FORECASTING VOLATILITY OF SGARCH(1,1) MODEL

• Main Author: Permata Sari, Dian
• Format: Theses
• Language: Indonesia
• Subjects: Matematika
• Online Access: https://digilib.itb.ac.id/gdl/view/33606
• Institution: Institut Teknologi Bandung

Forecasting volatility is an important aspects in financial markets because it can be used for risk management and asset allocation. There are various time series models that can be used to forecast volatility. The model that used in this thesis is Stochastic Generalized Autoregressive Conditional Heteroscedastic Order (1,1) model. The volatility in SGARCH(1,1) model can be expressed as a function of the observed and unobserved components. This thesis discusses about the ability of SGARCH(1,1) model to capture the empirical properties of returns such that can be used to predict the volatility in the future. This thesis will also discuss about the comparison of GARCH(1,1) and SGARCH(1,1) model. Data that used in this thesis are the stock returns of Google, IBM, Indofood, S&P 500 and Hangseng from 01/04/2010 until 12/31/2013. The results show that the empirical properties of the fifth stock returns are more adequately captured by SGARCH(1,1) than GARCH (1,1) model. Five step ahead volatility prediction of GOOGLE, Indofood, S&P 500 and Hangseng using SGARCH(1,1) is more accurate than GARCH(1,1) model.