基于符号序列分析的股市网络结构及金融波动研究

发布时间：2018-04-16 03:12

本文选题：符号时间序列分析 + 最小生成树　；参考：《天津大学》2014年硕士论文

【摘要】：随着国家经济的发展，中国与世界各国的经济联系越来越密切，同时全球金融市场的波动对我国经济的影响越来越大。采取科学的方法对股票市场的结构以及金融市场收益与波动进行分析，对于认识与分散金融风险具有重要的意义。在金融市场风险的度量与预测方面，金融计量学方法已经取得了丰硕的研究成果，但是金融系统是一个典型的非线性系统，金融计量学的方法很难全面而又准确地把握波动的本质。因此需要有新的方法和视角来认识与研究金融市场结构与金融波动等有关问题。作为金融计量学方法的补充，本文试图将符号时间序列分析法引入到非线性金融系统的结构与波动分析中。本文将符号时间序列分析法引入到股市网络结构分析中，，基于交易量和收益两个变量建立符号序列，通过符号序列编码序列之间的欧式距离建立距离矩阵，从而得到最小生成树和分层树。结合分层树和最小生成树的特点对网络结构进行分析,文中以沪深300指数成分股进行了实证分析。在金融收益与波动预测方面，引入生物信息学中的序列比对方法与符号时间序列分析方法相结合，提出一种新的预测方法。通过数据的符号化、选取合适的模式长度，基于符号比对的方式在历史符号序列中搜索与当前模式相似度最高的历史模式，然后将此历史模式用于下一个值的预测。文中以上证综指高频数据为样本，对收益以及波动进行了实证分析。本文第一章阐述了文章研究背景、意义以及涉及到的研究方法的研究现状，梳理了文章结构并且提出了本文的创新点。第二章概述了文中用到的理论基础，包括时间序列符号化、网络结构分析法、基于模式匹配的预测方法。第三章提出了基于符号时间序列分析法的股票市场网络结构分析，并以沪深300指数成分股为例进行了实证分析。第四章提出基于模式匹配的金融收益与波动的预测方法，并对文中用到的基于序列比对的相似性度量进行了详细描述，且以上证综指采样间隔为20分钟的高频数据为样本进行了实证分析。第五章总结了本文的工作，以及指出对未来研究中需要进一步改进的方向。
[Abstract]:With the development of national economy, the economic relationship between China and other countries is getting closer and closer. At the same time, the fluctuation of global financial market has more and more influence on China's economy.It is of great significance to analyze the structure of the stock market and the return and fluctuation of the financial market by adopting scientific methods to understand and disperse the financial risks.In the aspect of measurement and prediction of financial market risk, financial metrology has made great achievements, but the financial system is a typical nonlinear system.The method of financial metrology is difficult to grasp the essence of volatility comprehensively and accurately.Therefore, new methods and perspectives are needed to understand and study financial market structure and financial volatility.As a supplement of financial metrology, this paper attempts to introduce symbolic time series analysis into the structural and volatility analysis of nonlinear financial systems.In this paper, the symbolic time series analysis method is introduced into the stock market network structure analysis. The symbol sequence is established based on the two variables of trading volume and income, and the distance matrix is established by the Euclidean distance between the symbol sequence coding sequences.Thus, the minimum spanning tree and the hierarchical tree are obtained.Combined with the characteristics of hierarchical tree and minimum spanning tree, the network structure is analyzed, and the empirical analysis is carried out with the index of Shanghai and Shenzhen 300.In the aspect of forecasting financial returns and volatility, a new forecasting method is proposed by combining the methods of sequence alignment and symbolic time series analysis in bioinformatics.Through the symbolization of the data, the appropriate pattern length is selected, and the historical pattern with the highest similarity to the current pattern is searched in the historical symbol sequence based on symbol alignment, and then the historical pattern is used to predict the next value.Taking the high frequency data of Shanghai Composite Index as the sample, the paper makes an empirical analysis on the return and volatility.The first chapter describes the background, significance and research status of the research methods involved, combs the structure of the article and puts forward the innovative points of this paper.The second chapter summarizes the theoretical basis used in this paper, including time series symbolization, network structure analysis, pattern matching based prediction methods.In chapter 3, the network structure analysis of stock market based on symbolic time series analysis is proposed, and the empirical analysis of CSI 300 index component stock is given.In chapter 4, a method of forecasting financial returns and volatility based on pattern matching is proposed, and the similarity measures based on sequence alignment are described in detail.Based on the high frequency data with sampling interval of 20 minutes in Shanghai Composite Index, the empirical analysis is carried out.The fifth chapter summarizes the work of this paper and points out the direction of further improvement in the future research.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51

【参考文献】