上海股市收益序列的长期记忆性建模分析及预测  

Research on the Long Memory Model in Shanghai Stock Market Returns and Forecast

作　　者：张庆翠[1] 

机构地区：[1]天津大学系统工程研究所,天津300072

出　　处：《西北农林科技大学学报（社会科学版）》2003年第5期102-105,共4页Journal of Northwest A&F University(Social Science Edition)

摘　　要：应用ARFIMA模型研究上海股票市场收益序列的长期记忆性,揭示了上海股市收益具有较明显的长期记忆性,进而从另外一个角度证实了上海股票市场尚未达到弱式有效的结论。上海股市收益序列进行预测,并与传统的线性模型相比,分整的ARFIMA模型提高了股票收益序列长期预测的可靠性。The paper represents and models the long memory property in Shanghai stock returns sequence with ARFIMA model. We find there is strong evidence of long memory in Shanghai stock returns. Furthermore, from another point of view our results indicate that Shanghai stock market still does not achieve weak form efficiency; As compared to benchmark linear models, the estimated fractional ARFIMA model provides improved outofsample forecasting accuracy for the stock returns sequence over longer forecasting horizons.

关 键 词：长期记忆性 有效市场假说 分整自回归滑动平均模型 预测 

分 类 号：F830.91[经济管理—金融学]