基于高频数据的国债期货跨品种套利策略研究

发布时间：2018-03-21 03:48

本文选题：高频数据　切入点：国债期货　出处：《山西财经大学》2017年硕士论文　论文类型：学位论文

【摘要】：2015年3月20日,10年期国债期货在中国金融期货交易所挂牌上市,我国国债期货市场实现了从单品种到多品种的突破。随着10年期国债期货的上市以及国债期货市场流动性的不断提高,国债期货跨品种套利交易成为可能。由于10年期国债期货推出不久,鲜有学者对国债期货跨品种套利进行研究,投资者进行国债期货跨品种套利时能参考的依据不多,行之有效的国债期货交易策略也亟待丰富。因此,本文将对国债期货跨品种套利的理论依据和交易策略进行研究,力求为投资者提供有效的建议以及促进国债期货市场跨品种交易的发展。本文以中国金融期货交易所的5年期和10年期国债期货合约价格的高频数据为研究对象,分析合约之间的价格关系,采用门限自回归模型来确定无套利区间,并以此制定套利策略并分析套利效果。主要采用理论分析和实证研究相结合的研究方法进行研究。利用EXCEL、EVIEWS、R软件和MATLAB等计量工具来实现实证研究和模拟套利交易。首先选取5年期和10年期国债期货合约的5分钟高频数据进行价差分析和平稳性检验。然后基于门限自回归模型,采用合适的检验门限行为的方法,先确定滞后阶数,再参考sup-Wald检验方法进行门限行为检验;在参数估计方面,运用的格子搜索法估计门限自回归模型的门限值,得到无套利区间后,在此基础上设定套利交易信号构建套利策略。针对套利策略进行套利效果分析,计算其在样本内外数据中的收益表现,检验套利策略在国债期货跨品种套利交易中的有效性。研究结果表明,5年期国债期货和10年期国债期货价格数据之间既存在长期均衡关系,又存在门限协整行为,然后估计出门限自回归模型的两个门限值作为套利交易的信号,利用估计出的门限值所对应套利策略对样本内数据进行套利得到的效果良好,成功率达到69%,将既有套利策略运用在样本外数据进行套利的效果虽没有样本内的效果好,但是成功率也达到了61%,说明了所构建策略的有效性。
[Abstract]:In March 20th 2015, 10-year Treasury bonds futures were listed on the China Financial Futures Exchange. China's treasury bond futures market has achieved a breakthrough from single variety to multiple varieties. With the listing of 10-year Treasury bond futures and the increasing liquidity of the treasury bond futures market, Cross-variety arbitrage trading of treasury bonds futures is possible. Since not long after the introduction of 10-year Treasury bond futures, few scholars have carried out research on cross-variety arbitrage of treasury bonds futures, and investors can not refer to the basis for cross-variety arbitrage of treasury bonds futures. Therefore, this paper will study the theoretical basis and trading strategy of cross-variety arbitrage of treasury bond futures. In order to provide effective advice to investors and to promote the development of cross-variety trading in treasury bond futures market, this paper studies the high frequency data of 5-year and 10-year bond futures prices in China Financial Futures Exchange. The price relationship between contracts is analyzed and the threshold autoregressive model is used to determine the no-arbitrage interval. This paper makes a arbitrage strategy and analyzes the effect of arbitrage. It mainly adopts the research method of combining theoretical analysis with empirical research. It uses Excel EVIEWSR software and MATLAB to realize empirical research and simulate arbitrage trading. Firstly, the 5-minute high frequency data of 5-year and 10-year Treasury futures contracts are selected for price difference analysis and stability test, and then based on threshold autoregressive model, In the parameter estimation, the lattice search method is used to estimate the threshold value of the threshold autoregressive model, and the delay order is first determined by the appropriate method of testing threshold behavior, and then the threshold behavior is tested by referring to the sup-Wald test method, and the lattice search method is used to estimate the threshold value of the threshold autoregressive model. After obtaining the no-arbitrage interval, we set up the arbitrage trading signal to construct the arbitrage strategy, and analyze the arbitrage effect of the arbitrage strategy, and calculate the performance of the arbitrage in the data inside and outside the sample. The results show that there is a long-term equilibrium relationship and threshold cointegration between the price data of 5-year Treasury bond futures and 10-year treasury bonds futures. Then the two threshold values of the threshold autoregressive model are estimated as the signals of arbitrage trading. The arbitrage strategy corresponding to the estimated threshold value is used to arbitrage the data in the sample and the result is good. The success rate is 69%. The effect of applying the existing arbitrage strategy to the data outside the sample is not as good as that in the sample, but the success rate also reaches 61%, which shows the effectiveness of the strategy.
【学位授予单位】：山西财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F812.5;F724.5

【参考文献】