Research methods
Wallet clusters: confidence is not ownership
Shared timing and funding can support a lead without proving a shared operator.
Os artigos de investigação e as referências são publicados em inglês. A navegação está disponível em sete idiomas.
Neste artigo
- Name the observed relationship Was there a direct transfer, a shared funding source, repeated counterparties or similar timing? State the actual observation before naming the cluster. A large exchange withdrawal address can connect unrelated users, so shared infrastructure needs particular care.
- Look for independent support A stronger attribution combines multiple observations that do not merely repeat the same underlying signal. Public statements, repeated operational patterns and direct control evidence can change confidence, but each source has limitations. Record contradictory observations rather than silently discarding them.
- Publish the confidence boundary Use language that distinguishes a confirmed transaction relationship from a likely common operator and an unverified identity. Include alternative explanations and the period covered. Do not turn a graph's visual proximity into an allegation. A useful wallet map helps readers follow the evidence; it should not hide the assumptions behind attractive lines between addresses.
A cluster is an analytical grouping. It does not automatically identify a person or prove coordinated intent. This research workflow separates observed relationships from the attribution a researcher may infer from them.
Name the observed relationship Was there a direct transfer, a shared funding source, repeated counterparties or similar timing? State the actual observation before naming the cluster. A large exchange withdrawal address can connect unrelated users, so shared infrastructure needs particular care.
Look for independent support A stronger attribution combines multiple observations that do not merely repeat the same underlying signal. Public statements, repeated operational patterns and direct control evidence can change confidence, but each source has limitations. Record contradictory observations rather than silently discarding them.
Publish the confidence boundary Use language that distinguishes a confirmed transaction relationship from a likely common operator and an unverified identity. Include alternative explanations and the period covered. Do not turn a graph's visual proximity into an allegation. A useful wallet map helps readers follow the evidence; it should not hide the assumptions behind attractive lines between addresses.
Fontes
Ethereum documentation · accountsTechnical reference checked 9 September 2026. The review questions are editorial analysis, not findings about a named project.