Here's a reproducible example on how you can pass starting parameters
and bounds:
library(rugarch)
library(xts)
data("dmbp")
dmbp <- as.xts(dmbp, as.Date(1:nrow(dmbp)))
spec <- ugarchspec(mean.model = list(armaOrder = c(0,0), include.mean =
TRUE),
variance.model = list(model = "sGARCH", garchOrder =
c(1,1), external.regressors = dmbp[,2]))
setstart(spec) <- list("vxreg1" = 0.5)
setbounds(spec) <- list("vxreg1" = c(0, 1))
spec@model$start.pars
mod <- ugarchfit(spec, dmbp[,1])
I suspect this may also be related to a parameter scaling issue as well.
I'm working on a new implementation which should hopefully fix
this and many other issues (autodiff gradients/hessian, option for
multiplicative regressors, parameter scaling. standard error calculation
options etc), but won't have something out for at least a few months (it
will eventually be located in this repo: github.com/tsmodels/).
Regards,
Alexios
On 1/11/23 12:31 PM, Simon van Norden wrote:
Having trouble with bad initial parameter estimates in rugarch when
using an external regressor.
* Convergence code is 0 and convergence is rapid, but….
* Coefficient estimate on the external regressor is always 0.00000
and condH = NaN
* Results are robust to variable scaling and optimization algorithm
(tried them all.)
My hunch is that I’m on a saddle point (or similar ‘flat spot’.)
I’d like to try alternative initial parameter estimates for the
external regressor; can anyone point me to an example showing how to
pass this to the optimizer?
Would to happy to hear of other suggestions, or provide details, a
reprex, etc.
FWIW, this may be related to the question someone else posted back on
20-06-2022.
Thnx
*Simon van Norden*
*Professeur Titulaire*
Finance
3000, chemin de la Côte‑Sainte‑Catherine,
Montréal (Québec) H3T 2A7
*Telephone: 514 340 6781*
HEC Montréal <http://www.hec.ca/>
Agréments <http://www.hec.ca/decouvrez/choisir_hec/>
*hec.ca* <http://www.hec.ca/>
_______________________________________________
R-SIG-Finance@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should
go.
_______________________________________________
R-SIG-Finance@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should
go.
