Complex selection processes: Dealing with dependencies

Abstract

Deciding which alternative is the best out of a set of alternatives, where each has multiple attributes, can be a complicated task. In both practice and literature we find that the majority of decision making methods ignore the inherent dependencies that can exist between attributes. Doing so simplifies the task at hand but at the expense of reliability and effectiveness of the entire process. If we are to trust automated trading systems, we must have full trust in the ability of the software to understand and utilize our preferences. This paper presents a method for graphically structuring dependencies among attributes and defining the resulting utility function. An example of fund selection is then given, showing how the inclusion of such dependencies is essential in performing effective decision making.