False Positive Rate

A false positive rate measures how often a detection or matching system raises an alarm on something that, on closer inspection, wasn't actually a match or a risk at all. It's one half of a tradeoff every screening system has to make: tune a system to catch everything real (minimize false negatives, missed real matches) and it will also flag a lot of innocent noise (raise false positives); tune it to be conservative and quiet, and it will let real matches slip through unflagged. There's no setting that eliminates both error types at once, every system sits somewhere on that tradeoff curve, and the honest question to ask any vendor is where.

This matters most acutely in Sanctions Screening, where the industry widely acknowledges that a large majority of automated matches, commonly cited figures run well above 90% for unrefined systems, turn out, on manual review, not to be the sanctioned or politically-exposed person at all. That happens because names alone are a weak identifier: common names, transliteration variants, and incomplete source data mean a system tuned to catch every real match inevitably flags a flood of unrelated people who happen to share a name. That flood is precisely what fills a compliance team's Manual Review Queue and drives review headcount that scales with customer volume, not with actual risk.

Who actually built this

The underlying statistical concept, Type I error, predates any of this industry and belongs to general statistics, not to identity verification or compliance specifically. Biometric matching gave it a formal, standardized name and measurement methodology in ISO/IEC 19795, maintained by the international biometrics standards community. Applying the same idea to sanctions and watchlist name-matching is industry practice developed by compliance-technology vendors, Refinitiv, ComplyAdvantage, and NICE Actimize among the most cited, refining fuzzy-matching algorithms specifically to push this number down without pushing false negatives up. None of it is Solidus's design.

Solidus today

Solidus has no sanctions-screening false-positive rate to report, because Solidus does not run sanctions screening. For the identity-verification pipeline Solidus does run, document and face matching, liveness, no independent lab has measured a false-accept or false-match rate, and Solidus does not publish one, because doing so without real independent ground truth would be inventing a number rather than reporting one.

See also

Sanctions Screening is where false-positive rate is most consequential in KYC operations. Liveness Detection and Presentation Attack are where the biometric version of this measurement would apply to Solidus's own pipeline, if it existed yet. Pass Rate is the related, but distinct, operational metric, how many attempts clear the pipeline, not how many of the flags a screening step raises turn out to be wrong.