[en]This paper constructs a robust optimization framework of the uncertain worst-case return. The model defines an adjustable discrete uncertainty set which controls the conservatism of the optimal asset allocation. Without prior assumptions on the data generating process, the model also develops an a priori probabilistic guarantee of the robust solution. Unlike previous measures that depend solely on the uncertainty model, the new measure is also sensitive to asset allocation and investment horizon. We provide an application of international stock indexes portfolio protection during the 2008 financial crisis. Computational experiments and ex-post analysis provide evidence for the effectiveness of our model.[/en]