中国股市统计套利的可行性检验与套利信号的确定方法

发布时间：2018-04-27 01:33

本文选题：统计套利 + 协整　；参考：《南京大学》2014年硕士论文

【摘要】：统计套利是一种依赖市场做空机制,基于统计模型的量化投资策略。中国于2010年3月31日正式开展融资融券业务为实现统计套利提供了可能性。统计套利在国外资本市场已经是成熟的方法,成为了机构投资者的常用策略。国内已经有文献证明可以在中国股票市场实现统计套利。关于金融时间序列波动率模型目前主要有两大类：ARCH模型及其扩展和SV模型及其扩展。关于SV模型在统计套利的研究不是很多,本文旨在用实证数据检验GARCH模型和SV模型是否能中国股票市场实现统计套利,并比较GARCH模型和SV模型的模拟效果以及实际盈利能力。本文基于协整理论采用配对交易策略,本文用简单标准差、时变标准差确定交易信号和止损信号。其中时变标准差采用GARCH模型和SV模型进行模拟,以比较GARCH模型和SV模型对金融时间序列的刻画能力。本文收集沪深两市银行板块可进行融资融券业务的股票进行实证分析,检验模型拟合效果以及统计套利策略在中国股市的实际盈利能力。实证结果显示,三种方法都获得了稳定的收益,并且SV模型对时变标准差的拟合效果要优于GARCH模型。
[Abstract]:Statistical arbitrage is a quantitative investment strategy based on statistical model. China officially launched margin financing on March 31, 2010, which provides the possibility for the realization of statistical arbitrage. Statistical arbitrage is a mature method in foreign capital markets and has become a common strategy for institutional investors. Domestic literature has proved that the Chinese stock market can achieve statistical arbitrage. There are two kinds of financial time series volatility models: arch model and SV model. There are not many researches on SV model in statistical arbitrage. This paper uses empirical data to test whether GARCH model and SV model can realize statistical arbitrage in Chinese stock market, and compares the simulation effect and actual profitability of GARCH model and SV model. In this paper, based on the co-arrangement theory, the paired trading strategy is used. In this paper, simple standard deviation and time-varying standard deviation are used to determine the transaction signal and stop loss signal. The time-varying standard deviation is simulated by GARCH model and SV model to compare the ability of GARCH model and SV model to describe financial time series. In this paper, we collect the stocks in Shanghai and Shenzhen stock markets that can carry out margin trading, test the effect of model fitting and the actual profitability of statistical arbitrage strategy in Chinese stock market. The empirical results show that the three methods obtain stable returns and the SV model is better than the GARCH model in fitting the time-varying standard deviation.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51

【参考文献】