Bayesian

Multiple adverse factors, or a hard ship-specific signal, lift this hull above the fleet norm.

A coverage-weighted blend of the single component we could read for this hull; the weights renormalise over only the components present, so a thin read is never inflated and a hull is never credited a “safe 0” for a signal it has no row for. This headline is flagged low-confidence (a thin or structural-only read) and should not be treated as a verdict. Higher means riskier. Derived in-house from government-open port-State-control, flag, sanctions and our own vessel data; weight it by the coverage above.