Conformalized Decision Risk Assessment

In many operational settings, decision-makers must commit to actions before uncertainty resolves, but existing optimization tools rarely quantify how consistently a chosen decision remains optimal across plausible futures. This paper introduces CREDO -- Conformalized Risk Estimation for Decision Optimization, a distribution-free framework that quantifies the probability that a prescribed decision remains (near-)optimal across realizations of uncertainty. CREDO reformulates decision risk through the inverse feasible region -- the set of outcomes under which a decision is optimal -- and estimates its probability using inner approximations constructed from conformal prediction balls generated by a conditional generative model. By calibrating each ball to lie entirely within the inverse feasible region, CREDO obtains finite-sample valid lower bounds on decision optimality without parametric assumptions. The method avoids the conservatism of worst-case robust optimization, is compatible with modern generative models, and applies broadly to convex optimization problems. We establish theoretical validity guarantees, develop efficient computational procedures, and demonstrate through extensive numerical experiments that CREDO provides accurate, interpretable, and reliable assessments of decision reliability in both synthetic and application-motivated settings.