Robust investment policies with bound forecasts

We present a continuous minimax model for robust portfolio optimization based on worst-case analysis. The classical Markowitz framework is extended to continuous minimax with upper and lower bounds on the return scenarios and a discrete number of rival risk scenarios. The model integrates benchmark relative computations in view of scalable (not fixed) transaction costs. It evaluates worst-case optimal strategies in view of upper and lower bounds on forecast return and a discrete set of risk scenarios. Robustness arises from the non-inferiority of the min-max strategy. The robust optimal policies are obtained simultaneously with the worst-case scenario. We apply the model to a selection of investment problem and evaluate the ex-ante performance of the strategy using historical data.