金融高频数据的分析及实证研究

发布时间：2018-06-13 03:31

本文选题：高频数据 + “日历效应”　；参考：《中南大学》2013年硕士论文

【摘要】：随着计算机技术的发展和通讯技术的革新,金融高频数据的获取与操作变得可行和简单。由于高频数据包含了丰富的市场信息,及时从这大量数据中提取有效信息变得越来越重要。本文主要从周期性和长记忆性两方面来分析我国股市高频数据。首先,利用传统方法对我国股市高频数据进行初步的统计分析,得出我国股市高频收益率序列并不是正态分布,而是呈现出高峰厚尾性,且有着显著的“日历效应”。其次,介绍了“已实现”波动率及其扩展形式,分析得出,由于赋权“已实现”波动率充分考虑了“日历效应”,所以,赋权“已实现”波动率比“已实现”波动率和调整“已实现”波动率更有效。同时,在综合考虑测量误差和微观结构误差这两种误差的基础上得出“已实现”波动率和赋权“已实现”波动率的最优抽样频率为5min。最后,利用R/S分析法和修正的R/S分析法分别进行计算,其结果均表明赋权“已实现”波动率具有长记忆性。于是,在长记忆HAR-RV模型的基础上,本文提出了HAR-WRV模型,并以2011年度上证综指采集频率为5min的赋权“已实现”波动率和“已实现”波动率为例,通过实证分析得出,这两个模型均能很好地对波动率进行拟合和预测,且HAR-WRV模型的拟合和预测效果均优于HAR-RV模型。
[Abstract]:With the development of computer technology and the innovation of communication technology, the acquisition and operation of financial high frequency data becomes feasible and simple. Because the high frequency data contain abundant market information, it becomes more and more important to extract effective information from the large amount of data in time. This paper analyzes the high frequency data of Chinese stock market from two aspects of periodicity and long memory. Firstly, using the traditional method to analyze the high frequency data of Chinese stock market, it is concluded that the series of high frequency returns is not a normal distribution, but a peak and a thick tail, and has a significant "calendar effect". Secondly, the "realized" volatility and its extended form are introduced. It is concluded that, because the "realized" volatility fully considers the "calendar effect", Weighted "realized" volatility is more effective than "realized" volatility and adjustment "realized" volatility. At the same time, the optimal sampling frequency of "realized" volatility and weighted "realized" volatility is 5 mins on the basis of synthetically considering two kinds of errors: measurement error and microstructure error. Finally, by using the R- / S method and the modified R- / S method, the results show that the weighted "realized" volatility has long memory. Therefore, based on the long memory HAR-RV model, this paper puts forward the HAR-WRV model, and takes the "realized" volatility and the "realized" volatility of the Shanghai Composite Index in 2011 as an example. Both models can fit and predict volatility, and the effect of HAR-WRV model is better than that of HAR-RV model.
【学位授予单位】：中南大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F830.91;F224

【参考文献】