Selecting the best simulated design with the expected opportunity cost bound

In many applications, we wish to select the best simulated design among a set of design alternatives. While most work in the literature evaluates the quality of a selection procedure by the probability of correct selection of the best design, the expected opportunity cost of a potentially incorrect selection is also an important and useful measure, especially when the design performance reflects economic value. In this technical note, we consider the selection problem with an expected opportunity cost constraint. Based on some approximations, we derive for the selection problem an asymptotic closed-form allocation rule, which is easy to compute and implement. The numerical testing shows that the proposed selection procedure can enhance the simulation efficiency significantly.