A Hyperliquid system designed to absorb liquidations took on about $576m (£? not provided) of forced selling during the most intense minute of the October 2025 cryptocurrency crash, helping to keep the majority of those trades away from the platform’s public order book, according to a new research preprint.
The paper says approximately $641m was force-sold on Hyperliquid at 21:19 UTC on 10 October. Around $576m was directed to the venue’s backstop, while about $64m reached the public order book.
That distinction is important because a thinner order book can cause prices to fall further, potentially triggering additional closures of leveraged positions. Hyperliquid’s backstop is intended to interrupt that cycle by absorbing orders within the platform.
The research has not yet undergone peer review. Its direct measurements cover activity on Hyperliquid and do not represent the entire cryptocurrency market.
Under Hyperliquid’s liquidation rules, a position is initially subject to closure through market orders. In certain circumstances, however, a liquidator vault can assume the position instead. That vault operates as a component strategy within the Hyperliquidity Provider (HLP) protocol vault.
According to the study, the backstop absorbed 62.6% of forced-sale value away from the public order book after the liquidation event began. The selling was also concentrated over a short period: 87.8% of forced sales after the start occurred within 30 minutes, while 96.5% took place within an hour.
The paper recorded $733m in book-directed forced-sale value over a 15.7-hour period following the start of the event. Of that amount, $644m came during the initial “nucleation” phase.
The authors presented the 62.6% backstop figure as a separate off-book series. As a result, the figures describe different elements of the study’s measurement and should not be treated as a single combined liquidation total.
The researchers also assessed the cascade using a branching ratio. This measures the average number of further liquidations associated with each forced sale. A figure approaching one would suggest that a self-sustaining chain of liquidations was developing within the venue.
Hyperliquid’s structural estimate stayed below 0.2 in every regime measured by the study. It reached 0.195 during nucleation before falling to 0.140 at the peak. A separate amplification calculation produced a ratio of 0.122.
The authors say the findings indicate that Hyperliquid’s backstop reduced feedback within the platform at the height of the crash. They stress, however, that the conclusion applies only to activity inside Hyperliquid. Prices shared across different exchanges may still have contributed to the wider market’s liquidation cascade.
The research compares the mechanism with seven major Bitcoin perpetual futures cascades between 2022 and 2025. Part I of the study, previously reported by CryptoSlate, found no early-warning variable that remained consistent across all seven events. Part II focuses instead on the mechanism operating during a cascade.
Hyperliquid’s fill-log archive begins on 25 May 2025, meaning the October event is the paper’s only in-flight case study. The authors describe higher realised branching on platforms without a comparable backstop as a hypothesis that will require testing across multiple venues.
Hyperliquid is down 1.68% over the past 24 hours and is currently ranked ninth by market capitalisation.
The report was written by Liam Wright, also known as “Akiba”, a reporter, podcast producer and editor-in-chief at CryptoSlate. Wright believes decentralised technology has the potential to make…
CryptoSlate says it may use artificial intelligence tools to support research, editing and production workflows, while its editorial team remains responsible for published journalism. The publication also states that its writers’ opinions are their own and do not represent CryptoSlate’s views.
The information in the report is not investment advice, and CryptoSlate does not endorse any project mentioned or linked in the article. Buying and trading cryptocurrencies is considered a high-risk activity, and readers are advised to carry out their own due diligence before acting on the information. CryptoSlate accepts no responsibility for losses incurred through cryptocurrency trading and directs readers to its company disclaimers for further information.
CryptoSlate also invites readers to follow its signal service for market-moving updates.
