RUGARCH variance targeting issue

[stoyan.stoyanov]

Hi again,

I am running a TGARCH(1,1) model on a series of adjusted daily stock returns and get significantly different forecasts when using variance targeting and when not. However, I feel that the forecast should not be that different. Could it be that the model is "targeting" the wrong variable (omega instead of unconditional variance)? Again it could well be a problem with my data or model, but I do not see anything obviously wrong with either. That's why I'm attaching the output under both scenarios, as well as my data. The relevant parts of the code are below. Hopefully you can help. I just don't think that my long term prediction of sigma should be going to 0...

Code:

spec=ugarchspec(variance.model = list(model = "fGARCH", garchOrder = c(1,1),
                        submodel = "TGARCH", external.regressors = regressors, variance.targeting = TRUE/FALSE),
                 mean.model = list (armaOrder = c(3,3), include.mean = mean, archm = FALSE,
                        archpow = 1, arfima = FALSE, external.regressors = NULL, archex = FALSE),
                 distribution.model = "std", start.pars = list(), fixed.pars = list())

fit=ugarchfit(spec, data, out.sample = 100, solver = "solnp", solver.control = list(),
               fit.control = list(stationarity = 1, fixed.se = 0, scale = 0))

forecast=ugarchforecast(fit, data = NULL, n.ahead = 100, n.roll = 50, out.sample = 0,
               external.forecasts = list(mregfor = NULL, vregfor = NULL))

boot.pred=ugarchboot(fit, data = NULL, method = "partial", n.ahead = 10,
                n.bootfit = 100, n.bootpred = 500, out.sample = 0, rseed = NA, solver = "solnp",
                solver.control = list(), fit.control = list(),
                external.forecasts = list(mregfor = NULL, vregfor = NULL), parallel = FALSE,
                parallel.control = list(pkg = c("multicore", "snowfall"), cores = 2))

P.S. I know that I probably don't need an ARMA(3,3).

Data + Ouput files:
var.targeting.conditional.forecast.jpeg
var.targeting.bootstrap.output.jpg.jpeg
no.var_targeting.conditional_forecast.jpeg
no.var.targeting.bootstrap.output.jpg
var.targeting.fit.txt
no.var.targeting.fit.txt

Thanks,
Stoyan

[stoyan.stoyanov]

Data file:
data.csv

[alexios]

Hi Stoyan and thanks for catching another one.

Bug quashed (rev.427)

You can also submit bug reports directly to the rugarch bug forum list:
https://r-forge.r-project.org/forum/forum.php?forum_id=1117

Regards,

Alexios

PS if you want the unconditional variance this can be correctly
recovered by using the "uncvariance" method on the fitted object.

On 27/06/2012 15:50, stoyan.stoyanov wrote:

> Hi again,
>
> I am running a TGARCH(1,1) model on a series of adjusted daily stock returns
> and get significantly different forecasts when using variance targeting and
> when not. However, I feel that the forecast should not be that different.
> Could it be that the model is "targeting" the wrong variable (omega instead
> of unconditional variance)? Again it could well be a problem with my data or
> model, but I do not see anything obviously wrong with either. That's why I'm
> attaching the output under both scenarios, as well as my data. The relevant
> parts of the code are below. Hopefully you can help. I just don't think that
> my long term prediction of sigma should be going to 0...
>
> *Code:*
>
> spec=ugarchspec(variance.model = list(model = "fGARCH", garchOrder = c(1,1),
>                          submodel = "TGARCH", external.regressors =
> regressors, variance.targeting = TRUE/FALSE),
>                   mean.model = list (armaOrder = c(3,3), include.mean = mean,
> archm = FALSE,
>                          archpow = 1, arfima = FALSE, external.regressors =
> NULL, archex = FALSE),
>                   distribution.model = "std", start.pars = list(), fixed.pars
> = list())
>
> fit=ugarchfit(spec, data, out.sample = 100, solver = "solnp", solver.control
> = list(),
>                 fit.control = list(stationarity = 1, fixed.se = 0, scale =
> 0))
>
> forecast=ugarchforecast(fit, data = NULL, n.ahead = 100, n.roll = 50,
> out.sample = 0,
>                 external.forecasts = list(mregfor = NULL, vregfor = NULL))
>
> boot.pred=ugarchboot(fit, data = NULL, method = "partial", n.ahead = 10,
>                  n.bootfit = 100, n.bootpred = 500, out.sample = 0, rseed =
> NA, solver = "solnp",
>                  solver.control = list(), fit.control = list(),
>                  external.forecasts = list(mregfor = NULL, vregfor = NULL),
> parallel = FALSE,
>                  parallel.control = list(pkg = c("multicore", "snowfall"),
> cores = 2))
>
> P.S. I know that I probably don't need an ARMA(3,3).
>
> *Data + Ouput files:*
> http://r.789695.n4.nabble.com/file/n4634631/var.targeting.conditional.forecast.jpeg
> var.targeting.conditional.forecast.jpeg
> http://r.789695.n4.nabble.com/file/n4634631/var.targeting.bootstrap.output.jpg.jpeg
> var.targeting.bootstrap.output.jpg.jpeg
> http://r.789695.n4.nabble.com/file/n4634631/no.var_targeting.conditional_forecast.jpeg
> no.var_targeting.conditional_forecast.jpeg
> http://r.789695.n4.nabble.com/file/n4634631/no.var.targeting.bootstrap.output.jpg
> no.var.targeting.bootstrap.output.jpg
> http://r.789695.n4.nabble.com/file/n4634631/var.targeting.fit.txt
> var.targeting.fit.txt
> http://r.789695.n4.nabble.com/file/n4634631/no.var.targeting.fit.txt
> no.var.targeting.fit.txt
>
> Thanks,
> Stoyan
>
> -----
> Stoyan Stoyanov
> The University of Chicago Booth School of Business
> MBA Class of 2013
> [phone] | [hidden email]
> --
> View this message in context: http://r.789695.n4.nabble.com/RUGARCH-variance-targeting-issue-tp4634631.html
> Sent from the Rmetrics mailing list archive at Nabble.com.
>
> _______________________________________________
> [hidden email] 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.
>

_______________________________________________
[hidden email] 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.

[stoyan.stoyanov]

No worries. Does this mean that I should just not use variance targeting until the revision is out? I am actually trying to build something quite applied using the package, so I would be very happy if I could get my hands on a recent revision. Is one available anywhere?

Thank you,
Stoyan

[Michael Weylandt]

It's on R-forge: googling r-forge and rugarch should get you there, but you might need to build from source to get the bleeding edge. It usually takes r-forge about a day to build, but it can be unpredictable. Details in another recent thread on this list.

Best,
Michael

On Jun 27, 2012, at 10:32 AM, "stoyan.stoyanov" <[hidden email]> wrote:

> No worries. Does this mean that I should just not use variance targeting
> until the revision is out? I am actually trying to build something quite
> applied using the package, so I would be very happy if I could get my hands
> on a recent revision. Is one available anywhere?
>
> Thank you,
> Stoyan
>
> -----
> Stoyan Stoyanov
> The University of Chicago Booth School of Business
> MBA Class of 2013
> [phone] | [hidden email]
> --
> View this message in context: http://r.789695.n4.nabble.com/RUGARCH-variance-targeting-issue-tp4634631p4634643.html
> Sent from the Rmetrics mailing list archive at Nabble.com.
>
> _______________________________________________
> [hidden email] 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.

_______________________________________________
[hidden email] 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.

[alexios]

You should really be able to checkout the latest version from the svn
repository and built it. If you don't know how and are in a hurry, email
me off list your O/S and I  will send you a pre-built version.

-Alexios

On 27/06/2012 16:32, stoyan.stoyanov wrote:

> No worries. Does this mean that I should just not use variance targeting
> until the revision is out? I am actually trying to build something quite
> applied using the package, so I would be very happy if I could get my hands
> on a recent revision. Is one available anywhere?
>
> Thank you,
> Stoyan
>
> -----
> Stoyan Stoyanov
> The University of Chicago Booth School of Business
> MBA Class of 2013
> [phone] | [hidden email]
> --
> View this message in context: http://r.789695.n4.nabble.com/RUGARCH-variance-targeting-issue-tp4634631p4634643.html
> Sent from the Rmetrics mailing list archive at Nabble.com.
>
> _______________________________________________
> [hidden email] 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.
>

_______________________________________________
[hidden email] 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.