Minimax Expected Opportunity Loss: A New Criterion for Risk-Based Decision Making

Abstract

A risk measure, expected opportunity loss (EOL), is introduced to quantify the potential loss of making an incorrect choice in risk-based decision making. Different from Savage's (1951) minimax regret principle, EOL can account for the unbounded continuous random outcomes of alternatives and decision makers’ acceptable risk. This article studies the effects of the forms of loss function, correlation among outcomes, and the acceptable risk on the ranking results by considering the loss function in the power form. The results show that the loss functions and the outcomes correlations can significantly influence the rankings of alternatives in risk-based decision making.