摘要
研究了沪深股市收益率序列中的长记忆性,通过经验分布的概率密度函数对收益率的非正态分布进行了实证研究,然后应用R/S非线性估计方法以及AR和MA模型对结果的显著性进行了检验。主要针对平均循环周期的测定问题,引入了支持向量回归机的方法帮助测定,能够比较清楚地给出沪深股市中3个重点行业的平均循环周期。分析结果表明,中国行业市场普遍具有分形特征,其变化具有循环性。
The paper mainly addresses the long-term memory effect of returns serials in Shanghai and Shenzhen stock market.The non-normal distribution character of returns is studied through the probability density function based on the Empirical distribution.R/S non-linear estimate method,AR and MA models are used to prove the distinctiveness of the result.Aiming at the measuring of average cyclic period,the support vector regression method is presented to generate the relatively clear results of three key industrials in the stock market.The final analysis indicates that the whole Chinese industrials market has the universal behavior of fractal properties,and varys in periodic mode.
出处
《江南大学学报（自然科学版）》
CAS
2010年第2期244-249,共6页
Joural of Jiangnan University (Natural Science Edition)
基金
江苏省自然科学基金资助项目(BK2008098)
