The number everyone would put on the front page turns out to be a constant
433 published clusters over three days of Hyperliquid BTC. 71% sit at exactly 100% cannot-defend. Splitting at the median leaves 29% below and nothing above, because the median is the ceiling.
We built a product around a measurement nobody else publishes: how much forced exposure near liquidation is held by wallets that have no free collateral left and therefore cannot move their own liquidation price.
Then we set out to calibrate it — to test whether clusters we called more vulnerable were subsequently reduced more often. The contract for that test required a contrast check to run first, before any outcome was examined, with the power to stop the experiment.
It stopped the experiment
| clusters recorded | 433 |
| median cannot-defend | 100.0% |
| at exactly 100% | 71% |
| below the median | 29% |
| above the median | 0% |
Nothing sits above the median because the median is the ceiling. There is no high-vulnerability group to compare a low-vulnerability one against, so the question is unanswerable — not because the effect is absent, but because there is no variation to measure it with.
The obvious fallback, the share whose free collateral is merely thin rather than zero, fails identically: median 100.0, nothing above it.
What this means, and what it does not
The measurement is still correct. Those wallets genuinely cannot defend their positions, and the arithmetic most tools use to work this out still misclassifies one wallet in five.
But it cannot rank anything. A quantity that is 100 for almost every cluster cannot say which cluster is more exposed than another — which is exactly what the word "vulnerable" implies to a reader.
And it describes leverage, not today. A near-constant presented as a live figure invites you to read meaning into its level. There is none there.
The honest version of the claim
Almost all forced exposure near liquidation is undefendable, almost all of the time. That is a finding about how leveraged crypto markets are actually structured, and it is not a signal about what happens next.
It is also not something anyone else could tell you, because no other liquidation product measures defensibility at all — which means none of them has had occasion to discover that theirs would not vary either.
What would change it
A market in which traders hold real free collateral against near-liquidation positions would restore the spread and make the calibration answerable. The record keeps accruing for exactly that reason: it costs nothing to keep, and the question goes live the moment the market provides a contrast.