Comparison of predicting volatility of Tehran stock index in GARCH-MIDAS approach and quantile regression

This research is carried out to the GARCH-MIDAS model which is used with the aim of compensating for the shortcoming of conventional GARCH models; i.e., relying on symmetry in data frequency. Therefore, the advantage of GARCH-MIDAS model to GARCH models and of course other time series models is the combination of data that have different frequencies. For this purpose, stock returns are modeled based on a combination of daily and weekly volatility. Besides, the Quantile model is also one of the new models that focuses on the entire distribution instead of different frequencies, thereby does regression based on the distribution of the entire data and is not based on the characteristic of the normal distribution. The problem of the current research was formed from this difference between Garch-Midas and Quantile model, and the organization of the research was formed based on it. After describing the problem and assumptions in the first chapter, a review of the theoretical and empirical literature of the research was carried out, and in the third and fourth chapters, the research model, its description and regression were estimated. The findings of the research showed that the Garch-Midas model has a better fit than the quantile model and has a better modeling and forecast capability for the fluctuation in stock returns.