Paper: arXiv 2512.01112
Authors: Tarun Chitra
Abstract
Autodeleveraging (ADL) is a last-resort loss socialization mechanism for perpetual futures venues. It is triggered when solvency-preserving liquidations fail. Despite the dominance of perpetual futures in the crypto derivatives market, with over $60 trillion of volume in 2024, there has been no formal study of ADL. In this paper, we provide the first rigorous model of ADL. We prove that ADL mechanisms face a fundamental \emph{“trilemma”}: no policy can simultaneously satisfy exchange \emph{“solvency”}, \emph{“revenue”}, and \emph{“fairness”} to traders. This impossibility theorem implies that as participation scales, a novel form of \emph{“moral hazard”} grows asymptotically, rendering `zero-loss’ socialization impossible. Constructively, we show that three classes of ADL mechanisms can optimally navigate this trilemma to provide fairness, robustness to price shocks, and maximal exchange revenue. We analyze these mechanisms on the Hyperliquid dataset from October 10, 2025, when ADL was used repeatedly to close $2.1 billion of positions in 12 minutes. By comparing our ADL mechanisms to the standard approaches used in practice, we demonstrate empirically that Hyperliquid’s production queue overutilized ADL by \(\approx 28\times\) relative to our optimal policy, imposing roughly $653 million of unnecessary haircuts on winning traders. This comparison also suggests that Binance overutilized ADL far more than Hyperliquid. Our results both theoretically and empirically demonstrate that optimized ADL mechanisms can dramatically reduce the loss of trader profits while maintaining exchange solvency.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper presents a formal impossibility theorem and optimal mechanism design with heavy theoretical modeling, fitting the high math complexity category, while also providing rigorous empirical analysis on a real-world dataset with quantitative results, fitting the high empirical rigor category.
Research Flowchart
flowchart TD
A["Research Goal: Rigorously Model & Optimize ADL"] --> B["Methodology: Formal Modeling & Trilemma Proof"]
B --> C{"Can ADL satisfy Solvency, Revenue, & Fairness?"}
C -- Trilemma Proof --> D["Impossibility Theorem: No 'Zero-Loss' Socialization"]
C -- Constructive Analysis --> E["Optimization: Three Classes of Optimal ADL Mechanisms"]
D --> F["Empirical Validation: Hyperliquid Oct 10, 2025 Dataset"]
E --> F
F --> G["Key Findings: Hyperliquid Overutilized ADL by ~28x<br>(~$653M Unnecessary Haircuts)"]
G --> H["Outcome: Optimized ADL Reduces Trader Losses & Maintains Solvency"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper provides the first rigorous model of autodeleveraging (ADL) and identifies three classes of ADL mechanisms that can optimally navigate the solvency‑fairness‑revenue trilemma.
Source sentence
In this paper, we provide the first rigorous model of ADL. On the positive side, we show that three classes of ADL mechanisms can optimally navigate this trilemma to provide fairness, robustness to price shocks, and maximal exchange revenue.
Assumes 6
- The analysis assumes a large‑market limit (n →∞) with heavy‑tailed distributions for winner endowments and loser shortfalls. statistical
Source sentence
C.1 Setup and Assumptions We work in the large-market limit (n →∞) under the heavy-tailed assumptions characteristic of crypto markets. Assumption A (regular variation).
- Standard Law of Large Numbers and Extreme Value Theory scaling limits apply to aggregate endowment and deficit. statistical
Source sentence
Assumption B (LLN and EVT).
- The insurance fund evolves according to the dynamics given in equation (15) with affine liquidation fees. model
Source sentence
We assume that the fund evolves via eq. (15) and model liquidation fees as affine in notional size: τt(∆q) = τ fix + ϕmark pt ∆q + ϕexec pexec t ∆q, 22 where τ fix is a fixed fee and ϕmark, ϕexec are proportional rates on mark and execution price, respectively.
- A risk model g mapping leverage to a risk score exists and can be constructed via convex duality to maximize solvency per dollar of haircut. model
Source sentence
Given a stochastic process for price shocks, one can construct an optimal risk model (using convex duality) g∗that maximizes the total solvency gained per dollar of winner position that is haircut by an ADL policy.
Show 2 more
- ADL is triggered only when solvency‑preserving liquidations fail. market
Source sentence
It is triggered when solvency-preserving liquidations fail.
- Exchange revenue for funding the insurance fund comes from transaction, liquidation, and funding fees. execution
Source sentence
The insurance fund is typically funded using a portion of the exchange’s revenue from transaction, liquidation, and funding fees.
Data
- Hyperliquid October 10, 2025 ADL event crypto-onchain public
Source sentence
We analyze these mechanisms on the Hyperliquid dataset from October 10, 2025, when ADL was used repeatedly to close \$2.1 billion of positions in 12 minutes. We evaluate the proposed ADL mechanisms using a reconstruction13 of the October 10, 2025 liquidation cascade on Hyperliquid [99].
Costs
Modeled
Fees are modeled as τt(Δq)=τfix+ϕmark·pt·Δq+ϕexec·pexec·Δq with Hyperliquid parameters (0, 20 bps, 10 bps).
Source sentence
(15) and model liquidation fees as affine in notional size: τt(∆q) = τ fix + ϕmark pt ∆q + ϕexec pexec t ∆q, 22 where τ fix is a fixed fee and ϕmark, ϕexec are proportional rates on mark and execution price, respectively. Using the Hyperliquid fee schedule (τ fix, ϕmark, ϕexec) = (0, 20 bps, 10 bps) (see §2.2.2), the fees are τ4(∆qE) = 0 + 20 · 10−4 · p4 ∆qE + 10 · 10−4 · pexec E ∆qE ≈0.00365.
Finds 5
- ADL mechanisms face a fundamental trilemma: no policy can simultaneously satisfy exchange solvency, revenue, and fairness. theoretical
Source sentence
We prove that ADL mechanisms face a fundamental \emph{trilemma}: no policy can simultaneously satisfy exchange \emph{solvency}, \emph{revenue}, and \emph{fairness} to traders.
- As participation scales, a novel form of moral hazard grows asymptotically, making zero‑loss socialization impossible. theoretical
Source sentence
This impossibility theorem implies that as participation scales, a novel form of \emph{moral hazard} grows asymptotically, rendering `zero-loss' socialization impossible.
- Optimized ADL mechanisms can dramatically reduce losses of trader profitability while maintaining exchange solvency. performance
Source sentence
Our results show both theoretically and empirically that optimized ADL mechanisms can dramatically reduce losses of trader profitability while maintaining exchange solvency.
- The excess profits lost by profitable traders during the October 10 event are estimated between $45.0M and $51.7M. performance figure from elsewhere in the paper
Source sentence
Our methodology suggests the excess profits lost by profitable traders is between \$45.0M and \$51.7M.
- Approximately $653.6M of positions were closed in the October 10 event, and Binance over‑utilized ADL more than Hyperliquid. performance figure from elsewhere in the paper
Source sentence
In terms of the positions liquidated, this corresponds to roughly \$653.6M of positions being closed. This comparison also suggests that Binance overutilized ADL far more than Hyperliquid.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 30 related papers). Coverage is partial, so this is neither replication nor refutation.
2 papers in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (143)
- On law-invariant coherent risk measures (2001)
- Spectral measures of risk: A coherent representation of subjective risk aversion (2002)
- Short Communication: A Primer on Perpetuals (2023) — link
- The Express Lane to Spam and Centralization: An Empirical Analysis of Arbitrum’s Timeboost (2025) — link
- TimeBoost: Do Ahead-of-Time Auctions Work? (2025) — link
- Concave pro-rata games (2023)
- Selfish Routing and the Price of Anarchy (2005)
- Intrinsic Robustness of the Price of Anarchy (2015) — link
and 135 more
- Algorithmic Game Theory (2007)
- Composable and Efficient Mechanisms (2013) — link
- Perpetual Demand Lending Pools — → 2502.06028
- State of Crypto Perpetuals 2024 (2025) — link
- Exchange Review — June 2025 (2025) — link
- What Is Auto-Deleveraging (ADL) and How Does It Work? (2019) — link
- Futures Insurance Fund and Liquidation Fees (2025) — link
- Insurance Fund (2025) — link
- Insurance Fund (2025) — link
- Insurance Fund (2025) — link
- OKX vs Bybit: Which Is Better for Crypto Derivatives? (2025) — link
- Liquidations (2025) — link
- Auto-deleveraging (2026) — link
- Portfolio Margin (2025) — link
- Hyperliquid ADL is Sybil resistant (2025) — link
- Protocol Vaults: Hyperliquidity Provider (HLP) (2025) — link
- HyperMultiAssetedADL: Hyperliquid liquidation replay (2025)
- USD $ -Margined and COIN-Margined Futures REST API: Open Interest Endpoints (2025) — link
- Lighter: Order-Book Perpetuals Exchange Whitepaper (2025) — link
- Important Changes: Auto-Deleveraging (2016) — link
- Auto Deleveraging (2025) — link
- Towards Understanding Cryptocurrency Derivatives: A Case Study of BitMEX (2021) — link
- Hyperliquid and the JELLY attack: Context, vulnerability and team solution (2025) — link
- Hyperliquid Adds New Safeguard Update After XPL Price Spike (2025) — link
- Hyperliquid responds to XPL’s spike: liquidation then ADL per protocol; no bad debt (2025) — link
- ‘Largest Ever’ Crypto Liquidation Event Wipes Out 6 , 300 Wallets on Hyperliquid (2025) — link
- BitVC Swaps ‘Socialized’ Futures Losses for New Risk System (2015) — link
- What Is Auto-Deleveraging (ADL) and How Does It Work? (2025) — link
- Auto-Deleveraging (ADL) — Documentation (2025) — link
- Wintermute boss denies plans to sue Binance over the Oct. 10 crash (2025) — link
- Wintermute’s Evgeny Gaevoy says the market maker isn’t suing Binance and ’never had plans to’ (2025) — link
- Review of the Principles of Life-Office Valuations (1952)
- Traders say Binance cut their Bitcoin shorts — here’s why it happened (2020) — link
- Calculus, Vol. 2: Multi-Variable Calculus and Linear Algebra with Applications (1969)
- Extremes and Related Properties of Random Sequences and Processes (1983)
- Heavy-Tail Phenomena: Probabilistic and Statistical Modeling (2007)
- Modelling Extremal Events: For Insurance and Finance (1997)
- Central Counterparties: Magic Relighting Candles? (2019) — link
- Introduction to Online Convex Optimization (2019) — link
- Online Learning and Online Convex Optimization (2012)
- Measuring Asset Values for Cash Settlement in Derivative Markets: Hedonic Repeated Measures Indices and Perpetual Futures (1993)
- Continuous Auctions and Insider Trading (1985)
- Optimal Execution of Portfolio Transactions (2001)
- No-Dynamic-Arbitrage and Market Impact (2010)
- How Markets Slowly Digest Changes in Supply and Demand (2009) — link
- Elements of Information Theory (2006)
- Optimization of Conditional Value-at-Risk (2000) — link
- Conditional Value-at-Risk for General Loss Distributions (2002) — link
- Minimum Capital Requirements for Market Risk (2019) — link
- Skin in the Game: Risk Analysis of Central Counterparties (2025) — link
- Loss Sharing in Central Clearinghouses: Winners and Losers (2024) — link
- Moral Hazard in Teams (1982)
- Adapt or Die (2025) — link
- Options, Futures, and Other Derivatives (2017)
- Linear Perpetual: Funding (2024) — link
- On the Redington theory of immunization (1987)
- GMX Docs: Perpetual Exchange and GLP (Liquidity Provider) Mechanics (2025) — link
- Perpetual Protocol v2 (Curie) Documentation (2025) — link
- The Theory of Incentives: The Principal-Agent Model (2002)
- Robustness and Linear Contracts (2015) — link
- Multi-Agent Contracts (2023) — link
- Dynamic Noncooperative Game Theory (1999)
- Convex Optimization (2004) — link
- Improved Price Oracles: Constant Function Market Makers (2020) — link
- The Geometry of Constant Function Market Makers — → 2308.08066
- Marktform und Gleichgewicht (1934)
- Game Theory (1991)
- Security and Game Theory: Algorithms, Deployed Systems, Lessons Learned (2011)
- Dynamic Programming and Optimal Control (2017)
- Does a Central Clearing Counterparty Reduce Counterparty Risk? (2011) — link
- The Economics of Central Clearing: Netting, Asymmetric Information, and the Sharing of Default Risk (2011) — link
- Counterparty Credit Risk and Credit Value Adjustment: A Continuing Challenge for Global Financial Markets (2015)
- Recovery of Financial Market Infrastructures (2014) — link
- Online Convex Programming and Generalized Infinitesimal Gradient Ascent (2003)
- 2025 Q1 Crypto Industry Report (2025) — link
- E-mini S&P 500 Equal Weight futures: Classic index gets a new look (2024) — link
- E-mini S&P 500 Margins (2025) — link
- Update on the Collateral Ratio Under Portfolio Margin and the Leverage & Margin Tiers of USD-M Perpetual Contracts (2025) — link
- Fundamentals of perpetual futures — → 2212.06888
- Perpetual futures pricing — → 2310.11771
- Liquidations: DeFi on a Knife-edge (2021)
- An analysis of the market risk to participants in the compound protocol (2020)
- Liquidity risks in lending protocols: Evidence from aave protocol (2022) — link
- An empirical study of defi liquidations: Incentives, risks, and instabilities (2021)
- Multi-Period Trading via Convex Optimization (2017) — link
- Modern Actuarial Risk Theory: Using R (2005)
- Inequalities: Theory of Majorization and Its Applications (2011)
- The Concept of Comonotonicity in Actuarial Science and Finance: Theory (2002) — link
- The Concept of Comonotonicity in Actuarial Science and Finance: Applications (2002) — link
- Increasing Risk: I. A Definition (1970) — link
- Rules for Ordering Uncertain Prospects (1969)
- The Efficiency Analysis of Choices Involving Risk (1969) — link
- Stochastic Orders (2007) — link
- Model-Independent Bounds for Variance Swaps (2012) — link
- Markov Decision Processes: Discrete Stochastic Dynamic Programming (1994)
- Ruin Probabilities (2010)
- On the Time Value of Ruin (1998)
- Stochastic Control in Insurance (2008)
- Computing the Optimal Strategy to Commit to (2006)
- Robust Game Theory (2006)
- Inherent Trade-offs in the Fair Determination of Risk Scores (2017) — link
- Fairness and Machine Learning (2019) — link
- A Survey on Bias and Fairness in Machine Learning (2021) — link
- Fairness Through Awareness (2012) — link
- A Reductions Approach to Fair Classification (2018) — link
- Equality of Opportunity in Supervised Learning (2016) — link
- Social Choice and Individual Values (1951)
- Microeconomic Theory (1995)
- Drift Protocol v2: Liquidation Module (2024) — link
- ADL Is Broken. Socialized Loss Is the Future (2025) — link
- Second Interim Report of FTX Trading Ltd. and Affiliated Debtors (2023)
- Optimal Inventory Policy (1951) — link
- Credible Auctions: A Trilemma (2020) — link
- Towards Robust Distributed Systems (2000) — link
- Credible, Strategyproof, Optimal, and Bounded Expected-Round Single-Item Auctions for All Distributions (2022) — link
- Credible, Truthful, and Two-Round (Optimal) Auctions via Cryptographic Commitments (2020) — link
- Domestic Financial Policies under Fixed and under Floating Exchange Rates (1962) — link
- Brewer’s Conjecture and the Feasibility of Consistent, Available, Partition-Tolerant Web Services (2002) — link
- A Quantitative Arrow Theorem (2014) — link
- Capital Mobility and Stabilization Policy under Fixed and Flexible Exchange Rates (1963) — link
- The Sleepy Model of Consensus (2017) — link
- Credible, Optimal Auctions via Public Broadcast (2024) — link
- Foundations of Distributed Consensus and Blockchains (2024) — link
- Streamlined Blockchains: A Simple and Elegant Approach (2019) — link
- Optimal Stopping and Free-Boundary Problems (2006)
- Analysis of ADL Mechanisms and Execution Costs (2025) — link
- Execution Costs vs Mechanism Design in ADL (2025) — link
- A Dynamic Model of the Duration of the Customer’s Relationship with a Continuous Service Provider: The Role of Satisfaction (1998)
- Counting Your Customers: Who Are They and What Will They Do Next? (1987)
- A Joint Model of Usage and Churn in Contractual Settings (2013)
- Managing Churn to Maximize Profits (2020)
- Retention Futility: Targeting High-Risk Customers Might be Ineffective (2018)
- Learning to Play Stackelberg Security Games (2016)
- Three Strategies to Success: Learning Adversary Models in Security Games (2016)
- Autodeleveraging, Hyperliquid, and the $653m Debate (2025)
4 of 143 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.