Hyperliquid · perpetual

BTC

Where forced selling sits, how much of it is trapped, and how big it is against the book standing in front of it.

Cannot defend their position
98%
$79.9M of $81.5M in forced exposure sits with wallets holding zero free collateral. They cannot move their liquidation price without depositing from outside.
observed 78 positions read from the venue's own clearinghouse state
Nearest cluster
-0.50%
$1.1M at 78,402
Forced exposure
$81.5M
within ±10%
Largest cluster
9.8%
of the standing book

Who holds it

From a uniform sample of wallets discovered in our BTC trade recording — not the biggest accounts ranked by activity. It is uniform inside that frame, not a sample of every venue account. How the sample was drawn →

Top wallet
33.7%
of the sampled notional
Top 10 wallets
64.9%
of the sampled notional
Median position
$2k
mean is $170k

Put against the book: the top 10 sampled wallets' combined position is $220.3M 86.2% of the $255.5M standing within ±1%. Not a claim about what happens if they move — a comparison of two things we already measure separately.

WalletPositionSince last scanFree collateral
0x020ca6… $114.4M $0
0xa445a0… $21.0M +$151k $0
0x09bc1c… $18.9M $-16,969,241 $5.4M
0x1367df… $15.8M $2k
0x6f97d3… $14.8M +$4.6M $133k

The largest wallets in the sample, and the change in each since its own previous appearance (9.5h ago) — a fact about two observations, not a signal about what comes next. How this is measured →

What happens if a cluster is reached? Less than the raw move suggests in the tested sample. Forced flow at 15 minutes returned +40.07 bps — and a random minute in the same hour returned 44.52 bps. The scale shown below is consistent with that result: a typical cluster is small against the measured book. The full test →

How big is any of this, really?

Every liquidation map scales its clusters against each other, so the biggest one always looks like a wall. Against the liquidity actually standing in front of it, it usually is not.

Book standing within ±1% of spot $255.5M
largest cluster
9.8% $25.0M at 81,159
median cluster
0.19% $478k at 79,189

Drawn at true scale: the track is the whole book, the marks are those clusters inside it. This is the comparison a heatmap never shows you, and it is why reaching a cluster moves price about a tenth as much as the move it arrived on.

estimated hyperliquid l2Book, nSigFigs=3, fetched live. Levels reach ±2.47%; a cluster beyond that is divided by the book at spot, because the book at its own price is not observable

Where the forced selling sits

cannot defend — zero free collateral can still move its liquidation price % of book = share of the $255.5M standing within ±1% of spot
85,887 +9.0% $274k 0.11% 4w
85,493 +8.5% $114k 0.04% 2w
85,099 +8.0% $36k 0.01% 1w
84,705 +7.5% $219k 0.09% 4w
84,311 +7.0% $116k 0.05% 2w
83,523 +6.0% $539k 0.21% 3w
83,129 +5.5% $1.0M 0.40% 2w
82,735 +5.0% $492k 0.19% 3w
82,341 +4.5% $1.4M 0.54% 4w
81,947 +4.0% $1.6M 0.62% 4w
81,553 +3.5% $354k 0.14% 3w
81,159 +3.0% $25.0M 9.8% 7w
80,765 +2.5% $5.8M 2.3% 3w
80,371 +2.0% $605k 0.24% 4w
79,583 +1.0% $407k 0.16% 2w
79,189 +0.5% $478k 0.19% 2w
BTC now 78,796
78,402 -0.5% $1.1M 0.44% 5w
78,008 -1.0% $110k 0.04% 3w
77,614 -1.5% $2.9M 1.1% 2w
77,220 -2.0% $301k 0.12% 1w
76,826 -2.5% $24.0M 9.4% 3w
76,038 -3.5% $106k 0.04% 2w
75,644 -4.0% $12.6M 4.9% 5w
74,068 -6.0% $537k 0.21% 1w
72,886 -7.5% $73k 0.03% 1w
72,492 -8.0% $990k 0.39% 2w
72,098 -8.5% $403k 0.16% 2w
71,310 -9.5% $57k 0.02% 1w

Above spot is where shorts are bought back; below is where longs are sold. Bar length is exposure; the solid portion is the part held by wallets that cannot move their own liquidation price. Bar length is relative to the largest cluster here — the % of book column is the absolute reading, and it is the one that decides whether any of this can move the market.

calculated 0.5% buckets from 78 positions from the hourly scan of the 300 largest. Being large correlates with being safe: this tier holds 28% of positions within 10% of liquidation where the full scan holds 47% (F-0018). Coverage and dollar totals are unaffected

What changed since the last scan

The largest position moves between our last two deep scans, matched by wallet. This reports what happened — it is not a signal, and it is not read as one.

WalletChangeSizeFree collateral
0x004edc… position closed $23.9M → $0
0x5b5d51… position increased $100.1M → $118.7M $12
0xfc2713… free collateral increased +$5.8M $7.0M
0xfc2713… position increased $60.8M → $71.4M $7.0M
0x09bc1c… position reduced $30.8M → $24.4M $4.5M
0x09bc1c… free collateral increased +$2.5M $4.5M
0xecb63c… free collateral decreased $-6,014,214 $19.8M
0xecb63c… position reduced $23.4M → $22.5M $19.8M

166 changes of $50k or more, 5.9h since the second scan · deep tier only, both sides — mixing in the hourly tier would count a wallet re-ranking out of the top 300 as a position change, which it is not (F-0011). How this is measured →

What we cannot see

  • 66% of open interest — wallets outside this scan
  • cross-margin effects from positions in other assets
  • anything smaller than $50k, or anything that changed since the second scan above — the feed is two snapshots apart, not continuous
  • the book beyond ±2.47% — a cluster further out is measured against the book at spot, which is a proxy for one we cannot see

How every number here is produced →