Hyperliquid Liquidation Cliff Concentration
See whether one tracked wallet dominates pending liquidation value within 2% and 5% of the mark before treating a band as broad vulnerability.
ETH pending liquidation bands
| Side / distance | Pending value / wallets | Largest wallet | Top three | HHI / effective contributors | Without largest | Top contributors (up to three) |
|---|---|---|---|---|---|---|
| Downside longs Within 2% (inclusive) |
$899K 2 wallets |
74.3% | 100.0% | 0.618182 1.62 effective |
$231K | |
| Downside longs Within 5% (inclusive) |
$7.15M 6 wallets |
50.0% | 90.8% | 0.361023 2.77 effective |
$3.57M | |
| Upside shorts Within 2% (inclusive) |
$41K 1 wallet |
100.0% | 100.0% | 1.000000 1.00 effective |
$0 | |
| Upside shorts Within 5% (inclusive) |
$41K 1 wallet |
100.0% | 100.0% | 1.000000 1.00 effective |
$0 |
Bands overlap: the 5% band includes the 2% band. Wallets are combined separately within each side and band.
Dated inputs. Evaluated . Shared position rows loaded . ETH mark 2,330.4; native market context updated (2 s ago). This is context freshness, not a per-market exchange event timestamp. Oldest included position check: . The 15-minute check cutoff is reapplied on every render, even when the reader returns cached rows.
Fresh tracked coverage, separate from band concentration. $2.07B fresh tracked position notional ÷ $2.56B (2 × $1.28B one-sided native ETH OI) = 80.7%. The numerator uses |size| × mark for all valid fresh open positions, including those without a usable pending liquidation level. This describes a partial tracked sample, not whole-market coverage.
Exclusions. Of 1475 rows from the bounded current-position reader: 0 missing, non-finite or zero sizes; 1220 missing, future or older-than-15-minute checks; 0 invalid wallet addresses; 29 missing, non-finite or non-positive liquidation levels; 0 levels already at or beyond the mark; 219 pending levels beyond 5%. 255 valid fresh open positions contribute to tracked notional. The shared reader already removes closed positions, positions superseded by a later full-account keyframe and wallets outside its 48-hour polling bound. Rows unavailable to that reader cannot be counted here.
Method. Pending liquidation value is |position size| × recorded liquidation price, aggregated by wallet before any display cap. Long levels must be below the mark; short levels must be above it. Each wallet share divides its pending value by that side and band's sampled total. HHI is the sum of squared wallet shares on a 0–1 scale; effective contributors = 1 / HHI. One wallet gives HHI 1 and one effective contributor; two equal wallets give HHI 0.5 and two. “Without largest” subtracts the largest wallet's value from the band total; it is a removal calculation, not a forecast. Empty bands have unavailable concentration.
Interpretation. Source: recorded open positions from Hyperliquid's public API through the existing wallet tracker, plus native mark/OI context. Only a sample of wallets is tracked, with uneven polling. Cross-margin liquidation prices move with balances, collateral and other positions; a recent check can still be out of date. Pending levels are neither completed liquidations nor guaranteed forced orders. These measures provide no cascade probability or leverage recommendation.