On a local-search heuristic for a class of tracking error minimization problems in portfolio management

In this paper we describe a 2-phase simulated annealing heuristic approach for a special class of portfolio management problems: the problem of optimizing a stock fund with respect to tracking error and transaction costs over time subject to a set of complex constraints with a linear factor return model “feeding” the objective function with data. Our results on managing two real-world funds of a major German capital investment company have shown that this meta-heuristic provides proposals for the fund manager which are feasible with respect to the investment guidelines and excellent in quality in acceptable time. Thus the approach is ideally suited to be used routinely and interactively within a decision support system to assist the fund manager in his complex task of portfolio control and optimization.