基于协整的配对交易改进研究
发布时间：2018-07-17 19:30
【摘要】：我国沪深交易所在2010年3月开始推行融资融券交易试点,同年4月推出股指期货。这从根本上改变了以往我国股市投资者只能单边做多的局面,能够提高我国股市资金配置效率。两融业务的实施也使投资者能够实施更多的量化投资策略,从而更好地回避风险,提高收益。基于统计套利的配对交易策略是一种市场中性策略,通过识别配对股票价格的相对偏离,利用融资融券,做空相对高估的股票,做多相对低估的股票,并在两只股票价格差距回复至正常水平时进行平仓从而实现盈利。由于配对交易策略并不基于单只股票的涨跌,而是基于两只配对股票价格之间的相对差距,因此该策略能够平复市场风险,横跨牛、熊市实现稳定的套利。本文旨在验证配对交易对于中国股市的适用性,同时对常用的基于协整的配对交易步骤进行一定的改进,为国内投资者进行量化投资时提供一定的参考。本文首先回顾了之前学者们的研究成果,叙述国内外的研究现状和近年研究的方向。通过对近年国内相关文献的阅读,发现直接套用国外文献中将0.75倍标准差作为开仓阈值,2倍标准差作为止损阈值的做法可能并不适合国内的状况。然后本文论述了配对交易的分类、概念、常用的配对交易方法以及与之相关的数学模型。在介绍配对交易的常用方法时重点介绍了基于协整的配对交易方法具体过程。在此基础上,本文选取万德行业分类中的银行板块和地产经营类板块进行进一步实证分析。实证部分的数据被分为了样本内和样本外两个部分。2010年至2014年所有交易日数据作为样本内数据用于验证开仓阈值和止损阈值设置的合理性,取得适合于所选取股票对的最优开仓阈值和止损阈值的组合。2015年的所有交易日数据作为样本外数据,用以验证所选取的开仓阈值和止损阈值的有效性。在实证部分本文通过比较各板块内股票之间的相关系数来选取用于配对交易实证的股票对,对相关系数最高并通过协整检验的股票对建立误差修正模型,用于分析基于常数方差和基于GARCH模型的条件异方差两种情形下开仓阈值和止损阈值设置的合理性,并计算最优阈值下的收益情况。在实证部分选出了交通银行和宁波银行、云南城投和嘉凯城这两对组合作为配对股票对进行模拟实证交易。样本内实证结果表明传统的将常数(条件)标准差的0.75倍设置为开仓线,将常数(条件)标准差的2倍设置为止损线并不合理,开仓线与止损线的设置与配对股票有关,同时也与所选择的模型有关,通过对每组配对交易股票对计算最优阈值进而进行套利能够实现更高的收益。样本外的实证同样验证了通过上述方法所选择的开仓阈值和止损阈值组合不论是基于常数方差还是条件方差,即使是行情波动极为激烈的2015年,也能实现可观的收益,但是基于条件方差的配对交易并不一定能比基于常数方差的配对交易取得更高的收益。实证部分所得收益在两组股票对之间相差十分巨大,推测这样的结果可能与股票的价格波动剧烈程度有关;同时样本外实证部分所得收益也明显优于样本内所得的最优收益率,推测这可能与研究所选取的时间段有关。在本文的最后,还提出了一些研究中的不足,如没有验证所选时间段的合理性,没有证明止损阈值设置的必要性等。
[Abstract]:The Shanghai and Shenzhen Stock Exchange launched the pilot of margin trading in March 2010. The stock index futures were introduced in April of the same year. This fundamentally changed the situation that the investors in our stock market can only have a single side and can improve the efficiency of capital allocation in China's stock market. The implementation of the two thawing business also enables investors to implement more quantitative investment strategies. In order to better avoid risk and raise income, the paired trading strategy based on statistical arbitrage is a market neutral strategy. By identifying the relative deviation of the stock price, using margin trading, making short relatively overvalued stock, making a relatively undervalued stock, and holding a warehouse in two stock prices when the price gap is back to the normal level. The strategy is based on the relative gap between the two pairs of stock prices, which is based on the relative gap between the two pairs of stock prices, so the strategy can restore the market risk, cross the bull and bear market to achieve a stable arbitrage. The whole process of paired transaction is improved to provide some reference for domestic investors to quantify investment. This paper first reviews the research results of previous scholars, and describes the current research status at home and abroad and the direction of research in recent years. Through the reading of relevant domestic literature in recent years, it is found that 0.75 of the foreign literature will be directly applied. The double standard deviation as the opening threshold, the 2 times the standard deviation as the stop threshold may not be suitable for the domestic situation. Then this paper discusses the classification of the paired transactions, the concept, the common paired transaction methods and the related mathematical models. In the introduction of the common method of paired transaction, the cointegration based pairing transaction is introduced. On this basis, this paper makes a further empirical analysis of the bank plate and the real estate sector in the wad industry classification. The data of the empirical part are divided into two parts of the sample and out of samples from.2010 to 2014 as the sample data to verify the opening threshold and the stop threshold. The reasonableness of the value setting, the data of all trading days that fit the optimal opening threshold and the stop loss threshold of the selected stock for.2015 years as the sample data, to verify the validity of the selected opening threshold and the stop loss threshold. In the empirical part, this paper compares the correlation coefficient between the shares in each plate to select the correlation coefficient. The stock pair for the positive paired transaction is used to establish an error correction model for the stock with the highest correlation coefficient and through cointegration test. It is used to analyze the rationality of the opening threshold and the stop threshold setting in two cases of conditional heteroscedasticity based on constant variance and GARCH model, and calculate the income under the optimal threshold. The two combinations of the Bank of communications and Ningbo, the Yunnan city investment and the Jia Kai City are selected as the paired stock pairs. The empirical results show that the traditional constant (conditional) standard deviation is 0.75 times as the opening line, and the loss line is not reasonable until the 2 times the standard deviation of the constant (condition), the opening line and the stop loss. The setting of a line is related to the paired stock, and it is related to the chosen model. By calculating the optimal threshold to arbitrage each group of paired trading stocks, a higher return can be achieved. The empirical results also verify that the opening threshold and the stop threshold combination selected by the above method are based on the constant variance. It is a conditional variance, even if the volatility of the market is extremely intense in 2015, it can also achieve considerable profits, but the conditional variance based pair trading does not necessarily achieve higher returns than the pair based on constant variance. The income of the empirical part is very large between the two groups of stock pairs. At the same time, the income from the empirical part of the sample is also obviously superior to the optimal rate of return from the sample. It is presumed that this may be related to the time period selected by the study. At the end of this paper, some deficiencies in the study are put forward, such as no verification of the reasonableness of the selected time period and no proof of the stop loss. The necessity of setting the threshold and so on.
【学位授予单位】：浙江财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51
本文编号：2130699
[Abstract]:The Shanghai and Shenzhen Stock Exchange launched the pilot of margin trading in March 2010. The stock index futures were introduced in April of the same year. This fundamentally changed the situation that the investors in our stock market can only have a single side and can improve the efficiency of capital allocation in China's stock market. The implementation of the two thawing business also enables investors to implement more quantitative investment strategies. In order to better avoid risk and raise income, the paired trading strategy based on statistical arbitrage is a market neutral strategy. By identifying the relative deviation of the stock price, using margin trading, making short relatively overvalued stock, making a relatively undervalued stock, and holding a warehouse in two stock prices when the price gap is back to the normal level. The strategy is based on the relative gap between the two pairs of stock prices, which is based on the relative gap between the two pairs of stock prices, so the strategy can restore the market risk, cross the bull and bear market to achieve a stable arbitrage. The whole process of paired transaction is improved to provide some reference for domestic investors to quantify investment. This paper first reviews the research results of previous scholars, and describes the current research status at home and abroad and the direction of research in recent years. Through the reading of relevant domestic literature in recent years, it is found that 0.75 of the foreign literature will be directly applied. The double standard deviation as the opening threshold, the 2 times the standard deviation as the stop threshold may not be suitable for the domestic situation. Then this paper discusses the classification of the paired transactions, the concept, the common paired transaction methods and the related mathematical models. In the introduction of the common method of paired transaction, the cointegration based pairing transaction is introduced. On this basis, this paper makes a further empirical analysis of the bank plate and the real estate sector in the wad industry classification. The data of the empirical part are divided into two parts of the sample and out of samples from.2010 to 2014 as the sample data to verify the opening threshold and the stop threshold. The reasonableness of the value setting, the data of all trading days that fit the optimal opening threshold and the stop loss threshold of the selected stock for.2015 years as the sample data, to verify the validity of the selected opening threshold and the stop loss threshold. In the empirical part, this paper compares the correlation coefficient between the shares in each plate to select the correlation coefficient. The stock pair for the positive paired transaction is used to establish an error correction model for the stock with the highest correlation coefficient and through cointegration test. It is used to analyze the rationality of the opening threshold and the stop threshold setting in two cases of conditional heteroscedasticity based on constant variance and GARCH model, and calculate the income under the optimal threshold. The two combinations of the Bank of communications and Ningbo, the Yunnan city investment and the Jia Kai City are selected as the paired stock pairs. The empirical results show that the traditional constant (conditional) standard deviation is 0.75 times as the opening line, and the loss line is not reasonable until the 2 times the standard deviation of the constant (condition), the opening line and the stop loss. The setting of a line is related to the paired stock, and it is related to the chosen model. By calculating the optimal threshold to arbitrage each group of paired trading stocks, a higher return can be achieved. The empirical results also verify that the opening threshold and the stop threshold combination selected by the above method are based on the constant variance. It is a conditional variance, even if the volatility of the market is extremely intense in 2015, it can also achieve considerable profits, but the conditional variance based pair trading does not necessarily achieve higher returns than the pair based on constant variance. The income of the empirical part is very large between the two groups of stock pairs. At the same time, the income from the empirical part of the sample is also obviously superior to the optimal rate of return from the sample. It is presumed that this may be related to the time period selected by the study. At the end of this paper, some deficiencies in the study are put forward, such as no verification of the reasonableness of the selected time period and no proof of the stop loss. The necessity of setting the threshold and so on.
【学位授予单位】：浙江财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51
【参考文献】
相关期刊论文 前5条
1 黄晓薇;余湄;皮道羿;;基于O-U过程的配对交易与市场效率研究[J];管理评论;2015年01期
2 彭舒怡;;基于GARCH模型银行股配对交易研究[J];知识经济;2013年10期
3 王春峰;林碧波;朱琳;;基于股票价格差异的配对交易策略[J];北京理工大学学报(社会科学版);2013年01期
4 蔡燕;王林;许莉莉;;基于随机价差法的配对交易研究[J];金融理论与实践;2012年08期
5 崔方达;吴亮;;配对交易的投资策略[J];统计与决策;2011年23期
，本文编号：2130699
本文链接：https://www.wllwen.com/jingjilunwen/touziyanjiulunwen/2130699.html
