Dynamic portfolio optimization under multi-factor model in stochastic markets

摘要

We consider the modeling and solution of the multi-period portfolio selection problem in stochastic markets with bankruptcy risk control. This research differs from current results in the following ways: rather than in terms of return moments, the stochastic evolution of the market is directly described in terms of investment returns by a finite-state Markov chain; the multi-factor model is initially introduced in the modeling process to better control bankruptcy risk and to cope with large-scale portfolio selection problems; the stable distributions are adopted to describe factors' fluctuations to properly reflect the return distribution characteristics of risky assets; the bankruptcy risk in each period is flexibly controlled by utilizing the properties of the multi-factor model and restricting the portfolio loss caused by each factor; a specific bi-level programming method is proposed to find the analytical investment strategy; the practical significance and good performance of the stage-wise investment decision, when not optimal, are verified. Empirical results are finally provided to illustrate the suitability and practical performance of the new model and the derived explicit investment strategy.