Deperately seeking rolling forecast help

[EoinM]

Hi,

I am having some trouble with rolling volatility forecasts.

I have a file containing 5053 observations and I have estimated volatility using the EGARCH module included in eviews over the first half of the data. Then I am trying to use that equation (eq01) to do rolling forecasts of varying lengths to the end of the data. I have attached the code I have peaced together from the forum and the sample files. I runs but the output does not appear correct. If anyone can help I would be very grateful

Thanks

E

[EViews Esther]

From your program, I could not clearly figure out what you are trying to do. Can you answer the following questions: do you want to obtain 1-period-ahead forecast values? Or, do you want to store coefficient estimates? The following code will do 1-period ahead forecast and save the forecast values in the series fcast and store the coefficient estimates in coefmat.

Code: Select all

' set window size !window = 2500 ' get size of workfile !length = @obsrange ' declare equation for estimation equation eq01 %cmnd = _this.@command 'EDIT: get the equation specification ' set step size !step = 5 'calculate number of rolls !nrolls = @floor((!length-!window)/!step) 'matrix to store coefficient estimates matrix(6,!nrolls) coefmat 'where the number of coefficients is 6. series fcast 'series to store forecast estimates series fcastse series fcastvar %start = "@first" %end = "@last" !j = 0 ' move sample !step obs at a time for !i = 1 to !length-!window+1-!step step !step !j = !j +1 %first = @otod(@dtoo(%start)+!i-1) %last = @otod(@dtoo(%start)+!i+!window-2) smpl {%first} {%last} 'estimate equation - note I'm error catching here, any errors mean we'll keep going. !maxerr = @maxerrcount setmaxerrs !maxerr+1 _this.{%cmnd} 'estimate equation if @lasterrnum>0 then clearerrs else colplace(coefmat,eq01.@coefs,!j) 'store coefficients ' 1-period-ahead forecast %pers = @otod(@dtoo(%start)+!i+!window-1) 'start point %pere = @otod(@dtoo(%start)+!i+!window-1) 'end point if @dtoo(%end) < @dtoo(%pere) then 'check whether the forecast end point is greater than the workfile end point exitloop endif ' set smpl for forecasting period smpl {%pers} {%pere} eq01.forecast(f=na) audf1 aud_se aud_var ' store forecasts vars fcast = audf1 fcastse = aud_se fcastvar = aud_var endif next smpl @all

[EoinM]

Hi Esther,

Thanks a million for your response. I'm trying to get the forecasts and the coefficients.

Also, in the code where it says get the equation specification what should I be writing here?

Thanks

Eoin

[EViews Esther]

The 4th line lets you specify your equation specification, i.e.,

Code: Select all

%cmnd = _this.@command 'EDIT: get the equation specification

Please note that when you open your equation, "_this" command will recognize that you are interested in the "equation" object.

[EoinM]

Thanks Esther,

I am away from my PC with eviews on it at the moment but I will try this within the hour or two

Just to clarify, if I want to specify the equation EQ01, I merely open the equation and this command will run the forecasts based on the equation or do I need to edit the line. Apologies I'm a complete novice at programming and probably asking stupid questions. Again, I greatly appreciate your help

%cmnd = _this.@command 'EDIT: get the equation specification

[EViews Esther]

Yes, all you need to do is to open the equation of interest.

[EoinM]

Esther,

You are a lifesaver. Thank you for all your help!! It's working perfectly

Eoin

[Mo']

Hello Esther,

First of all, I want to thank everyone at the Eviews team helping out on this forum.

To the main problem, I am trying to follow the posts and your suggestions in a bid to reproduce Eoin's results but I find that even after opening the eq01 on the workfile before running he code you provided, I get an error message saying :

%CMND is not defined or is an illegal command in ''CMND = _THIS.@COMMAND''

Perhaps, I'm not getting something right.

Also, I'd like to find out if this code can work for other GARCH-type models on Eviews, I am interested especially in TGARCH and GARCH-M models.

Any suggestions would be very much appreciated.

Mo'

[EViews Esther]

Which version of EViews are you using? If you do not have EViews7, the program will not work.

If you have EViews7, then please upload your workfile and program.

[Mo']

Dear Esther,

I have been able to confirm that the code works for Eoin's data using Eviews7. Thank you.
I would like to find out if this code can work the same way(by opening the equation on Eviews7) for a different equation, say a GARCH,TGARCH or GARCH-M?

Regards,

Mo'

[EViews Esther]

Yes, it will work.

[Mo']

Thank you. I'd get on with it.

[Mo']

Esther,

thanks for your assitance thus far, I have been able to confirm that this code works, as I have compared it with simpler codes and obtained (almost) the same results for the one-period ahead static forecasts(changing to .fit,even though results look the same).
I'm a bit confused as to how to do the 5-period and 10-period ahead forecasts (i.e., forecasts for the 5-day and 10-day horizons as I am using a set of daily data). I am pretty sure it has to do with editing this part of the code(pasted below) but I am not sure about what do do. Please help.

Code: Select all

' 1-period-ahead forecast %pers = @otod(@dtoo(%start)+!i+!window-1) 'start point %pere = @otod(@dtoo(%start)+!i+!window-1) 'end point

Thanks.

[EViews Esther]

...how to do the 5-period and 10-period ahead forecasts

Code: Select all

' 5-period-ahead forecast %pers = @otod(@dtoo(%start)+!i+!window-1) 'start point %pere = @otod(@dtoo(%start)+!i+!window+4) 'end point ' 10-period-ahead forecast %pers = @otod(@dtoo(%start)+!i+!window-1) 'start point %pere = @otod(@dtoo(%start)+!i+!window+9) 'end point

[Mo']

Dear Eviews Esther,

I have tried to run a 5_step ahead forecast using the hint you provided and some further adjustments. But each time I try to run the code, I get an error message: Syntax error ''IF @ LAST < THEN''...I really have no idea why this keeps coming up, you've been extremely helpful, thanks in advance.

Code: Select all

' set window size !window =800 ' get size of workfile !length = @obsrange ' declare equation for estimation equation eq01 %cmnd = _this.@command 'EDIT: get the equation specification ' set step size !step = 1 'calculate number of rolls !nrolls = @floor((!length-!window)/!step) 'matrix to store coefficient estimates matrix(4,!nrolls) coefmat 'where the number of coefficients is 4. series fcast 'series to store forecast estimates series fcastse ' series fcastvar %start = "@first" %end = "@last" 'variable keeping track of how many rolls we've done !j = 0 ' move sample !step obs at a time for !i = 1 to !length-!window+1-!step step !step !j = !j +1 %first = @otod(@dtoo(%start)+!i-1) %last = @otod(@dtoo(%start)+!i+!window-2) smpl {%first} {%last} 'estimate equation - Error catching , any errors mean we'll keep going. !maxerr = @maxerrcount setmaxerrs !maxerr+1 _this.{%cmnd} 'estimate equation if @lasterrnum>0 then clearerrs else colplace(coefmat,eq01.@coefs,!j) 'store coefficients if {%end} < {%5pere} then 'check whether the forecast end point is greater than the workfile end point return endif ' set smpl for forecasting period smpl {%5pers} {%5pere} eq01.fit(f=na) r_f1 r_se r_var ' store forecasts vars fcast = r_f1 fcastse = r_se fcastvar = r_var endif next smpl @all