Try using model="VAR" instead of "AR"....will check to see if something
is amiss otherwise.
A.
On 24/03/2017 13:42, Josh Segal wrote:
Alexios,
Thanks again for your help.
I'm getting some counterintuitive results with the seriesSim output.
When I run your exact example above and then compute
cor(sim at msim$seriesSim[[1]]), I get a correlation matrix that has
non-diagonal values close to zero (ranging from -0.10 to 0.08).
When I measure the correlation of the original data
(cor(dji30ret[,1:5])) I get values from 0.33 to 0.66.
Shouldn't the simulation demonstrate higher unconditional
correlations? Am I misunderstanding something about the package, or
does this indicate a problem?
Thanks,
Josh
On Wed, Mar 22, 2017 at 4:06 PM, alexios galanos <alexios at 4dscape.com
<mailto:alexios at 4dscape.com>> wrote:
library(rmgarch)
data("dji30ret")
spec<-gogarchspec(mean.model=list(model="AR"),ica = "radical")
fit=gogarchfit(spec,dji30ret[,1:5])
sim=gogarchsim(fit,n.sim = 1000,m.sim=1,startMethod =
"sample",rseed = 10)
head(sim at msim$seriesSim[[1]])
str(sim at msim)
There are lots of examples and demos in the /inst/rmgarch.tests/
folder of the source package.
Alexios
On 3/22/2017 2:54 PM, Josh Segal wrote:
Hi everyone,
I'm trying to use the rmgarch package to estimate a
multivariate GARCH
model and then use those parameters to simulate paths
forward. I've gotten
as far as creating a goGARCHsim object (for example), but
can't figure out
how to access the simulated returns. I've looked through all
the methods
described in the documentation (page 58) but don't see
anything relevant.
I believe I am able to do this in the univariate case with
rugarch - is it
not possible in rmgarch?
Thanks for your help!
[[alternative HTML version deleted]]