摘要
复杂网络理论是研究股票市场内在结构和功能的有力工具,股票关联网络的拓扑性质和聚类结构对于理解网络的形成机制、发生在网络上的动力学行为具有重要意义。以中国上证180指数和深证100指数成分股票为研究标的,运用最小生成树算法和平面最大过滤图算法构建相应的股票关联网络,分析网络的基本拓扑统计性质和聚类结构。实证研究表明,平面最大过滤图关联网络为小世界网络,各关联网络内股票的影响强度服从幂律分布,股票之间存在的异类匹配模式揭示了市场内股票价格波动传导的过程,对最小生成树关联网络和平面最大过滤图关联网络的宗派和派系聚类分析能有效地挖掘股票之间的聚类结构信息,总体上看平面最大过滤图算法优于最小生成树算法,且实证结论对沪深股票市场具有普适性。
The complex network theory is a powerful tool for the study of inner structure and function of stock market. The topology property and clustering structure of stock correlation network is meaningful for understanding the forming mechanism and dynamic behaviors of the network. This paper uses the minimum-cost spanning tree （MST） and planar maximal filtering graph （PMFG） algorithm to construct corresponding stock correlation network, and study the networks＇ topology property and clustering structure. The original data are the component stocks of Shanghai 180 index and Shenzhen 100 index. The results demonstrate that the PMFG network is small-world＇ network, the stock influence strength obeys power distribution for all networks, the disassortative matching mode between stocks tells the process of price fluctuation conduction, the clan and clique clustering analysis of MST and PMFG can explore the clustering structure information among stocks effectively, as a whole, the PMFG outperforms MST algorithm, all the results are suitable for the two markets.
出处
《管理科学》
CSSCI
2008年第3期94-103,共10页
Journal of Management Science
基金
高等学校博士点基金(20060145001)
关键词
股票关联网络
复杂网络
最小生成树
平面最大过滤图
stock correlation network
complex network
minimum-cost spanning tree
planar maximal filtering graph
