高频金融时间序列的异象特征分析及应用——基于多重分形谱及其参数的研究  

Analysis on Anomalous Characteristics of High frequency Time Series and Its Applications——Based on Multifractal Spectrum and Its Parameters

作　　者：周孝华[1] 宋坤[1] 

机构地区：[1]重庆大学经济与工商管理学院,重庆400030

出　　处：《财经研究》2005年第7期123-132,共10页Journal of Finance and Economics

基　　金：国家自然科学基金资助项目(70473107)

摘　　要：文章首先从理论上推导出金融资产价格的高频时间序列出现大幅震荡前后多重分形谱所具有的异象特征,然后随机选取两只股票(民生银行、哈飞股份)各35天的5min高频交易数据对上述特征进行实证分析。结果表明,两只股票在持续大幅波动开始与结束时,其多重分形谱形态及参数的变化与理论上的异象特征相吻合。运用该研究方法可以对金融资产持续大幅波动的开始及结束做出一定预测。We first deduce the theoretical anomalous characteristics of multifractal spectra on high frequency financial time series during the period of significant price fluctuations. Taking two stocks’ 5min high frequency trading data (totaling 35 business days) as examples, we then demonstrate that multifractal spectra’s shapes and key parameters are consistent with the above theoretical anomalies. Using the method in this paper, we can forecast when abnormal fluctuations on prices of financial assets occur and finish.

关 键 词：多重分形谱 高频金融时间序列 大幅震荡 预测 

分 类 号：F830.59[经济管理—金融学]