Portfolio management under stable distribution hypothesis

Many evidences have come forth and suggested that empirical probability distributions of returns on securities conform better to Levy's stable distribution with infinite variance than to the normal distribution. In this paper, we develop a portfolio analysis framework for a stable market, and establish a multi-period planning model for portfolio management which considers taxation, contact cost, and constraints that should be considered by an institutional investor. To treat this large scale programming problem which involves many securities, multiple periods and nonlinearity, we give a decomposition-coordination algorithm for its solution. A numeric example shows the performance of the model and the algorithm.