Forecasting volatility for the stock market: a new hybrid model

Abstract

This study presents a new hybrid model that combines the grey forecasting model with the GARCH to improve the variance forecasting ability in variance as compared to the traditional GARCH. A range-based measure of ex post volatility is employed as a proxy for the unobservable volatility process in evaluating the forecasting ability due to true underlying volatility process not being observed. Overall, the results show that the new hybrid model can enhance the volatility forecasting ability of the traditional GARCH.

Keywords:

• forecasting model
• GARCH
• grey forecasting model
• volatility

2000 AMS Subject Classification :

• 68U01
• 68T01
• 91B02
• 91B28
• 91B84

Notes

The selecting criterion of δ is to produce the smallest forecast error rate 31. We choose δ to be 0.5 in our research.

The estimator in Parkinson 26 provides an efficient unbiased estimate of the volatility parameter from a geometric Brownian motion, but may be biased for other stochastic processes.

This evaluation measurement of variance forecasts is also adopted by Maheu and McCurdy 22.