🌋 Hyperliquid flags 100× gap in CEX liquidation reports
posted 13 Oct 2025
Hyperliquid co‑founder & CEO Jeff Yan argues CEX liquidation feeds aren’t comparable to on‑chain logs and can miss burst events by up to 100×. He cites Binance docs that publish only one liquidation per 1000ms (effectively per second), while Hyperliquid records every order, trade and liquidation on‑chain for anyone to audit.
Yan frames transparency and neutrality as the point of fully on‑chain execution: anyone can verify the chain’s behavior in real time, including whether liquidations were triggered and filled fairly. The remarks follow a volatile weekend that revived scrutiny of venue data quality and risk controls after a market‑wide cascade.
At the core of his claim is a line in Binance’s documentation: when many liquidation orders arrive within the same 1000‑millisecond window, the public stream pushes only the latest one; if nothing happens in that interval, it stays silent. During stress, liquidations come in bursts, so the visible count can undershoot reality by a wide margin, even though the internal engine may be processing far more.
For on‑chain systems like Hyperliquid, liquidation events are written to the ledger alongside orders and trades, creating an auditable trail and a real‑time solvency view. That difference matters for anyone who models risk or routes orders across venues. If public CEX feeds clip peak activity, risk engines read calmer conditions than traders actually face, hedges lag, and displayed liquidity appears sturdier than it is.
Community reaction split between support for on‑chain verifiability and questions about Hyperliquid’s own design. Developer Bojan Angjelkoski argued: “Being on‑chain doesn’t make it decentralized; validators merely attest and the code isn’t open source.” From the user side, a top trader known as The White Whale urged safeguards: “Consider false‑wick liquidation protection like Drift’s.”
Both points underscore that transparency of data and resilience of the mechanism are separate axes that users will judge.
Binance, meanwhile, has indicated it will compensate affected users from the recent turmoil and revisit elements of its reporting and risk management. Whether centralized venues publish higher‑fidelity feeds, add raw event backfills, or at least document rate limits and aggregation behaviors more prominently will set expectations for market participants and developers relying on those streams.
In practice, the industry now faces a choice: accept opaque streams that compress bursts into neat intervals, or move toward feeds – on‑chain or centralized – that expose the full texture of stress. Exchanges that document limits clearly, ship higher‑resolution public data, or let users reconcile streams against post‑hoc logs will help rebuild trust. If on‑chain venues continue to pair verifiability with sensible user protections, they gain a stronger case that markets, money and media can live in the open without sacrificing fairness.
