A compromise solution to mutual funds portfolio selection with transaction costs

This paper considers the portfolio selection problem with transaction costs which are assumed to be a V-shaped function of the difference between an existing portfolio and a new one. Under some assumptions on the variance–covariance matrix of returns, we derive a compromise solution to this portfolio selection problem by solving a linear program. The compromise solution is then extended to include a riskless asset which allows short sale. We also compare the results of our compromise solution with results derived by the direct utility function method, which is solved by a specially designed genetic algorithm in this paper. Results show that the expected return and the standard deviation of the compromise solution are both smaller than those solutions from the direct utility function method.