基于文化算法的股票投资收益分析

发布时间：2018-03-06 04:31

本文选题：文化算法　切入点：微分进化算法　出处：《西安工程大学》2012年硕士论文　论文类型：学位论文

【摘要】：目前，由于科学技术的相互交叉和渗透，单纯的依靠一个领域的理论和方法已经不能解决一些课题了，因此研究者开始把一个领域的研究成果运用到另一个领域中，逐渐的多个领域交叉发展的结果形成了许多边缘科学。文化算法作为一种新的进化算法，它是通过对以往进化理论的优势与缺陷进行细致的分析后，同时参照社会进化理论与已有的科研成果而提出的。文化算法是一种基于知识的双层进化系统，包括两个进化空间：一个是类似于其它进化算法的种群空间，包含着进化过程中的具体个体，另一个是信度空间，它是由进化过程中获取的经验知识组成的。文化算法通过将种群知识与已有的一些智能算法有机的结合起来，可以提高算法整体的性能。文化算法本质上是一种优化算法，可以应用于机器学习、网络安全性、优化设计、自动控制、模式识别和经济学等广泛领域。本文主要研究文化算法在经济学投资组合理论中的应用。本文先对文化算法的发展及现代投资理论的发展历程进行了阐述，随后又进行了以下的研究：首先，详细叙述了文化算法的计算框架，以及框架中两个空间（信度空间和种群空间）和各个功能函数（接受函数、影响函数和更新函数）的设计。随后又简单介绍了微分进化算法，包括算法的基本原理，各参数的设计，以及该算法的优缺点。然后结合以上两种算法的优点，本文将微分进化算法嵌入到文化算法框架中，得出了文化微分进化算法。然后借助11个测试函数分别对文化微分进化算法及另外两种算法进行了测试，并得出了满意的结果。其次，简单介绍了传统投资组合理论的研究方法，分析了以往理论的不足之处，而后重点介绍了现代投资理论的标志—马柯维茨投资理论，详述了该理论的基本思想，分析了组合投资可以降低整体风险的原理。再次，本文创造性的把文化微分进化算法应用于投资组合问题中。主要以马柯维茨投资组合理论为研究对象，构造了有风险偏好的投资组合模型，得出了相应的非线性规划模型，进而构造出了求解该模型的文化微分进化算法。然后应用测试数据进行仿真实验，结果表明文化微分进化算法比文化算法能更有效且快速的解决此类问题，在此也证明了本文提出的文化微分进化算法的有效性。最后，，通过对深圳交易所所有A股在2010年度的交易状况的分析，再运用相关软件及编程对数据进行筛选及计算，最终选出比较符合投资组合模型要求的6只股票，并把这6只股票的相应数据代入有风险偏好的投资组合模型。然后分别用文化微分进化算法、文化算法和微分进化算法求解该模型。实验结果表明，文化微分进化算法较其它两种算法能以更快的速度收敛到全局最优解，并且具有良好的鲁棒性，这也表现出了该算法在实际应用中的高效性。然后借助得出的最优投资比例，引入实例来进行实际的投资，并与不采用马柯维茨组合理论的投资结果进行比较，结果不但表明马柯维茨组合理论在投资活动中的重要性，更证明了文化微分进化算法良好的寻优特性，因此把该算法应用到经济投资领域中，可以为投资者做出更快捷且准确的投资决策。
[Abstract]:At present, the intercross and penetration of science and technology, simply rely on the theory and method of a field has been unable to solve some problems, so the researchers began to research a field applied to another area, the results of a number of areas of cross development gradually formed a lot of edge culture science. Algorithm is a new evolutionary algorithm, which is based on detailed analysis of advantages and disadvantages of the previous evolutionary theory, social evolution theory and referring to the existing research results and put forward. The cultural algorithm is a kind of double evolution system based on knowledge, including two evolutionary space: a similar population space in other evolutionary algorithms, including specific individuals in the process of evolution, the other is the reliability of space, it is the experience of knowledge acquisition by the evolutionary process. The composition of culture algorithm group knowledge General and some combination of intelligent algorithm organically, can improve the overall performance of the algorithm. The culture algorithm is essentially a kind of optimization algorithm, can be used in machine learning, network security, optimization design, automatic control, pattern recognition and other fields of economics. This paper focuses on the application of portfolio theory in Economics culture algorithm in.
This paper first expounds the development of cultural algorithm and the development of modern investment theory, and then the following research is carried out.
First of all, described in detail the calculation of the cultural algorithm framework, and the framework of the two space (belief space and population space) and all function (acceptance function, influence function and update function) is designed. Then a simple differential evolution algorithm is introduced, including the basic principle of the algorithm, design parameters, advantages and disadvantages the advantages of the algorithm. Then combined with the above two algorithms, the differential evolution algorithm is embedded into the cultural algorithm framework, the cultural differential evolution algorithm. Then the cultural differential evolution algorithm and the other two algorithms were tested by 11 test functions, and obtained satisfactory results.
Secondly, this paper introduces the research methods of traditional portfolio theory, analyzes the shortcomings of previous theories, and then introduces the symbol of modern investment theory Markowitz investment theory, describes the basic ideas of the theory, analysis of portfolio investment can reduce the risk of the whole principle.
Again, this paper creatively put cultural differential evolution algorithm is applied to the portfolio problem. Mainly based on Markowitz portfolio theory as the research object, constructs the model of portfolio investment risk preference, the corresponding nonlinear programming model, and constructs the culture solution of differential evolution algorithm of the model. Then the application of test data the simulation experiment results show that, the cultural differential evolution algorithm can be more effective and fast algorithm to solve such problems than culture, this also proves the validity of the cultural differential evolution algorithm is proposed in this paper.
Finally, through the analysis of the Shenzhen stock exchange all A shares in the 2010 annual transaction status, and then use the relevant software and programming for data selection and calculation, and ultimately selected 6 stocks compared with the requirements of the portfolio model, and the corresponding data into the portfolio model of these 6 stocks are then risk appetite. With the cultural differential evolution algorithm, cultural algorithm and differential evolution algorithm to solve the model. The experimental results show that the cultural differential evolution algorithm is compared with the other two algorithms can converge to the global optimal solution speed faster, and has good robustness, which also show the efficiency of the algorithm in the practical application of the optimal. The proportion of investment and with the help of the introduction of examples to the actual investment, and using Markowitz portfolio theory and investment results were compared, results not only show that the combination of Ma Kewei Bates. The importance of theory in investment activities also proves the good characteristics of cultural differential evolution algorithm. Therefore, applying this algorithm to the field of economic investment, we can make faster and more accurate investment decisions for investors.

【学位授予单位】：西安工程大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F830.91;TP301.6

【参考文献】