股票符号网络研究

利用符号网络来对股票市场进行研究,利用中国近期股市平稳震荡、牛市、熊市3个时期的数据, 首先使用股票收益率相关系数构建保留连边正负信息的符号网络, 其中正边采取优化阈值法, 负边采用固定阈值法, 发现网络中负边的比例较低且集中在银行股上. 之后重点关注牛市时期网络的特征, 分析了度及度分布、节点的受欢迎程度和特征向量中心性、平衡性、平均集聚系数和度相关性. 将其与传统网络进行对比, 发现负边的引入对节点的重要性有较大影响. 

Analysis of the stock market signed network

Complex networks have been widely applied to financial systems, but only a few utilize signed networks for the stock market. Signed networks are constructed to simulate the Chinese stock market in the present work. Signed networks were generated by calculating correlations between stock returns in the most recent fluctuates, in both bull and bear markets and setting threshold value by optimal threshold and fixed threshold methods. The proportion of negative edges was found low and the majority were between bank stocks. The main focuse of the paper is signed stock network in the bull market. Topological properties were analyzed. Degree distribution, popularity measures and eigenvector centrality are studied on the microcosmic scale,while structural balance, clustering coefficient and degree correlations are studied on the macroscopic scale. The work also compares the signed network with traditional unsigned positive network to investigate the impact of negative edges. These data indicate that negative edges have no influence on the degree distribution but affect the importance of nodes greatly.