Based On The Combination Of Multicriteria Decision Optimization Of Extending The Application Of Black - Litterman Model

Markowitz proposed a theory of portfolio selection in1952, based on the mean-variance model. Then the research of asset allocation went into the era of quantitative investment. However, this model can not blend to the investors’ subjective views well. In1992, the known Black-Litterman asset allocation model was created. It is based on the assumption of the semi-strong efficient market, combined with the capital asset pricing model (CAPM) and Bayesian rules. The BL model can easily interpret the investors’subjective views. In this paper, we used the market variables modified Black-Litterman model (BLm model) for the posterior distribution of excess return of A-share industrial indices.Existing research on the Black-Litterman model often uses Markowitz’s critical line algorithm in the optimization. This method assumes that all investors have the same preference when the market is equilibrium, and can not consider individual preferences. In order to deal with individual preferences, this paper uses multiple criteria decision making (MCDM) method for portfolio optimization.In this paper, we use the weighting method to translate multi-objective optimization problem into single-objective optimization problem. Firstly, we use interpolation to translate investors’ preferences for each goal into specific utility functions. Then, different utility functions were integrated into a total utility function, also as the objective function of optimization. We used genetic algorithm for optimization, which help getting ideal solutions in the NP-hard situation and the complex search space (non-convex and non-concave).Additionally, we verified that the robustness of the Black-Litterman model is extremely limited in the MCDM portfolio optimization. To enhance the robustness of the asset allocation, we need to rely on adding constraints, and the use of appropriate heuristic algorithm helps a lot.