Abstract:This study utilizes the Generalized Auto-regressive Conditional Heteroskedasticity (GARCH) model to analyze the Shanghai Stock Exchange Composite Index and the Shenzhen Composite Index. It computes the returns and logarithmic returns, fits the GARCH model with a student-t distribution, and calculates the standardized residuals of the stock. Tests for normality and heteroskedasticity of residuals are conducted, follows by the creation of auto-correlation and partial auto-correlation plots to aid in volatility forecasting. The findings indicate that both returns of stock distributions exhibit multiple peaks and significant fluctuations, revealing that there exists pronounced volatility clustering and strong abrupt changes. Historical fluctuations significantly affect current fluctuations, showing a clustering effect. Volatility in returns has been very persistent, reflecting strong speculation and elevated overall risk. Histograms and skewness assessments show a left-skewed distribution for both indices, with a kurtosis greater than three, indicating a "fat-tailed" distribution. The positive mean values imply that China's economy is growing constantly, while the low standard deviation implies the immaturity of China's stock market. These insights highlight the dynamic behaviors and risk profiles in Chinas major stock indices, offering a deeper understanding of their speculative nature and economic implications. The Advantages and disadvantages of GARCH models and limitations of a single model are acknowledged in the report.

BRC-GARCH-X Model: the Empirical Evidence in Stock Returns

Forecasting Stock Volatility Using Time-Distance Weighting Fundamental’s Shocks

Prediction of Volatility Based on Realized-Garch-kernel-type Models: Evidence from China and the U.S.

Application of GARCH Model on the Analysis of Stock Exponential Rate of Return

A GARCH Model with Modified Grey Prediction Model for US Stock Return Volatility

Volatility analysis based on GARCH-type models: Evidence from the Chinese stock market

The ZD-GARCH Model: A New Way to Study Heteroscedasticity

Threshold network GARCH model

A Time-Varying Coefficient Double Threshold GARCH Model with Explanatory Variables

Stock Indices Analysis Based on ARMA-GARCH Model

A new GJR-GARCH model for DOUBLE-STRUCK CAPITAL Z-valued time series

Applying GARCH Model to Forecast Chinese Stock Volatility

Functional‐coefficient Regression Models with GARCH Errors

A long short-term memory enhanced realized conditional heteroskedasticity model

The Application and Modeling for Conditional Heteroscedasticity Time Series

Utilizing the GARCH Model for Analysis and Prediction of Stock Market Trends

A Research into Stock Market Volatility Using Threshold GARCH Model

An Empirical Research in the Stock Market of Shanghai by GARCH Model

Quantile Estimation of Regression Models with GARCH-X Errors

Sparse Multivariate GARCH

A New GJR‐GARCH Model for ℤ‐valued Time Series