The Research On Portfolio Optimization Model And Strategy For Multi-objective

This paper proposed an innovative solution for the portfolio multiple-objective optimization. In chapter 2, an improved multiple-objective algorism based on target space partitioning was proposed. In chapter 3, the algorism was further improved and optimized; its intelligentization and innovation were also discussed and studied preliminarily. In chapter 4, a multiple-objective mathematical model was proposed and a series of derivative study were conducted. In chapter 5, an innovative evaluation method for multiple-objective optimization and an algorism-testing function were proposed; they were applied to evaluate the algorism proposed in this paper. Since the algorism was designed specifically to solve portfolio optimization problems, it showed great superiority in specific applications of portfolio decision-making and planning. Considering the requirements of portfolio for risk diversification, the stability and distribution evenness were highly stressed during the designing of the algorism and testing. Considering the investors or decision-makers might want to control and interfere with the decision-making process, the visual operation is realized through computer computation; this is also one important innovative point of this paper. Application approaches and examples were given in each chapter; the experimental results proved that this algorism could improve the efficiency of processing multiple-objective portfolio optimization, improve the way of decision-making and strengthen its scientificalness. It would be excellent if more great ideas could be generated in future study after this paper. The content of this paper is comprised of following parts:1.This paper proposed a multiple-objective method based on target space partitioning. The algorism also considered the evolution and preference, improved the target space partitioning method, added the secondary partitioning, introduced the new sorting rule and showed the corresponding mathematical analysis equation, detailed algorism steps and program expression. Through computer the algorism could generate partition picture automatically to achieve the visualization of the final solution; the visualization also guaranteed the maintainability, expansibility and scalability of the algorism. In terms of application, this paper showed the algorism's projection application relations in personal financial management portfolio and gave investors the mechanism for reference.2.On the basis of former designed multiple-objective method based on target space partitioning, the algorism's performance was improved and optimized. The tolerance concept was introduced to improve the algorism's overall optimization capability; the double-population saving mechanism was introduced to guide the algorism to approach the Pareto optimal region more rapidly and to save candidate sets for different requirements of other portfolios. The algorism's superiority in maintaining the solutions distribution evenly and convergence is also proved through comparison with former classical algorism and a large number of experiments.3.The classical algorism had too strict requirements for target functions and variables, therefore the good performance could only be obtained with specified constrains were set. To solve this problem, the economic model and multiple-objective algorism were established through mathematical methods and genetic algorism; the portfolio risks were also studied under uncertain investment circumstances. This algorism has much stronger adaptability than classical ones in portfolio area. It also avoided the generating of pseudo-effective solutions through strategy of lower layer results feedback to the upper layer.4.To solve the conflict and mutual influence among objectives in the portfolio, this paper proposed an innovative method of multiple-hierarchy objective optimization by combining gaming theory and analytic hierarchy process. Considering the special circumstances of China and the specific application of portfolio, the rigid decision-making was turned into flexible decision-making; and the objective judgments of gaming theory together with the subjective judgments of analytic hierarchy process were combined. Therefore the deviation of results is reduced and the scientificalness of decision-making is increased.5.This paper proposed an innovative method of portfolio fuzzy testing to solve the disequilibrium of multiple-objective evaluation system. Based on the multiple-objective gaming evaluation function, the fuzzy math evaluation is introduced to quantify the indicators according to the membership principle and combine the objective and subjective evaluation. Through the expert system, people's experience is fully utilized; and through the mutual conversion of qualitative factors and quantitative factors, all information is fully applied. With the current highly-concerned portfolio problem, this paper also showed the reference value this method can provide to deal with complex portfolio problems.6.In the end of this paper, a new testing function and the performance evaluation indicators were designed to test the former algorism and to concatenate the overall study of multiple-objective optimization method design.