基于随机矩阵和复杂网络理论的股票聚类分析-理论物理专业论文.docxVIP

万方数据 11 AbstractCorrelation Abstract Correlation analysis of economic and financial time series has been importantly and widely used in the field of economics and finance．Particularly in the financial sector,correlation among fmancial assets return rate is the core concept in modern portfolio theory．But in practice，the correlation information getting from the real data, for the most part，are noise．Therefore，how to separate the valuable correlation information from noise to be used and analyzed，such as in stocks of clustering analysis， is an interesting question． This article selects the stocks of CSI-1 00 as research object to study the correlations between different stocks’return ratio，and obtain the specific clustering information．Firstly,we extract principal information from the correlation matrix of return ratio and have a roughly grouping the stocks by the RMT．Then，through CNT based on the prior probability of maximum likelihood method，we construct corresponding reliability index to quantitatively optimize the grouping results． Eventually,The grouping results well accord with the classification of stock industry,at the same time，we also accurately shows the associated strength of stocks in different industry categories and the correlation between industry groups．Our study provides programming trading based on the portfolio theory with the basis of quantitative analysis． Keywords：Stock return rate；Cross—correlation matrix；RMT；CNT；Clustering analysis 万方数据 万方数据 IV 目录第一章绪论 目录 第一章绪论 1 1．1复杂系统简介 。1 1．2经济物理学简介 2 1．2．1经济学和物理学的关系 一2 1．2．2经济物理学的历史 ．．4 1．2．3经济物理学的定义 ．．5 1．2．4经济物理学的研究内容和方法 ．．7 1．3选题的意义 。9 第二章理论基础 ．11 2．1 金融市场及有效市场假说简介 ．11 2．1．1金融市场简介 11 2．1．2有效市场假说简介 12 2．2 金融时间序列的涨落和关联 13 2．2．1金融时间序列的涨落 13 2．2．2金融时间序列的自关联 1 5 2．2．3金融时间序列的互关联 15 2．2．4金融时间序列的关联性的应用 16 2．3 概率统计相关知识 18 2．3．1频率论和主观概率论 18 2．3．2条件概率与贝叶斯理论 18 2．3．3最大似然估计法 19 2．4聚类分析简介 20 第三章基于随机矩阵理论的关联系数矩阵分析 23 3．1随机矩阵理论简介 ．23 3．2 基于随机矩阵理论的关联系数矩阵分析 ．24 3．2．1金融时间序列数据来源 24 3．2．2股票收益率及其关联系数矩阵 25 V 万方数据 3．2．3关联系数矩阵的