Modeling of Volatility in the Stock Markets Returns: Classic and Bayesian GARCH Approaches for ISE -100

The accuracy of estimate in the investment risk is important for the potential investors as well as the expected return from the investment characteristics. One of the most fundamental issues on risk and return is to have the heavy tailed behavior of the residuals that makes it difficult to obtain an appropriate risk prediction model for index returns in financial studies. In this study, Istanbul stock exchange (ISE-100) daily index data between February 2007 and February 2017 is analyzed with the classic and Bayesian GARCH (1,1) models and it is aimed to compare the effects of them on return volatility with Student-t residuals. As a result of this study, no significant differences are found between the classical and Bayesian GARCH (1,1)-Student-t models both effects of shock on volatility and volatility persistence for stock return. This can be interpreted as both models not well differentiated from each other. In conclusion, classical and Bayesian GARCH (1,1)-Student-t estimation methods provide reliable results both in modeling volatility of returns and in estimating investment risk for investors and market regulators.