Understanding liquidations
Separate published forced closures, inferred pressure and observed coverage.
A liquidation is an exchange-managed reduction or closure of a position that no longer meets its margin requirements. The process depends on the venue, contract and margin mode. A liquidated long can create sell pressure; a liquidated short can create buy pressure. That does not guarantee a price move or a chain reaction.
Published events are a sample of activity
The exchange feed is not a complete census of all liquidations. Publication limits can omit events during busy periods. Recorded counts and USD sums describe received publications, and cannot establish the true total, the number of traders affected or how much leverage remains.
The liquidation intensity reference explains the normalized Research reading and its feed limits. The clustering reference explains a different statistic based on inter-event gaps. Its early values are influenced by initialization; a high value does not prove a self-exciting cascade.
Keep the heatmap separate
The liquidation heatmap projects pressure under model assumptions. A bright zone is not a confirmed inventory of positions waiting to close. Covered Hyperliquid position data and published liquidation prints are separate evidence sources with their own limits.
Inspect an event, then broaden the question
Open an archived event, read its measured window and replay access, and inspect the sequence. Then use the Research Workbench to test an explicit condition across a wider record. Selected dramatic examples alone cannot establish how often a cascade occurs.
One published study asks exactly that kind of question: after $5M of liquidations in an hour, does the move keep going? It reports every match beside an unconditional rate over the same markets and window. Its scope is its own; read it before applying the result elsewhere.
Looking for a calculation? Open the Reading library. For recorded exercises, browse guided lessons.