重标极差方法下时变Hurst指数的构建和实证研究_覃邑龙.pdfVIP

20 5 系 统 管 理 学 报 Vol. 20 No. 5 20119 Journal of Systems Management Sep. 2011 : 1005-2542( 2011) 05-0 20-07 Hurst 覃邑龙, 应益荣 ( , 200044) = 在时间序列的重标极差分析方法的基础上, 定义 一个时变H urst 指数的概念, 推导并证明 时变Hurst 指数的2 条性质, 并引入滑动分块自助法计算Hurst 指数的标准差以此为基础, 用 Hurst 指 数的下限及随机过程服从随机游走时的Hurst 的上限作为分析工具, 对Hurst 指数能否成功预测市场的反 转进行 研究实证研究表明: 时变Hurst 指数能成功地预测市场的反转 : 重标极差分析; 时变Hurst 指数; 滑动分块自助法; 长记忆性 : TP 18 : A The Construction and Empirical Testing of the Time-varying Hurst Exponent by the Rescale Range Analysis QI N Yi-long , YI N G Yi-rong ( School of Management, Shanghai University, Shanghai 200044, China) =Abstract This paper introduces a new concept of the time-varying Hurst exponent, based on the rescale range analysis of the time series. Then, it develops an algorithm to compute the stand deviation of the Hurst exponent by the moving block bootstrap method. On this basic, it explores how to obtain the Hurst exponent. s low er bound and upper bound and the Hurst exponent. s upper bound w hen the random process is random w alk on a certain confidence level. It also studies if the Hurst exponent can be used to predict the market reversal. Empirical results show that the time-varying Hurst exponent can successfully predict the market reversal. Key words: rescale range analysis; time-varying Hurst exponent; moving block bootstrap method; long memory , ( R/ S) , , Hurst 1951 , , Mandel- brot , , 0. 5, , M andelbrot[ 1] , : 2010-0 -21 :2