Modeling and Forecasting Distribution of Return on the Tehran Stock Exchange Index and Bitcoin with the GAS Time Variable Method

Predicting returns with the least error is one of the most important issues in financial markets that has been considered by many researchers in recent decades .Traditional linear and nonlinear models due to the inefficiency of linear models in market turbulence, the lack of correct extraction of the conditional distribution form of data due to the failure to record the conditional distribution dynamics in nonlinear models and the existence of limiting assumptions, it lacks the ability to predict returns in different market conditions. In order to eliminate the disadvantages of traditional models, in the present study using a new time-variable method called generalized autoregressive score (GAS) in order to predict the distribution of return of the total index of the stock exchange during the period 2010 to 2020 and for Bitcoin during the period 2014 to 2020. The results of modeling for the two assets by the new GAS model are compared with the results of the GARCH and AR models and their performance is tested for inside and outside the sample. The results show that in order to predict the daily return, the overall index of the new GAS model has a better performance and in order to predict the daily return of bitcoin, the GARCH model has been preferred.