Year of Graduation
Analysis of Forecasting Power of GARCH Models
In the paper, there were tested 54 ARCH-models for their ability to forecast. The models were tested on the foreign exchange market data. It was evaluated the accuracy of the predicted values: the values predicted by the volatility and VaR were compared with the realized values. It was shown that a Threshold ARCH model had statistically more precise forecasting values and was the best model for forecasting on the foreign exchange market.