Methodology and risk scoring
Name matching
Names are normalised (accents, titles and legal forms removed), compared token by token with Jaro-Winkler similarity and phonetic keys (so Shehbaz, Shahbaz and Shabaz match), with low weight for common name parts. Matches of 92% or more are strong; 84–92% are possible and count half.
Risk score
| Factor | Points |
|---|---|
| Sanctions match (strong / possible) | 60 / 30 |
| Terrorist or proscribed list | 60 / 30 |
| Wanted (Interpol, FBI, NCA) | 50 / 25 |
| Debarment (World Bank) | 25 / 12 |
| Convicted / charged (Wikidata) | 35 / 20 |
| Politically exposed person | 25, tier at least High |
| Relative or close associate of a PEP | 15, tier at least Medium |
| Offshore leaks | 20 + 5 per extra record, max 30 |
| Network linked to sanctioned / convicted parties | 15 |
| Network linked to offshore leaks | 10 |
| Adverse media (Wikipedia, news) | 10–15 |
| Country risk (High / Medium) | 15 / 8 |
Tiers: 0–24 Low · 25–49 Medium · 50–74 High · 75+ Critical. PEP and RCA floors follow FATF Recommendation 12.
Limitations
Public sources are incomplete and can be wrong. A match is a lead, not proof. Verify identity before any decision.