$138M and a 60% Shrink: How Far Has Unwinding Gone?
On August 19, TradingBeats reported Hyperliquid's MU-USDC perpetual open interest near $138 million, down over 60% from almost $300 million in July and the lowest in about two and a half months. Liquidation in the past 24 hours hit long positions but remained small: $77,700 for Kvyadav1:vishnuvamsi.eth, $55,900 for 0x4a...8f7b and $55,500 for 0x59...7716. Concentration is notable: the top address holds about $9.65 million, or 7%; the top five hold $35.92 million, or 26.1%; the top ten hold $54.36 million, or 39.5%. Price fell nearly 7% in a day with all liquidations on the long side and none on the short side. This is the verified factual baseline. The important question is not whether the event is bullish or bearish, but which customers, assets, services and time windows are affected. Search users also need to know whether funds remain accessible, whether published figures can be reproduced and what action a platform should take next. Reporting, statements by involved parties and analytical conclusions must remain separate, and any detail absent from the reviewed page is left unclaimed rather than reconstructed from assumption.
Top Ten at 39.5%: Why Structure Matters More Than Volume
Falling open interest combined with concentrated long liquidations suggests leveraged longs are being cleared passively. In thin liquidity, additional short pressure or more long stop-losses can amplify downside. The top ten controlling 39.5% means their reductions, transfers or liquidations will move price and funding materially. Monitoring totals without structure misses the real risk, and a single large account can dominate both price discovery and fee flows in low-liquidity conditions. Risk should be traced across the customer, account, wallet, counterparty and final asset. One alert establishes an association, not proof that the customer knowingly participated in misconduct. Amount share, direction, historical behavior, control of the sending address and subsequent interaction all affect the conclusion. A blanket restriction can create widespread false positives and encourage risky actors to fragment activity, so reviewers need both confirming and falsifying evidence with explicit conditions for escalating or closing a case.
A Concentration Dashboard: OI, Liquidation Side and Funding on One Timeline
Trustformer KYT can build a concentration dashboard tracking top-1, top-5 and top-10 shares, directional mix and rate of change, with alerts for rising concentration or abnormal single-address accumulation. Liquidation monitoring should separate sides and preserve addresses and amounts. Open interest, price, funding and liquidation data belong on one timeline to distinguish active from passive unwinding, and thresholds should tighten as open interest and liquidity decline together. Trustformer KYT should assign one case identifier and preserve source data, rule version, transaction hashes, entity labels and analyst reasoning. A tiered response is more defensible: monitor low-risk activity, request source-and-purpose evidence for medium-risk cases and restrict funds only when high-risk indicators converge. Daily replay should measure false positives, missed cases, handling time and appeal outcomes. The program must also compare activity before, during and after the event window, identify the entities responsible for deviations and document every override, creating an auditable decision trail for customers, compliance committees, regulators and external reviewers. Control effectiveness should be tested against changing counterparties, products and transaction patterns. Entity clustering must distinguish common infrastructure from common ownership, and data confidence should be shown beside every label. Periodic sampling by a second analyst prevents automated scores from becoming unsupported final judgments.