股市时间序列的多重分形消除趋势分析  

Analysis of Multifractal Detrended Fluctuation in Stock Market Time Series

作　　者：袁平平[1] 于建玲[1] 商朋见[1] 

机构地区：[1]北京交通大学理学院,北京100044

出　　处：《北京交通大学学报》2007年第6期69-72,共4页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：科技部"973"基金资助项目(2004CB318005)

摘　　要：为了探索股票时间序列的无标度性,应用多重分形消除趋势分析的方法(MF-DFA)来研究沃尔玛指数(WMT)日收盘价.研究结果表明,沃尔玛指数日收盘价的变化具有多重分形的特性,广义Hurst指数显著依赖于波动函数的阶数,并随之变化;尺度函数表现出明显的非线性性质;多重分形谱呈现单峰钟形图像.To detect the scale qualities of stock market time series, we apply Muhifractal Detrended Fluctuation Analysis（MF-DFA） to WMT index daily closing quotation. The research results show the change of WMT index daily closing quotation has muhifractal properties. The generalized hurst exponent obviously depends on the order of fluctuate function and changes with it; The curve of scaling function clearly departs from a straight line, i.e. it shows clearly nonlinear property; The muhifractal spectrum displays the commonly observed bell shape. This will provide an important theoretic foundation for reseaching muhifractality in theory of finance.

关 键 词：股票市场 多重分形消除趋势分析 时间序列 多重分形谱 

分 类 号：O211.61[理学—概率论与数理统计]