The Regularization Aspect of Optimal-Robust Conditional Value-at-Risk Portfolios

In portfolio management, Robust Conditional Value - at - Risk (Robust CVaR) has been proposed to deal with structured uncertainty in the estimation of the assets probability distribution. Meanwhile, regularization in portfolio optimization has been investigated as a way to construct portfolios that show satisfactory out-ofsample performance under estimation error. In this paper, we prove that optimal- Robust CVaR portfolios possess the regularization property. Based on expected utility theory concepts, we explicitly derive the regularization scheme that these portfolios follow and its connection with the scenario set properties.