Multi-scale Research On Financial Volatility Based On Symbolic Time Series Analysis

Financial volatility is the intrinsic properties of financial markets, which isclosely related to the functionality and stability of the financial markets. And it hasalso been a measure of market efficiency and indicators of the perfection degree of thedevelopment. The financial volatility means the uncertainty and risk in the market.Regardless of market investors or market regulators, it is of great significance to studythe characteristics of financial volatility and understand comprehensively andcorrectly, grasp the characteristics. The financial markets essentially is a nonlinearsystem, where many factors affect the volatility with complex changes. Fractal andchaos analysis and symbolic time series analysis, which are nonlinear system analysismethods are gradually introduced to the analysis of financial markets. It is hoped to bemore effectively and comprehensively to reveal the characteristics of financialvolatility. Numerous studies have found that there indeed exists multi-scalephenomenon in the financial markets. The results obtained by using a single timescale analysis of financial volatility tend to be one-sided. This paper is a study offinancial market volatility characteristics with the combination of multi-scale analysisand symbolic time series analysis method, to understand the characteristics offinancial volatility comprehensively and accurately.Firstly, this paper discusses the background and significance of the study on thevolatility in financial markets. It summarizes the financial volatility research, focuseson the basic theory of symbolic time series analysis methods and waveletmulti-resolution analysis to provide a theoretical basis for the thesis unfolds. Andwavelet multi-resolution analysis and symbolic time series method are applied to theanalysis of realized volatility calculated by Shanghai Composite Index and ShenzhenComponent Index series. They are broken into detail components with different scales.Using symbolic analysis of the original sequence and detail components, the symbolsequence histogram can identify the principal patterns and abnormal pattern ondifferent time scales. The results can provide the basis of the investment strategy andrisk management for different types of investors. This paper uses the rank orderdiagram of symbolic time series to study the differences and similarities betweendifferent sequences and the same sequence in different scales.The method of rank order diagram of symbolic time series shows superiority of symbolic time seriesanalysis, which reduces a lot of unnecessary trouble by simple calculation. Euclideannorm, and Statistics amount of2, relative entropy, and distance of rank order insymbolic time series statistics are quantitative analysis in the differences betweendifferent series on different scales. They can describe the differences with specificvalues.This paper is a part of the project "Study on Financial Volatility on SymbolicTime Series Analysis "(item number:70971097),which is supported by NationalNatural Science Foundation...