Comparison of forecasting precision with ARCH &G ARCH models by using the prices of Iraqi crude oil export values

Abstract

Many time series data are characterized by a large number of random changes, which makes them suffer from heteroscedasticity, where classical analyzes require homogeneity of variance, us research is interested  in to study of time series that suffer from Volatility in the form of periods of high and low Which is concerned with the element of risk or uncertainty in the market because it's  importance in the subject of building   dynamics models of time series , and then the use of those models in forecasting the values of the phenomenon studied, In this research were used  the conditional autoregressive models of heteroscedasticity when the errors follow a normal distribution that takes into account the Volatility that occur in prices, it began with the diagnosis stage using the ARCH effect test to detect The problem of heteroscedasticity in the remainder of the estimated model, followed by the estimation stage, then examining the suitability of the model  by using several measures for precision, and then the forecasting process. for using the values of the prices of Iraqi crude oil exports for a series of (136) observations for the period between the period 1/1/2011 to 4/1/2022.  It appears that the best model GARCH(1,1) because it achieved the lowest values of the precision measures