The Portfolio Research Based On The Multi-index Sorting Black-Litterman Model

Domestic investors pay more attention to asset allocation with the rapid development of China’s financial industry. Research shows that reasonable asset allocation can be good to reduce the risk. It is necessary to study how to allocate asset in complex economic form. In general mean-variance model that proposed by Markowitz in1995was the beginning of modern investment theory. This model was first to use quantified idea to asset allocation. Then a large number of scholars take in-depth study on this issue. The first part introduces the extended mean-variance model based on statistical estimation and risk control requirements.In recent years, A lot of practice prove that subjective view of investors also play a crucial role in asset allocation. For this reason, Black and Litterman propose a model base on the capital asset pricing model and the inverse optimization theory in1992. Black-Litterman model embed the investor’s subjective view into traditional mean-variance model. The second part of this paper describes the Black-Litterman model and weaknesses, and gives the improvement and expansion of the model by foreign scholars.The view is difficult to quantify in the Black-Litterman, this problems have restricted their extensive use. For which many scholars take extends research on this problem. The third part of this article will propose a new ways to solve this issue. We improve and extend the Black-Litterman model on two aspects. On one hand, we get the evaluation of asset through multiple indicators of the asset, then, take some sorting methods to obtain quantitative view according to the ranking information of asset. On the other hand, we discuss some algorithms at the stage of combinatorial optimization. The traditional approach are based on Markowitz’s optimization algorithm, this paper this paper discusses the stochastic optimization algorithm proposed by Tze Leung Lai.This paper selects ten stocks in Shanghai Stock Exchange Market as the freely configured assets. Obtain matrix view through multiple index ranking method and use GARCH model to predict parameters of view. After that we can gain the weight of the ten stocks via stochastic optimization method. Finally, the weight of multi-phase stock portfolio can be computed by using the rolling sliding window method and weight retention ten days. After that, we can compare our model with other models. The empirical results show that this model has many advantages compared with the other model, can provide a new idea to research the problem of portfolio.