GARCH Prediction not possible

Hello everyone,

I have a very small problem that is bothering me in my GARCH estimation/prediction: The estimation part works perfectly, however I encounter a problem when I want to predict/forecast the conditional variance for the next period. Here is part of my code:

Mdl = garch(1,1);
opts = optimset('fmincon');
opts.Algorithm = 'interior-point';
EstMdl = estimate(Mdl,return,'options',opts);
Vf1 = forecast(Mdl,1,'Y0',return);

Now, in the last line of code, I run into the problem... The error message is as follows:

Conditional variance constant must be specified.

    Error in Code (line 774)
    Vf1 = forecast(Mdl,1,'Y0',return);

But I cannot understand why this problem appears... The returns consist of a vector (60,1) and everything should normally be perfectly fine..

Thank you guys very much already! I appreciate your help a lot!

Answer 1

[Roger Wohlwend]

You invoke the function forecast with Mdl, which is just a model specification and thus contains no coefficients. You have to use EstMdl instead. That's the estimated model with the coefficients.

Vf1 = forecast(EstMdl,1,'Y0',return);

[Clemens Mueller]

Oh.. Yes, I agree this was a rather stupid mistake. I apologize, but thank you very much for pointing this out to me