| Summary: | Abstract
Purpose
This paper focuses on comparing different models used in volatility forecasting and attempting to decide a “best” model for China’s security market.
Design/methodology/approach
Two investigated return series that are Shanghai Composite Index and Shenzhen Composite Index in China Security Market respectively covering 10 years period data from 01/01/2001 to 31/12/2010. Afterwards, using eight most popular models which are random walk (RW), historical mean model (HM), Simple Moving Average Model (MA), Exponential smoothing model (ES), Exponentially Weighted Moving Average (EWMA), GARCH (1, 1) along with its family models (i.e. Threshold-GARCH, Exponential-GARCH(1,1) forecast a seven months forecasting horizon at daily, weekly monthly frequency respectively for both series. Finally, accuracy of volatility forecasting of several of the methods are checked by Loss Functions, Diebold & Mariano, and Clarke-West.
Findings
Although MA model and EWMA perform well, unfortunately, no one can conclude a “best” volatility forecasting techniques consistently for both series cross different data frequency for China security market .Additional China security market presents leverage effect.
|
|---|