An effective challenge of a model carried out by qualified personnel who are organisationally separate from the model developers and users, covering conceptual soundness, ongoing monitoring, and outcomes analysis.
In practiceUnder SR 11-7 and equivalent supervisory expectations, independence is a function of incentives, reporting lines, and authority, not merely organisational charts. Validators must have the standing to demand changes, withhold approval, or escalate to the model risk committee. The work covers data quality, assumptions, code review, benchmarking against alternative methods, sensitivity analysis, and a written validation report with findings tracked to closure. Findings are rated and feed into the firm-wide model risk appetite.
A retail bank's validation team rejects a proposed deep-learning collections model on the basis that benchmark testing against the incumbent gradient-boosted model shows no lift after controlling for sample drift, and requires the developer to rerun on a stratified out-of-time sample.
This definition is maintained by Moweb partners and used in live client engagements. For how Independent validation applies to your estate, or to challenge a working definition, speak to a partner.