Crowding

Nominally independent vaults are often in the same trade. When the market moves against them they do not leave one by one — they leave together, through the same liquidity, turning a controlled loss into a cascade. This page measures how much of that is happening, and is honest about how much of it we can see. Click an instrument to see exactly which vaults hold it.

Vaults with live positions

352

Hyperliquid, currently observable

Gross exposure observed

$191.15M

longs and shorts, not netted

Long: $101.86M
Short: $89.29M
Net: +$12.56M

Instruments

177

Priced: 177
One-sided: 28

Flagged

≥5% of open interest and ≥80% one-sided

Exit over a day: 0
Exit within a day: 0

Holding more than a quarter of a day of the asset's own trading, across the 177 instruments we could price. This is the list that traps somebody.

MEME

one side: 67%
of OI: —
to exit: 0.8d
10 vaults: $155.3k

These are different lists on purpose. Coverage measures the share of the contract; exit time measures whether the asset can absorb the position leaving. A crowd can be most of a contract and still walk out in an hour.

Does being in a crowd cost anything?

121 events across 112 vaults

Vaults entering a crowd went on to underperform matched vaults that did not by -135281.76% at the worst point of the following 20 days. Shuffling the event dates produces a path that extreme 26% of the time, so this is not yet distinguishable from noise.

Day	Entered a crowd	Matched controls	Difference	Events
-5	1347.03%	887.77%	459.26%	119
-4	1071.39%	205.99%	865.40%	119
-3	450.37%	71.20%	379.17%	119
-2	292.07%	28.83%	263.24%	119
-1	36.11%	6.62%	29.49%	119
+1	62.24%	0.66%	61.58%	119
+2	3663.23%	12.43%	3650.79%	119

What this cannot tell you

We see on-chain and venue-side only. Anything held on a centralised exchange is invisible, so true crowding is at least as bad as shown, never better.
Coverage is usually low. Most instruments here show us holding a low single-digit share of open interest. That is a statement about our reach, not about safety.
Book clusters use single-linkage grouping. Members are chained by pairwise similarity, so a cluster average can sit below the joining threshold. It is transparent and arguable, which we prefer to a clever method nobody can interrogate.
Positions are snapshots. A strategy that flips intraday can look uncrowded at every observation and still be crowded in between.