Screening methodology

The steps

Resolve the address format and select the network, or probe the plausible networks in turn.
Retrieve transfers and counterparties from the indexer for that network.
Attribute each counterparty: match against sanctions address sets, against the interim MiCA register, and against known service labels.
Categorise the flow into five exposure classes and normalise to percentages.
Weight the classes, apply the sanctions and register adjustments, and clamp to 0–100.
Derive the five Article 20 rating factors from the same signals.

Exposure categories and weights

A euro that reaches a sanctioned entity moves the score far more than a euro settled with an authorised provider, which pulls the score down. The weights are the whole model, so they are stated here in full.

Category	Weight
Sanctioned entity	+100
Mixer or privacy service	+72
Unregistered exchange	+46
Unattributed	+22
Authorised CASP	−12

The weighted average across categories is then adjusted: a listed counterparty adds 24 points, an absent MiCA authorisation adds 8, and proximity to a flagged service adds up to 18 points depending on the shortest hop distance. A direct designation on any consulted list short-circuits the calculation and returns 100.

Addresses with fewer than eight observed transfers are pulled toward the middle of the range, because a two-transaction history does not support a confident verdict in either direction.

Banding

Band	Score
High	70 and above
Medium	40 to 69
Low	below 40

Article 20 rating factors

Regulation (EU) 2024/1624 asks firms to rate a customer on geography, product, delivery channel, transaction pattern and customer type. Each factor is derived from a distinct signal so that the rationale printed beside it is defensible on its own.

Factor	Derived from
Geographic risk	Share of flow touching providers without an EU authorisation
Product risk	Share of flow through mixers, privacy services and sanctioned entities
Delivery channel	Shortest hop distance to a flagged service
Transaction pattern	Overall score and the observed transfer count
Customer type	Share settled with authorised CASPs, and register match

Limits

Attribution depends on the labels the indexer publishes. An unlabelled counterparty is counted as unattributed, not as clean.
The keyless data tier caps how many transfers can be retrieved per address, so very active addresses are sampled rather than read in full.
Sanctions lists carry addresses only where the designating authority published them. Absence from a list is not evidence of legitimacy.
Cross-chain flow through bridges is resolved only where the bridge counterparty is itself attributable.
When an upstream source is unreachable the report is marked degraded and the affected findings are shown as unavailable rather than as a negative result. See data sources.

Reproducibility

Scoring is a pure function of the retrieved data. The same inputs always produce the same score, and each report records which source answered, when, and with which list version, so a result can be reproduced or challenged after the fact.