金融时间序列的波动率分析

发布时间：2018-01-13 06:13

本文关键词：金融时间序列的波动率分析　出处：《华东师范大学》2013年硕士论文　论文类型：学位论文

【摘要】：本文对中国股市的波动率进行了研究,分别选取上证指数和深圳综指的对数收益率序列为样本序列。结合假设检验和统计推断的分析方法,在正文中验证了样本数据的自相关性,平稳性等统计特征,并发现数据并非是正态分布的,而是表现出一种高峰厚尾的性质。在第五章中具体选取了两组样本数据,分别对其拟合七组模型,挖掘波动率的条件异方差,非对称性等性质。结果发现随着模型的改进,拟合结果有所改善。就方差模型而言,GARCH模型的拟合效果要好于ARCH模型；TGARCH模型和EGARCH模型的拟合效果较为接近,且均优于GARCH模型；进一步,在均值模型中加入分数差分的概念,使用ARFIMA模型,在此基础上再建立方差模型,其结果要优于均值模型为ARMA模型的情况。最后发现ARFIMA-TGARCH模型是所有模型中拟合效果最好的,这是因为模型充分提取了序列的长记忆性,杠杆效应等信息。
[Abstract]:In this paper, the volatility of Chinese stock market is studied. The logarithmic return series of Shanghai Stock Exchange Index and Shenzhen Composite Index are selected as sample series, combined with hypothesis test and statistical inference analysis method. The autocorrelation and stationarity of the sample data are verified in the text, and it is found that the data are not normal distribution. In Chapter 5th, two groups of sample data were selected to fit seven groups of models respectively, and the conditional heteroscedasticity of volatility was mined. The results show that with the improvement of the model, the fitting results are improved. In terms of variance model, the fitting effect of GARCH model is better than that of ARCH model. The fitting effect of TGARCH model and EGARCH model is close and better than that of GARCH model. Furthermore, the concept of fractional difference is added to the mean value model, and then the variance model is established by using ARFIMA model. The result is better than the mean model is ARMA model. Finally, it is found that the ARFIMA-TGARCH model is the best fit among all the models. This is because the model fully extracts the information of long memory, leverage effect and so on.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224;O212.1

【参考文献】