基于时序图模型结构估计的股票联动研究

发布时间：2018-03-11 22:21

本文选题：图模型　切入点：时序SPACE算法　出处：《数理统计与管理》2012年05期 　论文类型：期刊论文

【摘要】：我国股市个股价格同时上涨或同时下跌的联动现象极为普遍,传统上使用向量自回归、协整、有向非循环图等方法主要用于少量股票或市场之间的联动性研究,不适于直接对大规模个股之间的联动关系进行研究。文章关注大规模时序图模型结构建立及估计方法,通过将ADL方法引入SPACE算法,提出了可以估计高维低样时序图模型的ADL-SPACE算法;设计模拟实验考察了算法中惩罚参数λ值的设置对于节点自回归相关性捕获的有效性;在实证研究中,文章使用了ADL-SPACE算法对个股联动研究了三方面的内容:1.基于个股联动的代表性行业之间的联动性;2.设计了我国A股市场中行业联动强度,对行业内外联动性进行综合评价和分析;3.采用一阶滞后个股基于时序图模型结果构造了投资组合,模拟显示收益预期表现良好。以上研究均表明时序SPACE图模型方法在大规模股票的联动探测中有较好的应用前景。
[Abstract]:It is very common for individual stock price to rise or fall at the same time in Chinese stock market. Traditionally, the methods of vector autoregression, cointegration and directed acyclic graph are mainly used to study the linkage between a small number of stocks or markets. It is not suitable to study the linkage relationship between large scale stocks directly. This paper focuses on the method of establishing and estimating the structure of large-scale sequential diagram model, and introduces the ADL method into the SPACE algorithm. A ADL-SPACE algorithm is proposed to estimate the high-dimensional and low-sample sequence diagram model. Simulation experiments are designed to investigate the effectiveness of setting the penalty parameter 位 in the algorithm for node autoregressive correlation acquisition. In this paper, ADL-SPACE algorithm is used to study the linkage of individual shares in three aspects: 1.Based on the linkage between representative industries. 2. The intensity of industry linkage in A share market of China is designed. A comprehensive evaluation and analysis of the interaction between the industry and the outside is carried out. The first order lag stock is used to construct the investment portfolio based on the results of the timing chart model. The simulation results show that the expected return performance is good. All the above studies show that the time-series SPACE model method has a good application prospect in the large-scale stock linkage detection.
【作者单位】：中国人民大学应用统计科学研究中心&中国人民大学统计学院;
【基金】：中国人民大学应用统计科学研究中心2009年重大项目(08jjd910248) 中国人民大学科学研究基金(中央高校基本科研业务费专项资金资助)项目(10XNI014)
【分类号】：F832.51;O211.6

【参考文献】