Page 1 of 1

SIMPLE BUT IMPORTANT DOUBT

Posted: Wed May 07, 2014 5:44 pm
by mingxiao
Hi to everybody,
I`m new in coding and EVIEWS, but I must write a work using it.
I must compare the volatility forecasting performance of ARMA(1,1) and GARCH(1,1) and its extensions. The data are daily return of a market index.
Well, I estimated the ARMA(1,1) and the GARCH(1,1) both based on the SQUARE DAILY RETURN calculated by EXCEL.
Then to forecast and get the values of forecasting evaluation criteria I SIMPLY used the button "FORECATS", between ESTIMATE and STATS.
Is doing so right?
I´ve seen the it´s possible to use a code to do that, but I don´t know how to write it. I could learn but is it necessary?
I mean, what I get by doing as I did?
Please help me, mine is a simple question for you but I don´t have the necessary background.
I don´t need detailed answers, it´s enough if you say to me "you are wrong, you should use the code", or you are doing well.
Thanks in advance

Re: SIMPLE BUT IMPORTANT DOUBT PLEASE HELP

Posted: Wed May 07, 2014 6:14 pm
by EViews Gareth
If you want to forecast, then yes, hitting the forecast button is fine. EViews has two ways to do everything, either through code, or with a mouse. Using the mouse is absolutely fine.

Re: SIMPLE BUT IMPORTANT DOUBT

Posted: Thu May 08, 2014 4:42 am
by mingxiao
If you want to forecast, then yes, hitting the forecast button is fine. EViews has two ways to do everything, either through code, or with a mouse. Using the mouse is absolutely fine.
Thanks very much, you have been very polite.

But making use the square daily returns for estimating both ARMA and GARCH is right?
I mean, obviously I can estimate GARCH from the time series of square daily returns (or from the time series of realized volatility) and forecast by it the future values of the time series, but I wonder whether in order to compare the two models I should do this or is more correct using the conditional variance that I can obtain by the GARCH estimated from daily returns.