Rolling VaR estimates

[Stu]

.

[Stu]

。

[trubador]

Suppose you have a return series called rt. Then, for EWMA you can try:

Code: Select all

!alpha = 0.02 'you should provide an appropriate value for this parameter smooth(s,!alpha) rt^2 ewma

And as for GJR with t-distribution:

Code: Select all

arch(1,1,thrsh=1,tdist) rt c

Finally, to obtain the lower %1 quantile of a series y:

Code: Select all

scalar q1 = @quantile(y,0.01)

Please note that, all these commands (and many more) are available in the users guide.

[Stu]

.

[Stu]

I run the programs, but it keeps showing up error messages. And have tried to adpat the codes, but couldnt figure out what the problem is!!
Could someone please help, by perhaps even giving some hints at the problems? I very much appreciate your time!

Thanks.

[Stu]

。

[Stu]

。

[EViews Gareth]

Code: Select all

%smpl = @str(!i) + " 1250" scalar x = @elem(ewma, %smpl)

[Stu]

Gareth,

When i run it, it shows up 'Illegal date 1 1250 in “scalar x = @elem(ewma, "1 1250")"

What could be the problem here?

[EViews Gareth]

Sorry, I posted without thinking.

I don't understand what you're trying to do with that line. @elem takes a single date, not a range of dates.

[Stu]

。

[Stu]

。

[EViews Gareth]

Code: Select all

scalar x = @elem(ewma, "1251")

[Stu]

.

[trubador]

I tried to modify your code so as to perform what you want and to give you the idea for further adjustments. However, I suggest you to go over An Introduction to Eviews Programming before moving forward.

Code: Select all

!window = 1250 !length = @obsrange !alpha = 1-0.94 '1-alpha is the coefficient of estimated variance !scale = 10000 'it might be a good idea to adjust the scale series x 'create the output series beforehand matrix(!window,!length) results 'you can create a matrix to store all the resulting smoothed series ' move sample !step obs at a time for !i = 1 to !length-!window smpl @first+!i-1 @first+!i+!window-2 smooth(s,!alpha) !scale*returns^2 ewma stom(ewma,ewmat) 'convert series to a vector colplace(results,ewmat,!i) 'place the vector into the matrix x(!window+!i) = ewma(!window+!i) next smpl @all