基于复杂网络的时间序列双变量联动波动

选择伦敦金与Au9999下午收盘价格作为样本数据研究时间序列双变量之间的联动波动规律.依据粗粒化方法,将伦敦金与Au9999价格的联动波动状态转化为由5个{P,N,M}字符组成的字符串,每个字符串代表5天的价格联动波动模态.将模态作为节点,模态之间的转化为边,构建价格联动波动复杂网络.运用复杂网络理论对时间序列双变量联动波动模态的统计、变化规律和演化机制进行分析.结果表明:时间序列双变量联动波动模态分布具有幂律性、群簇性和周期性,其联动波动模态主要通过少数几种模态进行转换与演化.本方法不仅可以研究不同类型时间序列双变量联动波动,同时可为多变量联动波动研究提供思路.

Linkage Fluctuation in Double Variables of Time Series Based on Complex Networks

To study linkage fluctuation of double variables of time series,we took London Gold and Au9999. Model of linkage fluctuation is consisted of characters {P,N,M},using coarse graining process. Nodes of complex network are 5-symbol strings. Linkage fluctuation complex network is composed of all models and link edges between them. It indicates that models have power-law distribution,clustering and periodicity. Transmission and evolution are finished mainly by few models. It provides an analyzing method for many kinds of double variables of time series and ideas for general law of multi-variables of time series,as well.