An amendatory branch and bound algorithm for mad model with concave transaction costs and bounded assets constraints

This paper is concerned with a portfolio optimization problem under transaction costs and bounded constraints. The mean-absolute deviation (MAD) portfolio optimization model can be transformed into a set of linear programming by using epsi-deviation piecewise linear functions to estimate transaction cost functions. An amendatory branch and bound algorithm is proposed to obtain epsi-deviation approximate efficient portfolio. In order to compare the algorithm proposed by Konno and Wijayanayake with the amendatory algorithm, a computational experiment from the real stock data in the Shanghai Stock Exchange is offered. The empirical results show that the amendatory algorithm needs less calculation than the previous algorithm, while getting the same optimal portfolio. Copyright © 2010 Watam Press.