股市复杂网络的聚类结构分析

发布时间：2018-03-13 12:36

本文选题：复杂网络　切入点：社团结构　出处：《华南理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：在经济一体化的今天，各个行业间的经济合作越来越紧密，彼此间的影响越来越密切。作为国民经济晴雨表的证券市场，其安全问题越来越突出，风险管理难度越来越大。在这样一个大背景下，研究各个行业股票之间的关联关系有着迫切的理论和现实意义。随着1998年《“小世界”网络的群体动力行为》和《随机网络中标度的涌现》两篇开创性论文的发表，复杂网络理论研究已经成为学术界的研究热点之一。十余年来，国内外学者利用复杂网络理论并结合其它学科进行交叉研究取得了可观的研究成果。已有学者证明股票市场是一个复杂系统，区别于传统金融学基于有效市场假说（EMH）的研究方法，利用复杂网络理论研究股票市场是从宏观的角度研究其拓扑结构和整体特性。利用这一个全新的角度研究中国股票市场具有重要的现实意义。经过2008年金融危机后，在2011年至2013年间，，中国股市整体上仍然一路下挫。为了进一步研究中国股票市场的内部结构，探讨其运行机制，本文笔者利用复杂网络理论并运用R软件及其igraph软件包来研究中国股票市场，并做了以下工作：1.结合复杂网络理论，利用最小生成树算法（MST）构建了中国股票关联网络模型。然后探索性分析了所有股票在不同时间段内的收益率分布情况，发现整体上呈现尖峰厚尾的分布。2.利用Newman快速算法对股票关联网络进行社团划分，并通过对比各个社团内股票间的平均相关系数，发现股市在严重下跌时的平均相关系数明显大于股市稳定时，且行业聚集程度明显加大。3.通过比较分析网络中各个节点的度、介数、接近度和特征向量，发现金融和能源行业的股票在网络中处于中心位置，这也和现实中该两个行业在经济体系中占有重要地位的情况相吻合。此外，还发现两个重要的统计特征，一个是具有高“介数中心性”的节点是连接多个社团的重要的节点；另一个是同时具有高“介数中心性”与高“特征向量中心性”的节点具有强相关性。
[Abstract]:In today's economic integration, the economic cooperation among various industries is getting closer and closer, and the influence of each other is becoming more and more close. As a barometer of the national economy, the security problems of the securities market are becoming more and more prominent. Risk management is becoming more and more difficult. Under such a background, it is of urgent theoretical and practical significance to study the relationship between stocks in various industries. With the publication of two groundbreaking papers in 1998, "the dynamic behavior of small World" Network and "the emergence of scale in Random Networks", the theory of complex networks has become one of the research hotspots in academic circles for more than ten years. Scholars at home and abroad have made considerable achievements by using the theory of complex network and combining with other disciplines to carry out cross-research. Some scholars have proved that the stock market is a complex system. Different from the traditional approach of finance based on efficient market hypothesis (EMH), Using complex network theory to study the stock market is to study its topological structure and overall characteristics from the macroscopic angle. It is of great practical significance to use this new angle to study the stock market in China. After the financial crisis of 2008, between 2011 and 2013, the Chinese stock market as a whole continued to fall. In order to further study the internal structure of China's stock market and discuss its operating mechanism, In this paper, the author uses the complex network theory and R software and igraph software package to study the Chinese stock market, and does the following work: 1. Combining with the complex network theory, Based on the minimum spanning Tree algorithm (MST), a Chinese stock correlation network model is constructed, and then the yield distribution of all stocks in different time periods is analyzed. It is found that the whole distribution of peak and thick tail. 2. Using Newman fast algorithm to divide the stock association network, and by comparing the average correlation coefficient among the stocks in each community, It is found that the average correlation coefficient of the stock market in the severe decline is obviously higher than that in the stable stock market, and the degree of industry aggregation is obviously increased .3.Through comparing and analyzing the degree, the medium, the proximity and the characteristic vector of each node in the network, Found that stocks in the financial and energy sectors are at the centre of the network, which is consistent with the fact that the two industries are important in the economic system. Two important statistical features have also been found. One is that the node with high "centricity" is an important node connecting multiple communities, the other is that the node with high "centricity" and "eigenvector centrality" has strong correlation.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51;O157.5

【参考文献】