Financial time series modeling and analysis

Author: Peng Zuoxiang
Publisher:
Publish Date: 2006-04-01
Features: The structure and main content of this book are as follows:
Chapter 1 introduces the problem statement, research approach, and organizational structure of the paper.
Chapter 2 conducts an economic analysis of investors' decision-making behavior in financial markets to explain the statistical characteristics of high-frequency financial time series, such as fat tails, volatility clustering, conditional variance heterogeneity, and long memory. This chapter also explores the reasons behind these well-known financial phenomena.
Chapter 3 employs extreme value theory to estimate and test the parameters of fat-tailed distributions in high-frequency financial time series, specifically the tail index, and discusses its application in risk management. Using extreme value theory and related knowledge, this chapter locally fits the distribution or density of returns to effectively estimate and predict value-at-risk, avoiding the invalidity of estimates and predictions caused by distorted overall fitting. In the modeling process of Chapter 3, a combined approach of methodological research and practical analysis is adopted.
Chapter 4 discusses the estimation of long-memory parameters in financial time series, focusing on models involving fractional integration parameters (ARFIMA), Gaussian semi-parametric methods, and GPH non-parametric estimation methods, and applies them to empirical analysis of long memory in the Shanghai and Shenzhen stock markets.
Chapter 5 covers unit root or stationarity tests in time series.
Chapter 6 systematically analyzes the ADF unit root test for financial time series with GARCH-error, representing a further deepening and innovation of the topics covered in Chapter 5. The empirical analysis in Chapter 6 suggests that the existence of pseudo-GARCH phenomena may be attributed to the in the arbitrary and unsystematic specification of GARCH models.