A robo-advisor system: Von Neumann–Morgenstern expected utility modelling via pairwise comparisons

Abstract We introduce a robo-advisor system that recommends personalized investment portfolios to users, using a Von Neumann–Morgenstern expected utility model elicited from pairwise comparison data. The robo-advisor system comprises three fundamental components. First, we employ a static preference questionnaire approach to generate pairwise lotteries comparison questions. Next, we design three optimization-based preference elicitation approaches to estimate the nominal utility function pessimistically, optimistically and neutrally. In the preference elicitation process, we assume that the user’s pairwise choices are exactly consistent with her/his true utility, without any response error. Finally, we compute portfolios based on the nominal utility using an expected utility maximization model. We conduct a series of numerical tests on a simulated user and some human participants to evaluate the efficiency of the proposed robo-advisor system.