基于DTW-MST模型的全球股市网络拓扑结构研究

Study on the topology structure of global stock markets networks based on DTW-MST models

基于次贷与欧债危机下的全球最具代表性的40个股指日收益率,运用DTW(动态时间规整法)和MST(最小生成树)构建和解析了全球股市关联网络及其动态拓扑结构.结果表明:DTW法解决了非线性、长度不一致的数据间的相关性测度问题;全球股指显示出较强的地理聚集性,金融危机发生时的股指网络变得相对更松散,股指间的关联性更弱,网络结构性更差,但欧债危机之后的网络结构比次贷危机之后的要好;金融危机后,股指间的关联性在增大,股指网络的中心聚集性在增强,美国股市的影响力有所削弱,中国股市一直处于边缘地位.

Based on the global most representative 40 stock index daily yields under the subprime mortgage and the European debt crisis,the DTW and MST is applied to construct and analyze the correlation network of global stock markets and its dynamic topology structure.The results show that:the dynamic time warping method may solve the problem of correlation measurement between the data of nonlinear and different lengths;the global stock indexes show stronger geographical aggregation,the stock indexes networks became relatively looser when the financial crisis occur,the correlation between stock indexes is weaker,the network has worse structure,but the network after the European debt crisis is better structure than it after the subprime crisis;after the financial crisis,the correlation between the stock indexes is increasing,the central aggregation of stock index networks is strengthening,the influence of stock markets in America may have been weakened,China stock market has long stayed in peripheral status.