A score you can argue with.
Every risk score decomposes into the signals that produced it, with weights, evidence and a model version — because a number nobody can interrogate is not analysis, it is decoration.
How is a crypto address risk score calculated?
A crypto address risk score combines direct and indirect exposure to categorised counterparties, weighted by how many hops separate them and how much value moved. A score is an analytical indicator, not a determination of guilt, and it should prompt a review rather than substitute for one.
Six signals, one score.
Switch address to see the same model produce a very different picture.
What each signal measures.
| Signal | Weight range | What it measures |
|---|---|---|
| Transaction exposure | 0–26 | Share of value with a path to a category flagged high risk, weighted by hop distance. |
| Known risk exposure | 0–22 | Direct interaction with addresses in a maintained risk set, with the set and version recorded. |
| Counterparty risk | 0–16 | Risk carried by the addresses this address transacts with most. |
| Behavioural signals | 0–12 | Patterns such as rapid fan-out, peel chains, round-value structuring or dormancy breaks. |
| Entity exposure | 0–10 | Exposure to attributed services whose category carries elevated risk. |
| Sanctions signal | 0–6 | A match against a published designation in the exposure path. |
The weights above are the platform default. An organisation with a different risk appetite can adjust them under a platform agreement — and every score records the model version used, so a change of policy never silently rewrites past findings.
Score to level, and what a level should mean to you.
| Level | Score | Suggested handling |
|---|---|---|
| Low | 0–39 | Nothing observed that warrants special handling. Ordinary monitoring continues. |
| Elevated | 40–74 | Something warrants a look. Appropriate for review, not for automatic refusal. |
| High | 75–100 | Material exposure observed. Appropriate for escalation under your own policy. |
Your obligations are set by your regulator, your policy and your risk appetite — not by this table. The platform provides the measurement; the decision remains yours.
What a risk score is not.
A risk score is an analytical indicator derived from observable on-chain activity. It does not independently establish unlawful conduct, and attribution is asserted only where a source, evidence trail and confidence level exist.
A risk model that cannot be inspected cannot be challenged, and a model that cannot be challenged has no place in a decision that affects somebody's account or their case. Publishing the decomposition is the point.