Autodeleveraging: Impossibilities and Optimization

TL;DR

The paper formalizes autodeleveraging (ADL) in perpetual futures as a last-resort mechanism to preserve exchange solvency when liquidations fail, revealing a fundamental trilemma among solvency, revenue, and fairness. It develops a rigorous, venue-agnostic model that separates severity (how much to socialized) from haircuts (who pays), and proves that no static policy can simultaneously optimize all three objectives in large markets with heavy-tailed risks. It shows pro-rata fairness is unique under axiomatic and welfare criteria, while queue-based ADL concentrates losses on the top winner and is less fair; dynamic, risk-aware strategies (RAP, MDIC) better balance solvency, fairness, and revenue, albeit with trade-offs. Empirically, using Hyperliquid’s October 10, 2025 event data, the authors demonstrate substantial inefficiencies in production-queue ADL and show that near-optimal dynamic policies could have reduced excess haircuts by roughly 98% across observed shocks, while maintaining solvency. The work also develops a Stackelberg dynamic framework to optimize long-term exchange value and demonstrates a phase transition where dynamic control is necessary to preserve the trilemma’s objectives, providing practical guidance for ADL design and policy evaluation in large crypto derivatives venues.

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.

Figures (13)

• Figure 1: Sorted equity profiles for stylized liquidation examples. Negative positions (red) appear on the left, positive positions (green) on the right. Dashed lines highlight the bankruptcy level and liquidation triggers.
• Figure 2: ADL severity example comparing queue and pro-rata coloring. Purple shading equals the negative equity mass (deficit) while blue shading shows the haircut mass allocated to winning traders. The queue panel’s dashed blue block at rank 2 highlights residual equity when the queue method allows partial closures; exchanges that close winners fully ( e.g. Hyperliquid) would shave this bar completely. Haircut mass matches deficit mass in each panel, illustrating severity $\theta=0.50$.
• Figure 3: ADL overshoot ($H_t - D_t$). Dashed: per-shock overshoot. Solid: cumulative overshoot. The static policy $\pi_{\text{queue}}$ accumulates $\approx\$653$M of excess haircutting. Dynamic policies ($\pi_{\text{md}}, \pi_{\text{vec}}$) keep overshoot negligible by construction, while $\pi_{\text{exp}}$ maintains it within a bounded constant factor.
• Figure 4: Outstanding negative equity (cumulative residual deficits). Lower is better. Despite its aggression, $\pi_{\text{queue}}$ fails to clear the deficit effectively due to feedback loops. $\pi_{\text{exp}}$ achieves competitive solvency without the welfare cost.
• Figure 5: Largest winner haircut (USD). $\pi_{\text{queue}}$ places a massive, spikey burden on single participants. Dynamic policies ($\pi_{\text{md}}, \pi_{\text{exp}}$) distribute the load, respecting fairness constraints.

Theorems & Definitions (61)

• Proposition 2.1: Trilemma, Informal
• Proposition 5.1: Informal
• Proposition 5.2: Informal
• Proposition 5.3: Queue Maximizes Damage to Top Winner
• Proposition 6.1
• Proposition 6.2: Informal
• Proposition 7.1: Informal
• Proposition 8.1: Informal
• Proposition 8.2: No-Wait Condition
• Proposition 8.3: Informal