基于混频数据支持向量回归的股指期货套利策略

发布时间：2018-01-02 06:40

本文关键词：基于混频数据支持向量回归的股指期货套利策略　出处：《浙江工商大学》2017年硕士论文　论文类型：学位论文

【摘要】：金融二级市场上股指期货不仅是单边投机的重要标的,更是实现套期保值和双边套利的重要手段。股指期货可以对A股市场起到价格发现以及平稳趋势的作用。但是在近期过往时间段内(尤其2015年中),中国股票二级市场经历了巨大的波动,这波动不仅表现在股票自身价格的大幅起落,也表现在股票现货与期货之间的价差的激烈起伏。按照套利理论,由于股票现货和期货之间由于强相关性和价格收敛性质的存在,二者之间价差应当稳定在一定区间范围内。市场上期现基差的剧烈波动,对套利交易者来说,既意味着隐藏风险也蕴含着获利机会。对基差范围的准确预测是套利交易有效进行的重要保证。近年由于大数据概念、统计学习及深度学习等方法的蓬勃发展,越来越多的预测模型被运用到量化投资领域当中。支持向量回归这一机器学习方法由于基于结构风险最小化的理念和核函数的高维映射能力,使其成为外推能力优异的预测方法。本文选取不同频率的行情数据作为输入特征,通过支持向量回归方法,既预目标测交易日中的基差值又预测基差波动幅度。通过叠加基差值和基差波动幅度,就可以得到基差的波动范围,即无套利区间。之后本文在此无套利区间上形成套利交易策略。利用1分钟实时行情数据的突破和回复,作为交易系统的开平仓条件。最后通过实证数据来验证交易策略,检测其总体收益情况以及在不同行情阶段的表现。
[Abstract]:Stock index futures in the secondary financial market are not only the important targets of unilateral speculation. Stock index futures can play a role in price discovery and stable trend of A share market. But in the recent past time period (especially middle of 2015). China's secondary stock market has experienced tremendous fluctuations, which is not only reflected in the volatility of the stock price itself, but also in the sharp fluctuation of the price difference between stock spot and futures. According to arbitrage theory. Due to the existence of strong correlation and price convergence between stock spot and futures, the price difference between them should be stabilized in a certain range. The sharp fluctuation of the current basis in the market is for arbitrage traders. Accurate prediction of the basis range is an important guarantee for the effective carrying out of arbitrage trading. In recent years, due to the rapid development of big data concept, statistical learning and in-depth learning, etc. More and more prediction models are applied to the field of quantitative investment. Support vector regression (SVM) is a machine learning method based on the concept of structural risk minimization and the ability of high-dimensional mapping of kernel functions. This paper selects the different frequency quotation data as the input feature, and adopts the support vector regression method. The base difference and the base fluctuation range can be obtained by adding the base value and the base fluctuation range. That is, no arbitrage interval. Then this paper forms a arbitrage trading strategy in this no-arbitrage range. Using the one-minute real-time market data breakthrough and recovery. Finally, through empirical data to verify the trading strategy, to test its overall earnings and performance in different stages of the market.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51

【参考文献】