Why Can't Crypto Platforms Guarantee Zero Losses with Autodeleveraging?

Perpetual futures platforms, like those handling Bitcoin or Ethereum derivatives, rely on a mechanism called autodeleveraging (ADL). When a standard liquidation fails to cover a trader's loss, ADL steps in to "steal" part of winning positions, preserving the exchange's solvency.

A recent study on arXiv built the first mathematical model of ADL and uncovered a true trilemma: an exchange cannot simultaneously guarantee solvency, high revenue, and fairness to traders. In practice, the larger the platform, the higher the risk of a moral hazard: operators may rely on ADL to cover mistakes, while winning traders suffer unjust cuts.

The authors identified three classes of ADL mechanisms that best balance these goals, ensuring fairness, price‑shock resilience, and maximal exchange profit. Testing their proposals on Hyperliquid data from October 10, 2025, they found the platform used ADL 28 times more than needed, stripping about $653 million of legitimate trader gains. Binance appears to have over‑used ADL as well.

The takeaway is clear: optimizing ADL can dramatically cut unfair trader losses while keeping the platform safe. A reminder that even the most sophisticated crypto systems have limits and trade‑offs to manage.

The theoretical model introduced by the authors starts from a game‑theoretic formulation where the exchange and traders are players with opposing payoffs. Solvency is modeled as a budget constraint: the exchange must keep a positive capital buffer after each liquidation. Revenues stem from trading fees and ADL penalties, while fairness is quantified through an average loss function for winning traders.

The trilemma arises from an impossibility theorem proving that, for any ADL policy, at least one of the three conditions must be violated under high volatility. Formally, if solvency (C1) and revenue maximization (C2) are enforced, fairness (C3) is sacrificed, as the policy disproportionately hits profitable positions. Conversely, a fair policy reduces the exchange's profit margins, threatening its ability to absorb market shocks.

The three optimal ADL mechanism classes are: (i) dynamic‑threshold ADL, which adjusts cut rates based on current volatility; (ii) proportional‑reduction ADL, which distributes cuts linearly across winning positions; (iii) hybrid ADL, combining dynamic thresholds with a “safety‑queue” mechanism to cap maximum cuts.

The empirical analysis on Hyperliquid uses tick‑by‑tick data to reconstruct the full liquidation and ADL sequence. Comparing the real policy (over‑use of ADL) with the three optimal policies shows a 92% reduction in unjustified cuts and a 15% increase in net exchange revenue, thanks to better risk allocation. These findings suggest that finer regulation of ADL policies could boost transparency and trust in crypto futures markets without compromising platform financial stability.

Finally, the authors propose a real‑time monitoring framework based on stress‑test metrics to dynamically adjust ADL parameters. This approach would allow preventive intervention before the combination of volatility and trading volume pushes the system toward the moral‑hazard state described by the trilemma.