A compromise solution to mutual funds portfolio selection with transaction costs (Q5952507)

Mean-variance analysis is a classical way in modern finance to describe and characterize how agents behave under market risk and uncertainty when choosing optimal portfolios that maximize their expected utility of wealth. This article considers the portfolio selection problem modified with some assumptions, based on empirical evidence, on the correlation coefficient structure of the returns of securities in the portfolio. In addition, the authors incorporate transaction costs that are supposed to be proportional to the absolute value of the exchanged assets. The portfolio selection problem is formulated as a bi-objective non-linear programming problem and a compromise programming method is proposed to find a solution. Finally, some numerical examples are provided to compare the compromise solution with the ones obtained by applying a genetic algorithm to maximize several direct utility functions.