Page 1 of 1
RMSE with mv-garch
Posted: Sun Apr 05, 2015 8:22 am
by BNK_BNK
Hi,
I am a new joiner of RATS ,
I'm using the multivariate GARCH , EGARCH , GJR-GARCH to measure the volatility spillover on index and index futures.
After I got the result from these three models, I try to compare them by using RMSE , then I try to use @uforeerrors but it doesnt work.
So what should I do to get RMSE of each model.
thx,
Bnk
Re: RMSE with mv-garch
Posted: Sun Apr 05, 2015 12:06 pm
by TomDoan
RMSE of what? You can't really compute an RMSE for volatility because the volatility isn't observable.
Re: RMSE with mv-garch
Posted: Mon Apr 06, 2015 2:40 am
by BNK_BNK
Hi,
according to this attachment file in page92, they try to find out the best fit model
by comparing RMSE of different type of GARCH family that I try to replicate this method .
Thx,
Re: RMSE with mv-garch
Posted: Mon Apr 06, 2015 9:05 am
by TomDoan
From what I can tell, the RMSE's being computed are for the returns themselves. They use an AR(1) model for the mean of the returns and (for that part of the analysis) are comparing the forecasts using the different estimates for the AR(1) that come out of the different GARCH models. Note that for that analysis, the actual volatility estimates aren't important---it's just what, one hopes, is the improved efficiency of the estimates of the AR(1) coefficients by taking into account the heteroscedasticity. I don't know if that was what you were expecting. (Based upon the title of the paper, it's not what I would have guessed).