The liquidation contagion map
When a major coin prints a top-decile liquidation cluster, the shock rarely stays put. This map is the measured anatomy of how those cascades have spread: every directed edge is source → responder, built from unique shock events in an append-only journal, carrying its sample size and a day-clustered 95% confidence interval. Descriptive history, bound to the corpus since 2026-05-20 — never a prediction, never a trade instruction.
13 cascade edges serving
430 unique shock events
corpus since 2026-05-20
shock floors: BTC ≥$4.0M · ETH ≥$3.0M liquidated in an hour
The edge map — tap any edge for its full stats
Tap or hover an edge — every stat carries its n and its 95% CI.
Edge width ∝ |shock-beta| · arrow points at the responder · blue nodes are shock sources · ETH appears twice by design — shock source (top) and responder (grid)
Every edge, with its sample size
| Edge | Shock-beta | 95% CI | Same-bar | Next-bar | n (shocks) | Days |
|---|---|---|---|---|---|---|
| BTC → SUI | 1.39 | [1.22, 1.55] | -0.21% | +0.10% | 192 | 67 |
| BTC → ADA | 1.29 | [1.11, 1.49] | -0.16% | -0.01% | 192 | 67 |
| BTC → LINK | 1.28 | [1.18, 1.37] | -0.15% | +0.07% | 192 | 67 |
| BTC → AVAX | 1.25 | [1.13, 1.38] | -0.17% | +0.07% | 192 | 67 |
| BTC → SOL | 1.25 | [1.15, 1.34] | -0.17% | +0.12% | 192 | 67 |
| BTC → ETH | 1.21 | [1.12, 1.32] | -0.14% | +0.04% | 192 | 67 |
| BTC → DOGE | 1.14 | [1.04, 1.25] | -0.19% | +0.08% | 192 | 67 |
| BTC → XRP | 1.06 | [0.96, 1.16] | -0.07% | +0.06% | 192 | 67 |
| ETH → LINK | 1.03 | [0.97, 1.08] | -0.07% | +0.10% | 238 | 73 |
| ETH → SOL | 0.99 | [0.92, 1.05] | -0.11% | +0.11% | 238 | 73 |
| ETH → OP | 0.97 | [0.70, 1.25] | -0.12% | +0.07% | 72 | 26 |
| ETH → ARB | 0.83 | [0.60, 1.10] | +0.04% | +0.09% | 69 | 25 |
| BTC → BNB | 0.83 | [0.75, 0.92] | -0.12% | +0.04% | 192 | 67 |
The honesty block — verbatim from the measurement artifact
Basis: historical-not-forecast · corpus since 2026-05-20 ·
430 unique shock events (one immutable journal row per
source×shock-hour, deduped, no look-ahead).
Disclaimer, verbatim: “Long-horizon liquidation-cascade contagion: for each source→responder pair, on hours where the source printed a large liquidation cluster (>= its USD floor) AND moved >= 0.4%, the responder's mean same-bar / next-bar move and the shock-beta (mean response/source elasticity) with a day-clustered 95% CI. DESCRIPTIVE — how assets HAVE co-moved on shock bars, bound to the corpus since 2026-05-20; not a forecast, not financial advice. A pair below n>=10 & day-clusters>=7 shows 'watching' with no beta.”
Not a signal. Nothing on this page has cleared the
promotion kill-gate (net_usd>0 ∧ ci95_low>0) that guards every
Hunter Killer strategy claim — this is market structure, published
because measured honesty is the product. It is not a forecast and not financial
advice.
Want the machine-readable feed? The same artifact serves
as JSON at /market/contagion/longrun for Pro subscribers, with the
trailing-window sibling at /market/contagion.
Measurement stamp — artifact generated 2026-08-16T22:09:04Z · basis historical-not-forecast · since 2026-05-20