An optimal opportunity cost selection procedure for a fixed number of designs

The expected opportunity cost is an important quality measure for the selection for the best simulated design among a set of design alternatives. It takes the case of incorrect selection into consideration and is particularly useful for risk-neutral decision makers. In this paper, we characterize the optimal selection rule which minimizes the expected opportunity cost by controlling the number of simulation replications allocated to each design. The observation noise of each design is allowed to have a general distribution. A comparison with other selection procedures in the numerical experiments shows the higher efficiency of the proposed method.