Page 1 of 1

Forecast Evaluation for GARCH-type models

Posted: Thu Aug 03, 2017 8:21 am
by Alecs
I want to forecast variance and I do not understand how to obtain the RMSE, MAE, MAPE for the variance.

Step-by-step approach:
1. Use Log Returns
2. Fit ARIMA using AIC
3. Add GARCH(1,1) and estimate coefficients in sample.
4. Proc -> Forecast, select my out-of-sample period and static forecasting
5. I get the forecast for the mean equation and the table with the loss functions, but that is for the mean equation, right?

How do I get it just for the GARCH?

I have also tried:
1. Fit ARMA -> save residuals
2. alt+q+e mean equation "residuals c" Method: ARCH and select a GARCH(1,1)
3. same thing with the sample sizes
4. and again I get the error measurements for the mean (which is a flat line now)

Also, do I need include "residuals c" in the mean equation or should I just put "residuals"?

Thank you!