An Empirical Investigation of Comparative Performance of GARCH Family Models and EWMA Model in Predicting Volatility of TSEC Weighted Index of Taiwan.

The act of modeling and forecasting stock market volatility has become essential to risk management practice; it has emerged as one of the most popular themes in financial econometrics and has been mainly and constantly used in the valuation of financial assets and the Value at Risk, as well as the pricing of options and derivatives, are all part of the process. The purpose of this paper is to contrast the GARCH (Generalized Auto Regressive Conditional Heteroscedasticity) family models -- GARCH, PGARCH, EGARCH, GJR-GARCH, MGARCH, and IGARCH -- with the EWMA (Exponentially Weighted Moving Average) models in the prospect of establishing the best algorithm to forecast TSEC volatility weighted index from Taiwan stock market. We employ daily returns for the range between July 1997 and July 2023. The sample data spans a long time period i.e. 26 years of data with 6375 observations, which includes dramatic events such as the financial crisis of 2008, the COVID-19 epidemic, and the war between Russia and Ukraine. We observe that the asymmetric model PGARCH with a Student's t distribution develops the best model for assessing the volatility of the chosen index, but PGARCH with a normal error distribution produces the best estimating performance outcomes and hence outperforms the EWMA model. The present empirical investigation also seeks to illustrate the potential for investment returns as well as the associated risk. Our findings may have implications for risk management in Taiwan, together with a deeper exploration of the TSEC weighted index from Taiwan stock market volatility dynamics, given the scarcity of prior such studies. [ABSTRACT FROM AUTHOR]