of its validator set and the decentralization of its sequencer fall short of protocols that run on the Ethereum Layer 1, and the discretionary intervention at the validator level during the JELLY incident (discussed in Section 2.3.2) is precisely a symptom of this tension [17]. Chapter 9 analyzes in depth the gains and losses of its architectural choices.

The difference between centralized and decentralized perpetual futures exchanges is not merely one of technical architecture but a fundamental divide in trust paradigm. Table 2-3 compares the two models along five dimensions—asset custody, transparency, core risk, trading experience, and trust paradigm—providing a structured reference for the analysis of DEX-specific risks that follows.

| Feature | Centralized exchange | Decentralized exchange |
| :--- | :--- | :--- |
| Asset custody | The exchange's centralized wallet; users must trust the platform | Smart contract or self-custody; non-custodial |
| Transparency | Low; relies on audits and proof of reserves | Very high; all trades and funds are verifiable on-chain |
| Core risk | Operational opacity, asset misappropriation, and a single point of failure | Smart-contract vulnerabilities, oracle risk, and maximal extractable value (MEV) attacks |
| Trading experience | Smooth, low-latency, feature-rich | Rapidly improving, but limited by the performance of the underlying chain |
| Trust paradigm | Platform trust | Architectural trust |

**Table 2-3.** Core differences between centralized and decentralized perpetual futures exchanges (Data source: compiled by the author)

While achieving non-custodial operation and on-chain transparency, DEXs introduce several risks that do not exist, or are milder, on CEXs. Oracle risk arises because DEXs rely on external price oracles (such as Chainlink or Pyth) to obtain the spot prices used to calculate the mark price and trigger liquidations. The BIS analysis of the DeFi oracle problem notes that an on-chain financial system's reliance on external price sources reintroduces issues of data quality, price-source manipulation, and governance trust at the protocol layer [26]. In low-liquidity token markets, an attacker can influence the oracle's quote by manipulating a few price sources, triggering a liquidation in their own favor. Oracle price updates carry an inherent delay (typically on the order of seconds), which in extremely volatile markets can open a significant gap between the mark price and the true market price, throwing off the timing of liquidations. An interruption in oracle service can pause a DEX's liquidation system, so that accumulated risk is released all at once when the oracle recovers.

MEV-based attacks are the second class of risk unique to DEXs. On a public blockchain, pending transactions are visible to everyone in the mempool, and block builders can extract value by reordering transactions. A sandwich attack—in which an MEV searcher, on detecting a user's large trade, inserts its own transactions before and after it to profit from the price slippage—worsens the user's execution price. Liquidation opportunities are visible to all, so multiple liquidators compete to execute, producing gas competition, and part of the liquidation profit is converted into validator income. The sandwich attack targets mainly DEXs based on the automated market maker (AMM) model; for DEXs that use a central limit order book (such as Hyperliquid), MEV risk shows up more as front-running and information-advantage problems at the sequencer level, tied directly to how decentralized the sequencer is. On a CEX, the exchange controls trade ordering and there is no public-mempool MEV problem, but a DEX exposes trade ordering to all block producers.

Smart-contract vulnerabilities are the third class of risk unique to DEXs. All of a DEX's logic is coded in smart contracts, and once a contract is deployed its logic is public to all, so attackers can audit it in search of exploitable vulnerabilities. Security reports from firms such as Immunefi show that DeFi and on-chain trading protocols have long suffered fund losses from smart-contract vulnerabilities, private-key leakage, and economic attacks [27]; the DeFi policy recommendations of the International Organization of Securities Commissions (IOSCO) likewise list governance, code, oracles, and cross-chain dependencies as important risk surfaces [28]. Although auditing and formal verification keep improving, zero vulnerabilities remains an unattainable goal, especially in a perpetual futures system where modules such as liquidation logic, margin calculation, and multi-collateral management interact in highly complex ways. Upgradeable smart contracts fix vulnerabilities but raise the trust question of who holds the authority to upgrade; if that authority rests with a few, the promise of decentralization weakens. Non-custodial operation eliminates the risk of platform misappropriation but introduces contract-vulnerability and admin-key risk, and the admin authority of an upgradeable contract is essentially a form of *code custody*. Chapter 28 systematically discusses this tension in the context of trust-stack design.

Regulatory-compliance risk likewise cannot be ignored. In 2023, the enforcement action by the Commodity Futures Trading Commission (CFTC) against several DeFi protocols, including Opyn, ZeroEx, and Deridex, showed that decentralization is not grounds for regulatory exemption [29]. Satisfying anti-money-laundering and know-your-customer (AML/KYC) requirements in a permissionless environment is a dilemma both technical and philosophical: there is a fundamental contradiction between mandatory identity verification and permissionless access. Protocol developers and governance participants may face personal legal liability, a risk that deeply constrains the long-term development of the DEX ecosystem.

DEXs also face two further risks. Governance-concentration risk arises when a few token holders or a team control key protocol parameters (such as the margin ratio, liquidation thresholds, and rate parameters), which may amount to de facto centralization. Cross-chain-bridging risk arises when assets are bridged to an app-chain: once a bridge contract is breached, user assets face systemic loss.

Together, these risks make up the DEX's trust-tax substitution: the CEX's trust tax is reliance on a centralized operator, while the DEX's is reliance on the correctness of code, the reliability of oracles, and the fairness of block producers. Chapter 1's risk-surface-transformation framework is confirmed once again: decentralization does not eliminate risk but transforms it from one surface to another.

### 2.4.5 Regulatory evolution

The regulatory evolution of perpetual futures passed through four phases. In the vacuum period, from 2016 to 2020, regulators had little awareness of or attention to crypto derivatives, the market grew in an almost unregulated environment, and exchanges such as BitMEX and Binance offered high-leverage trading worldwide without identity verification or anti-money-laundering review. In the enforcement-based period, from 2020 to 2023, regulators began to state their positions through enforcement actions: in 2020 the CFTC filed suit against BitMEX [30], and the collapse of FTX directly accelerated the tightening of regulation. The hallmark of this period, however, was the absence of a clear rule framework, so participants did not know what was compliant, only what would be prosecuted. In the framework-building period, from 2023 to 2025, major economies shifted from enforcement to legislation. The European Union's Markets in Crypto-Assets (MiCA) regulation applied in phases: the rules for asset-referenced tokens and e-money tokens from June 30, 2024, and the rules for crypto-asset service providers (CASPs) and other main provisions from December 30, 2024 [31]. MiCA primarily governs the issuance of, and service providers for, crypto assets not regulated as financial instruments under existing EU financial-services law; a crypto derivative, if it constitutes a financial instrument, is more likely to be governed under the MiFID II framework. ESMA's guidelines explain that the boundary between MiCA and MiFID II depends on the rights conferred by the asset, the contract structure, and the trading arrangement, rather than on the single factor of *whether it is on-chain* [12]. MiCA thus covers the spot side and the CASP framework, while MiFID II covers the financial-instrument and derivatives side. In the United States, the jurisdictional dispute between the U.S. Securities and Exchange Commission (SEC) and the CFTC is the core obstacle, and Congress is accelerating legislation to clarify which crypto assets are securities and which are commodities. Hong Kong implemented a licensing regime for virtual asset service providers (VASPs) in 2023, and Singapore applies a differentiated regime to crypto derivatives through its Monetary Authority. The compliance-transition period began in 2025, as market participants undertake compliance retrofits under a clear framework and compliance becomes a necessary condition for market access.

Yet a fundamental tension exists between the global nature of the perpetual futures market and the territoriality of regulation. When major economies tighten regulation, trading volume may shift to offshore jurisdictions with looser rules, and FTX's exploitation of the Bahamian regulatory environment is a cautionary precedent. The persistence of regulatory arbitrage may weaken the substantive effect of the compliance transition, and progress in international regulatory-coordination mechanisms—such as the cross-border coordination frameworks of the Financial Stability Board (FSB) and IOSCO—will be the decisive factor.

The dual effect of regulation must be confronted directly. In the short term, it constrains the market: it limits the leverage available to retail traders, raises exchanges' compliance costs, and suppresses some innovation—DeFi protocols in particular face the technical and philosophical dilemma of satisfying identity-verification requirements in a permissionless environment. In the long term, it enables the market: a more compliant, more transparent market with better risk controls is a prerequisite for attracting institutional investors at scale; regulatory clarity reduces uncertainty and lowers the compliance-risk premium, and consumer-protection measures reduce fraud and manipulation, strengthening the market's overall credibility and sustainability. Chapter 26 develops a systematic economic analysis of regulation.

## 2.5 Chapter summary

The 400-year evolution of derivatives is, in essence, three leaps in trust technology. Forward contracts rely on personalized trust; their radius of trust is limited to a circle of acquaintances, and market size is strictly constrained because the cost of trust is too high. Futures contracts achieve institutionalized trust through standardization and the CCP, extending the radius of trust to strangers worldwide, but they depend on a centralized institution and introduce the inherent contradiction of concentrating risk in order to manage it. Perpetual futures introduce algorithmic anchoring through the funding rate and margin rules, reducing reliance on the traditional CCP; but in the CEX setting they still depend on platform rules and operation, while in the DEX setting they shift trust to code, oracles, sequencing, and governance. Each leap responds to the trust bottleneck left by the previous one, while also generating a new trust tax.

The three structural conflicts between traditional futures and digital assets—temporal mismatch, liquidity fragmentation, and trust mismatch—make it infeasible to transplant traditional futures directly into the digital-asset world. The two core design innovations of perpetual futures (abolishing the expiry date and the funding rate) and their market outcome (the concentration of liquidity in a single contract) address these three conflicts directly; the causal logic of their mutual dependence (a soft anchor replacing a hard anchor) is analyzed in Section 2.2.4.

The core theoretical contribution of this chapter is cost awareness. Abolishing the expiry date eliminates roll costs but weakens three economic functions: the hard anchor of expiry convergence becomes the soft anchor of the funding rate, the periodic repricing window gives way to continuous position-holding, and the termination date for market-making risk becomes indefinite. These three costs explain many of the distinctive phenomena in the perpetual futures market. The threefold economic nature of the funding rate—the continuous form of the traditional cost of carry, a real-time market-sentiment indicator, and the foundation for an on-chain-native carry asset—upgrades perpetual futures from a trading tool into financial-information infrastructure and the foundation for a new asset class. The leverage-cycle dynamics reveal that the liquidation cascade is an endogenous, Minsky-style cycle of the perpetual futures market, faster than the traditional credit cycle and lacking central-bank-style tools to manage it. The October 2025 event (roughly $19 billion by both the reported-liquidation and open-interest-contraction measures, detailed in Section 2.4.3 [21] [23]) provides direct evidence for this dynamic.

Perpetual futures are both a product of the trust tax—the trust mismatch of traditional futures gave rise to them—and a restructuring of it: CEX perpetuals concentrate trust in platform rules, risk control, and asset custody, while DEX perpetuals shift trust to code, oracles, sequencing, and governance. Compression is not free, and perpetual futures introduce their own trust tax: the threefold cost of abolishing the expiry date, the endogenous fragility of the leverage cycle, the new risk surfaces of smart contracts and oracles, and the transitional cost of regulatory uncertainty.

The next chapter closes out Part 1. Chapter 1 established the analytical framework of the trust tax, and this chapter set out the logic and cost of the birth of perpetual futures; Chapter 3 now argues that perpetual futures are not merely a crypto-native tool but represent a general direction in the evolution of financial infrastructure, pushing the narrative of trust-tax compression toward a more macroeconomic perspective. As the core object of analysis in this book, perpetual futures run through all subsequent chapters: from mechanism deconstruction to price discovery, from the arbitrage ecosystem to the economics of liquidity, from volatility dynamics to market-quality assessment, and from decentralization engineering to the future in the age of AI.

## References

[1] CoinGecko. (2026). *2025 annual crypto industry report*. https://www.coingecko.com/research/publications/2025-annual-crypto-report

[2] CoinGecko. (2026). *The rise of perpetuals and perp DEXs*. https://www.coingecko.com/learn/rise-of-perpetuals-and-perp-dexs

[3] Alexander, C., Choi, J., Park, H., & Sohn, S. (2020). BitMEX bitcoin derivatives: Price discovery, informational efficiency, and hedging effectiveness. *Journal of Futures Markets*, *40*(1), 23–43. https://doi.org/10.1002/fut.22050

[4] Robertson, K., & Zhang, R. (2025). Price discovery in bitcoin spot and futures markets. *Journal of International Money and Finance*, *159*, 103415. https://doi.org/10.1016/j.jimonfin.2025.103415

[5] Hull, J. C. (2022). *Options, futures, and other derivatives* (11th ed.). Pearson.

[6] Black, F., & Scholes, M. (1973). The pricing of options and corporate liabilities. *Journal of Political Economy*, *81*(3), 637–654. https://doi.org/10.1086/260062

[7] Garbade, K. D., & Silber, W. L. (1983). Price movements and price discovery in futures and cash markets. *The Review of Economics and Statistics*, *65*(2), 289–297. https://doi.org/10.2307/1924495

[8] Shiller, R. J. (1993). Measuring asset values for cash settlement in derivative markets: Hedonic repeated measures indices and perpetual futures. *The Journal of Finance*, *48*(3), 911–931. https://doi.org/10.1111/j.1540-6261.1993.tb04024.x

[9] Hayes, A. (2016, May 13). *Announcing the launch of the perpetual XBTUSD leveraged swap*. BitMEX Blog. https://www.bitmex.com/blog/announcing-the-launch-of-the-perpetual-xbtusd-leveraged-swap

[10] CME Group. (2025). *Bitcoin futures and options on futures*. https://www.cmegroup.com/markets/cryptocurrencies/bitcoin/bitcoin.html

[11] Merton, R. C. (1973). Theory of rational option pricing. *Bell Journal of Economics and Management Science*, *4*(1), 141–183. https://doi.org/10.2307/3003143

[12] European Securities and Markets Authority. (2024). *Guidelines on the conditions and criteria for the qualification of crypto-assets as financial instruments*. https://www.esma.europa.eu/document/guidelines-conditions-and-criteria-qualification-crypto-assets-financial-instruments

[13] Financial Conduct Authority. (2020). *FCA bans the sale of crypto-derivatives to retail consumers*. https://www.fca.org.uk/news/press-releases/fca-bans-sale-crypto-derivatives-retail-consumers

[14] Ackerer, D., Hugonnier, J., & Jermann, U. (2025). Perpetual futures pricing. *Mathematical Finance*. Advance online publication. https://doi.org/10.1111/mafi.70018

[15] Schmeling, M., Schrimpf, A., & Todorov, K. (2023; revised 2025). *Crypto carry* (BIS Working Papers No. 1087). Bank for International Settlements. https://www.bis.org/publ/work1087.htm

[16] Ethena Labs. (2026). *Ethena documentation*. https://docs.ethena.fi/

[17] Hyperliquid Community Docs. (2025). *March 26 JELLY incident*. https://hyperliquid-co.gitbook.io/community-docs/introduction/roadmap/2025-26-03_incident

[18] Board of Governors of the Federal Reserve System. (2023). *Regulation T: Credit by brokers and dealers*. https://www.ecfr.gov/current/title-12/chapter-II/subchapter-A/part-220

[19] Minsky, H. P. (1986). *Stabilizing an unstable economy*. Yale University Press.

[20] Danielsson, J., Shin, H. S., & Zigrand, J.-P. (2012). Endogenous and systemic risk. In *Quantifying Systemic Risk* (pp. 73–94). University of Chicago Press. https://doi.org/10.7208/chicago/9780226921969.003.0004

[21] FTI Consulting. (2025, December 24). *Crypto crash Oct 2025: When leverage met liquidity*. https://www.fticonsulting.com/insights/articles/crypto-crash-october-2025-leverage-met-liquidity

[22] CoinGlass. (2026). *2025 crypto derivatives market annual report*. https://www.coinglass.com/learn/2025-annual-report-en

[23] Ali, Z. (2025, October 15). Anatomy of the Oct 10–11, 2025 crypto liquidation cascade: Macroeconomic triggers, market microstructure, and systemic risk lessons. *SSRN Electronic Journal*. https://doi.org/10.2139/ssrn.5611392

[24] Brunnermeier, M. K., & Pedersen, L. H. (2009). Market liquidity and funding liquidity. *Review of Financial Studies*, *22*(6), 2201–2238. https://doi.org/10.1093/rfs/hhn098

[25] Hyperliquid. (2025). *Hyperliquid documentation*. https://hyperliquid.gitbook.io/hyperliquid-docs

[26] Bank for International Settlements. (2023). *The oracle problem and the future of DeFi* (BIS Bulletin No. 76). https://www.bis.org/publ/bisbull76.pdf

[27] Immunefi. (2024). *Crypto losses in 2023*. https://immunefi.com/research/

[28] International Organization of Securities Commissions. (2023). *Policy recommendations for decentralized finance*. https://www.iosco.org/library/pubdocs/pdf/ioscopd754.pdf

[29] Commodity Futures Trading Commission. (2023). *CFTC orders operators of three DeFi protocols to pay penalties for illegal derivatives trading* [Press release]. https://www.cftc.gov/PressRoom/PressReleases/8774-23

[30] Commodity Futures Trading Commission. (2020). *CFTC charges BitMEX owners with illegally operating a cryptocurrency derivatives trading platform* [Press release]. https://www.cftc.gov/PressRoom/PressReleases/8270-20

[31] European Parliament & Council of the European Union. (2023). Regulation (EU) 2023/1114 on markets in crypto-assets (MiCA). *Official Journal of the European Union*. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32023R1114