Forcasting and Modeling Stock Returns Volatility in Tehran Stock Exchange Using GARCH Models

The present study aim is modeling and comparing predictive power of GARCH models in forcasting stock returns volatility in Tehran Stock Exchange. Therefore was selected time period from 03/20/2009 to 03/19/2017 based daily returns of total price index (TEPIX) including 1900 observation and reviewed GARCH ، EGARCH ، PGARCH ، GJR ، GARCH-M ، FIGARCH and FIEGARCH models with time series approach and under the normal distribution assumption and evaluated Predictive performance of those models based Mean Squared Error (MSE), Median Squared Error (MedSE), Mean Absolute Error Statistic (MAE), Root Mean Square Error (RMSE) and Theil Inequality Coefficient (TIC) measures. The results showed the FIGARCH models have least error In terms of three criteria MSE, MAE and RMSE Which shows the Fitness of this model is better than all other models to predict the volatility of stock returns. It was also found that returns of the total stock price index (TEPIX) have cluster volatilities which means that with low fluctuations, have small volatility in subsequent periods and severe volatility with high fluctuations.