Portfolio Selection with minimum transaction lots: an approach with dual expected utility

In this paper we analyse the portfolio selectionproblem with minimum transactionlots in the context of non-expected utility theory. We assume that the decisionmaker ranks the alternatives by using a specific DualExpectedUtility. This functionallows portfolio values less or equal a fixed benchmark tobe weighted inadifferent way from values greater than the fixedbenchmark. Under normallydistributedreturns and opportunechoice ofthe benchmark, the suggested approach leads to an NP-complete problemandhas the advantage ofusing mixed linear programming to obtainthe optimal portfolio. We also show resultsobtained by implementing the model on the Italian stock market. (keywords: dual expectedutility, portfolio selection, NP-completeness, linear programming with mixed variables)