基于高频数据的中国股票收益率研究

发布时间：2018-04-02 12:31

本文选题：稳定分布　切入点：高频数据　出处：《华东师范大学》2013年硕士论文

【摘要】：基于上海证券交易所每0.5秒更新一次的高频数据,利用稳定分布拟合中国股票市场收益率,探究稳定分布的四个参数与不同抽样频率之间的关系,以及某个抽样频率下,不同取样时间段里稳定分布四个参数的变化情况。本文内容以及结果如下：第一,介绍了稳定分布的相关理论、性质；第二,基于稳定分布,利用高频数据对不同频率下的中国股票市场收益率进行实证分析,实证结果表明不同频率下,特征指数α没有明显趋势,可以给出具体的α估计值；偏度参数p变化不规则,但将不同频率下的p MLE估计值组成序列,检验结果表明p估计值显著大于或者小于零,说明尽管抽样频率不同,但每个频率下的β值都能反映这段时间里收益分布的偏向情况；随着抽样频率的增大,尺度参数逐渐递减趋近于0,表明频率越大,在这个频率下的股票市场波动越小；将不同频率下的μ估计值组成序列,除了沪深300指数的μ的估计值显著大于零外,其余的样本股、上证50指数的μ的估计值显著小于零。第三,基于稳定分布,对频率为五分钟的沪深300指数收益率进行参数时变性分析。采用的是沪深300指数五分钟数据,时间区间为2005年4月8日至2012年8月31日,将这段时间按照季度分为30个时间段。实证结果表明,特征指数并没有明显的趋势。经过检验,α=1.600就可以刻画这些时间段内样本收益率分布的尾部特征；偏度参数表现出不规则的波动性,与宏观经济指标进行因果分析,检验结果表明宏观指标并不能引起偏度参数的变化；不同时段样本下尺度参数c的MLE先波动上升,再波动下降,呈现倒“V”型。对它和沪深300指数成交量季度变化率进行因果分析,得出成交量的变化是引起c变化的原因,即存在由成交量到c的单向因果关系；稳定分布的位置参数μ随着时间的推移有震荡减小的趋势,对位置参数和沪深300指数成交量变化率进行因果分析,得出成交量是引起μ变化的原因,即存在由成交量到μ的单向因果关系。最后,基于稳定分布,给出了风险度量工具ES的数值计算表达式。
[Abstract]:Based on the high - frequency data updated every 0.5 seconds of Shanghai Stock Exchange , the relationship between the four parameters of stable distribution and the different sampling frequency and the variation of four parameters of stable distribution in different sampling periods are explored by using the stable distribution to fit the yield of China ' s stock market . The contents and results are as follows :

Firstly , the relevant theories and properties of stable distribution are introduced .

Secondly , based on the stable distribution , the empirical results show that there is no obvious trend of the characteristic index 伪 at different frequencies by using high frequency data , and the specific 伪 estimation value can be given .
The p - value of the p - MLE in different frequencies is larger than or smaller than zero , which shows that although the sampling frequency is different , the 尾 - value at each frequency can reflect the deviation of the income distribution during the period .
As the sampling frequency increases , the scaling parameters gradually decrease to 0 , indicating that the higher the frequency , the smaller the volatility of the stock market at this frequency ;
The value of 渭 in Shanghai - Shenzhen 300 - index was significantly greater than zero except that 渭 of Shanghai - Shenzhen 300 index was significantly greater than zero , and the estimated value of 渭 in Shanghai - Shenzhen 300 index was significantly less than zero .

Thirdly , on the basis of the stable distribution , we analyze the parameters of the Shanghai - Shenzhen 300 - index yield with a frequency of five minutes . The results show that the time interval is from April 8 , 2005 to August 31 , 2012 , and the time interval is divided into 30 periods according to the quarter . The empirical results show that the characteristic index has no obvious trend . After examination , 伪 = 1.600 can depict the tail features of the sample yield distribution within these periods .
The results show that the macroscopic index can not cause the variation of the bias parameter .
The first fluctuation of the MLE of the scale parameter c in different period samples increased , then the fluctuation decreased , and the inverted " V " type was presented . The causal analysis was carried out on the rate of change of the yield of the 300 index in Shanghai and Shenzhen .
On the basis of the stable distribution , the numerical expression of the risk measurement tool ES is given .

【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】：F224;F832.51

【参考文献】