Risk Engine
Configurable weighted aggregation and hard rules convert module signals into Low / Medium / High risk while preserving module-level reason codes and human-review overrides.
Module weights
Illustrative only; production weights require validation, governance and approval.
Aggregated decision
Weighted score is only one part of the decision. Hard rules, data-quality gates and human-review rules can override the numeric class.
Human-accountability gate
High-impact adverse-media / criminal / sanctions ambiguity cannot trigger final decline without review and documented rationale.
Low Risk
All mandatory checks pass, no material finding, data quality sufficient.
Medium Risk
One or more review thresholds, ambiguities or evidence-quality exceptions.
High Risk
Confirmed prohibited business, high-confidence sanctions concern or material risk rule.
| Rule ID | Rule | Type | Threshold | Action | Human gate |
|---|---|---|---|---|---|
| R-WEB-001 | Confirmed prohibited website category | Hard rule | Confidence ≥ 0.90 + evidence | High risk | Yes before decline |
| R-ADV-004 | Adverse-media candidate | Review rule | Entity 0.60–0.89 | EDD | Mandatory |
| R-DOC-003 | Cross-document identity mismatch | Data-quality rule | Material field mismatch | EDD / request clarification | Mandatory |
| R-IMG-002 | Unsafe uploaded image | Intake rule | Category threshold exceeded | Reject upload, not merchant | No merchant-risk inference |
Interactive Rule Studio available
Configure hard rules, review rules and data-quality gates, then run what-if simulations before promotion.
Open Rule Studio →