For each Hyperliquid perp, the price levels at which leveraged positions blow up. Built from live open interest + typical leverage distribution — no insider data, no guesses about who holds what. Use it to spot the next squeeze before it hits.
Rank by |funding APR| × √OI / √volume — big crowded position, one-way funding, thin tape. The classic squeeze cocktail.
| Asset | Mark | OI | Funding APR | Setup | Score | |
|---|---|---|---|---|---|---|
| Loading... | ||||||
Recent fills (≥$10K nominal) absorbed by Hyperliquid's HLP vault — the engine of last resort that takes the other side when positions get force-closed. Not 100% liquidations (HLP also market-makes), but the bigger the fill, the higher the chance.
Hyperliquid is the most transparent perp DEX in the world: every order book event, funding rate, open interest, and trade is on-chain. What's not publicly indexed is the per-position leverage choice of each trader — that data exists but isn't queryable at aggregate scale.
We take three observable facts: open interest (USD), funding rate (signal of long/short bias), and max leverage tier per asset. We then assume a documented leverage distribution typical of perp markets:
And a 60/40 long bias on the OI split, which we'll refine if Hyperliquid publishes per-asset crowd skew.
For each leverage band, the liquidation price is approximately mark × (1 − 0.95/leverage) for longs and the symmetric for shorts (we use 0.95, not 1.0, to account for maintenance margin being roughly 50% of initial margin on Hyperliquid).
The ladder isn't claiming this exact $X gets liquidated at $Y price. It's claiming this is the order of magnitude, and the relative shape — where the densest clusters sit — is what actually matters when you're trading the cascade. Coinglass and friends use a similar synthetic model under the hood. We just say so.
Indexing every Hyperliquid position-level update from the explorer (subgraph or websocket). On the roadmap — but it'd be 10x the data we'd otherwise need to store and the value-add over the synthetic model is modest. For now, transparency beats false precision.