Evaluating Volatility Forecasting Performance Measure with Generalized ARCH Models
Hemanth Kumar P.1, S. Basavaraj Patil2
1Hemanth Kumar P., PhD Scholar, Department of Computer Science Engineering, VTU RRC, Belgaum, Karnataka, India.
2Dr. S. Basavaraj Patil, Department of Computer Science, VTU RRC, Belgaum, Karnataka, India.
Manuscript received on October 22, 2017. | Revised Manuscript received on October 28, 2017. | Manuscript published on November 05, 2017. | PP: 18-23 | Volume-7 Issue-5, November 2017. | Retrieval Number: E3083117517/2017©BEIESP
Open Access | Ethics and Policies | Cite 
©The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/

Abstract: Volatility Forecasting is an interesting and challenging problem in current financial instruments. There are many financial risks and rewards directly associated with volatility. Hence forecasting volatility becomes the most discussed topic in finance. In this research we apply various univariate conditional heteroskedasticity models for forecasting volatility. The various extensions of the standard Generalized ARCH models such as SGarch, CSGarch, Egarch, IGarch and GJRGarch are used for forecasting. Volatility values are forecasted for 10 days in advance and values are compared with the actual values. Mean square error is computed between Garch forecast and actual values for the 10 days. The model with lowest MSE values over 10 forecasted periods is selected as the best performing model. The forecasted results for 10 days show that GJRGarch ranks top in the accuracy of forecasting and IGarch at the bottom. GJRGarch outperforms all other univariate ARCH models.