portfolio optimization interdiction: a novel risk-averse vision for risk management

Portfolio optimization is a crucial problem in financial management. since there are several types of risks, decision makers cannot manage their portfolios effectively based solely on historical data. on the other hand, the use of predicted data contains high errors which limits its practical applicability. the most well-known optimization model is markowitz’s model, which minimizes risk level through a quadratic objective function while maintaining a fixed level of profit. however, this approach cannot adequately manage risks arising from different unexpected events. this paper proposes a bilevel optimization model from the perspective of a riskaverse decision maker. in the first level, different events are modeled under various scenarios. then, the decision maker operates at the second level, observing these events and determining how to set their portfolio. we utilize concepts from interdiction problems to convert the bilevel problem into a single-level optimization problem that can be solved using optimization solvers such as gurobi and cplex.