Multi-period mean variance portfolio selection under incomplete information.

Zhang, Ling; Li, Zhongfei; Xu, Yunhui; Li, Yongwu

This paper solves an optimal portfolio selection problem in the discrete-time setting where the states of the financial market cannot be completely observed, which breaks the common assumption that the states of the financial market are fully observable. The dynamics of the unobservable market state is formulated by a hidden Markov chain, and the return of the risky asset is modulated by the unobservable market state. Based on the observed information up to the decision moment, an investor wants to find the optimal multi-period investment strategy to maximize the mean-variance utility of the terminal wealth. By adopting a sufficient statistic, the portfolio optimization problem with incompletely observable information is converted into the one with completely observable information. The optimal investment strategy is derived by using the dynamic programming approach and the embedding technique, and the efficient frontier is also presented. Compared with the case when the market state can be completely observed, we find that the unobservable market state does decrease the investment value on the risky asset in average. Finally, numerical results illustrate the impact of the unobservable market state on the efficient frontier, the optimal investment strategy and the Sharpe ratio. Copyright © 2016 John Wiley & Sons, Ltd.

FINANCIAL markets; DISCRETE-time systems; HIDDEN Markov models; MATHEMATICAL optimization; MARKOV processes

Applied Stochastic Models in Business & Industry, 2016, Vol 32, Issue 6, p753

1524-1904

Academic Journal

10.1002/asmb.2191