Autodeleveraging as Online Learning

TL;DR

This paper provides a concise formalization of ADL as online learning on a PNL-haircut domain: at each round, the venue selects a solvency budget and a set of profitable trader accounts to cover shortfalls up to the solvency budget, with the aim of recovering exchange-wide solvency.

Abstract

Autodeleveraging (ADL) is a last-resort loss socialization mechanism used by perpetual futures venues when liquidation and insurance buffers are insufficient to restore solvency. Despite the scale of perpetual futures markets, ADL has received limited formal treatment as a sequential control problem. This paper provides a concise formalization of ADL as online learning on a PNL-haircut domain: at each round, the venue selects a solvency budget and a set of profitable trader accounts. The profitable accounts are liquidated to cover shortfalls up to the solvency budget, with the aim of recovering exchange-wide solvency. In this model, ADL haircuts apply to positive PNL (unrealized gains), not to posted collateral principal. Using our online learning model, we provide robustness results and theoretical upper bounds on how poorly a mechanism can perform at recovering solvency. We apply our model to the October 10, 2025 Hyperliquid stress episode. The regret caused by Hyperliquid's production ADL queue is about 50\% of an upper bound on regret, calibrated to this event, while our optimized algorithm achieves about 2.6\% of the same bound. In dollar terms, the production ADL model over liquidates trader profits by up to \$51.7M. We also counterfactually evaluated algorithms inspired by our online learning framework that perform better and found that the best algorithm reduces overshoot to \$3M. Our results provide simple, implementable mechanisms for improving ADL in live perpetuals exchanges.

Figures (13)

• Figure 1: Observation model and replay workflow used for policy evaluation on October 10, 2025.
• Figure 2: Lifecycle of an ADL round (Figure 2a--2b).
• Figure 3: Horizon sensitivity of production overshoot $O(\Delta)$, where $\Delta$ is the markout/holding-time evaluation offset.
• Figure 4: Cumulative fairness component by policy.
• Figure 5: Per-round policy performance against needed budget targets.

Theorems & Definitions (38)

• Proposition 1: Deficit-weighted severity bound with explicit constants
• Proposition 2: Static regret bound
• proof
• Proposition 3: Dynamic regret with comparator variation
• proof
• Proposition 4: Loss-specific gradient and diameter constants
• proof
• Proposition 5: Sharper dynamic bound for path-dependent rounds
• proof
• Definition 1: Queue allocation map