Robust Stochastic Outperformance under Kullback-Leibler Ambiguity

We study the worst-case probability that $Y$ outperforms a benchmark $X$ when the law of $Y$ lies in a Kullback-Leibler neighbourhood of the benchmark. The max-min problem over couplings admits a tractable dual (via optimal transport), whose optimiser is an exponential tilt of the benchmark law. The resulting solution reduces to a one-parameter family indexed by regulizer $\lambda$, which controls the KL information budget and induces an increasing transfer of probability mass from lower to higher outcomes. The formulation can be evaluated from the baseline distribution and may serve as a distribution-wide scenario generation environment as well as a basis for robust performance assessment. Keywords: exponential tilting; optimal transport; stochastic dominance; stress testing.