Attribution Confidence

Attribution Confidence

Attribution confidence is a measure of how certain a risk-screening tool is that a given address label — an exchange, a mixer, a scam operation — is actually correct, rather than a guess based on incomplete or circumstantial evidence. It is what separates a confirmed finding from a possible one.

What it means

Some attributions are close to certain: a sanctions list published by a government agency names a specific address explicitly, leaving essentially no ambiguity. Others are inferred through behavioral clustering or heuristic pattern-matching — an address that behaves like an exchange hot wallet probably is one, but without a direct confirmation, that attribution carries meaningfully lower confidence.

Presenting confidence honestly matters because treating every label as equally certain leads to two opposite failure modes: over-trusting a shaky inference causes false positives that erode a tool's credibility, while suppressing lower-confidence findings entirely hides genuinely useful early signal from a user who understands how to weigh it appropriately.

Real-world example

A government-published sanctions list entry for an address carries very high attribution confidence, while an address flagged as "likely" belonging to a mixing service purely because of its transaction pattern, with no direct confirmation, carries meaningfully lower confidence, and a scan result should communicate that difference.

Related terms

• Suspicious Service
• Provenance
• Risk Score

In FreezeRadar

FreezeRadar's scoring methodology page documents how attribution confidence factors into a wallet's overall risk score, so a low-confidence signal never carries the same weight as a confirmed one.

Check a wallet now