基于Copula方法的国债期货市场相依结构研究

发布时间：2018-03-16 18:26

本文选题：尾部相关性　切入点：M-Copula　出处：《天津大学》2016年硕士论文　论文类型：学位论文

【摘要】：对金融市场之间各组成部分相关性的分析,是认识市场运行规律的基础,有效的相关性分析可以用来优化市场结构,提高经济运行效率。传统金融资产间相关性的分析以线性模型为主,但最新的研究显示,金融资产间的存在着非线性及非对称的相关性,这种相关关系无法利用传统的计量方法实现测度。而Copula函数可以对具有该类型特征的金融资产进行刻画,并对于尾部相关性的测度具有突出的优势,可以满足对金融变量相关性分析的需求。本文对Copula函数的理论、特征及主要类别进行了阐述,并在传统Copula函数基础上进行扩展,在利用Copula函数对金融资产间相依结构的分析中,整合常用的三类Archimedean族Copula函数,加入惯性权重的粒子群算法对其参数进行优化,构建了基于惯性粒子群算法的M-Copula模型。利用该模型研究了我国5年期跨期国债期货合约结算价日涨跌幅之间的尾部相关关系,同时利用基于惯性权重的粒子群算法对混合Copula模型的参数进行选取和确定后,比较了混合Copula函数与传统Copula函数对跨期国债期货合约上尾部以及下尾部相关关系的拟合能力。研究结果表明:基于惯性权重粒子群算法的混合Copula函数较好地刻画了跨期国债期货合约结算价日涨跌幅尾部非线性、非对称的相关特征,且根据测度结果可以判定,跨期国债期货合约结算价日涨跌幅间的上尾部相关性更为明显,并依据对该特征的把握构建了能够产生超额收益的交易策略。
[Abstract]:The analysis of the correlation between the components of the financial market is the basis of understanding the law of market operation. Effective correlation analysis can be used to optimize the market structure. The traditional analysis of the correlation between financial assets is based on linear model, but the latest research shows that there are nonlinear and asymmetric correlations among financial assets. However, Copula function can describe the financial assets with this kind of characteristics, and it has a prominent advantage for the measurement of tail correlation. In this paper, the theory, characteristics and main categories of the Copula function are expounded, and extended on the basis of the traditional Copula function. In the analysis of the dependent structure of financial assets by using the Copula function, Integrating three kinds of Copula functions of Archimedean family, the parameters are optimized by PSO with inertial weight. The M-Copula model based on inertial particle swarm optimization algorithm is constructed. Using this model, the tail correlation between the daily rise and fall of the settlement price of China's 5-year Treasury bond futures contract is studied. At the same time, the parameters of hybrid Copula model are selected and determined by particle swarm optimization algorithm based on inertial weight. This paper compares the fitting ability of mixed Copula function and traditional Copula function to the relationship between upper tail and lower tail of the futures contract. The research results show that the hybrid Copula function based on inertial weight particle swarm optimization is a better description of the relationship between the upper tail and lower tail of the futures contract. Non-linearity in the tail of the daily rise and fall in the settlement price of the futures contracts of the inter-Treasury bonds, On the basis of the asymmetric correlation characteristics, and according to the measurement results, it can be judged that the upper tail correlation between the daily rise and fall of the settlement price is more obvious, and based on the grasp of this characteristic, the trading strategy which can produce excess returns is constructed.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F224;F724.5

【相似文献】