GARCH (1.1)

Eigminas

New Member
Hi,

I have a question. When calculating volatility using GARCH (1.1) using for instance 1260 (daily returns) observations in excel, according P. Christoffersen the oldest volatility in the sample is equal to simple standard deviation of all 1260 observations. Then all the next days volatility is calculated using GARCH (1.1) model with alfa, beta and omega parameters. The likelihood is estimated and using SOLVER in excel the likelihood is maximized (maximum likelihood estimation). Then SOLVER changes the parameters and likelihood too and according to that the volatility is adjusted.
My question is whether I have to use alfa, beta and omega parameters obtained from 1260 observations when calculating the new volatility when the most recent new return is added and the oldest is removed from the sample or I have to use SOLVER again when sample is updated with a new daily return? The thing is that I use 1260 returns to calculate volatility using GARCH (1.1) and to calculate VaR. How should I calculate the volatility updating the sample with new daily return?

Thanks,
 
Top