uGARCHfit-class {rugarch} | R Documentation |
class: Univariate GARCH Fit Class
Description
Class for the univariate GARCH fit.
Objects from the Class
A virtual Class: No objects may be created from it.
Extends
Class GARCHfit
, directly.
Class rGARCH
, by class GARCHfit
, distance 2.
Slots
fit
:Object of class
"vector"
Holds data on the fitted model.model
:Object of class
"vector"
The model specification common to all objects.
Methods
- coef
signature(object = "uGARCHfit")
: Extracts the coefficients.- cofint
signature(object = "uGARCHfit")
: Similar to the stats S3 methodconfint
, extracts coefficient confidence intervals taking additional optional argumentsparm
andlevel
, as well asrobust
(default: FALSE) indicating whether to use the robust covariance matrix for the calculations.- vcov
signature(object = "uGARCHfit")
: Extracts the covariance matrix of the parameters. Additional logical option of ‘robust’ indicates whether to extract the robust based covariance matrix.- infocriteria
signature(object = "uGARCHfit")
: Calculates and returns various information criteria.- nyblom
signature(object = "uGARCHfit")
: Calculates and returns the Hansen-Nyblom stability test (1990).- gof
signature(object = "uGARCHfit", groups = "numeric")
: Calculates and returns the adjusted goodness of fit statistic and p-values for the fitted distribution based on the Vlaar and Palm paper (1993). Groups is a numeric vector of bin sizes.- newsimpact
signature(object = "uGARCHfit")
: Calculates and returns the news impact curve.- signbias
signature(object = "uGARCHfit")
: Calculates and returns the sign bias test of Engle and Ng (1993).- likelihood
signature(object = "uGARCHfit")
: Extracts the likelihood.- sigma
signature(object = "uGARCHfit")
: Extracts the conditional sigma values.- fitted
signature(object = "uGARCHfit")
: Extracts the fitted values.- residuals
signature(object = "uGARCHfit")
: Extracts the residuals. Optional logical argumentstandardize
(default is FALSE) allows to extract the standardized residuals.- getspec
signature(object = "uGARCHfit")
: Extracts and returns the GARCH specification from a fit object.- uncvariance
signature(object = "uGARCHfit", pars = "missing", distribution="missing", model = "missing", vexdata = "missing")
: Calculates and returns the long run unconditional variance of the GARCH fit given auGARCHfit
object.- uncvariance
signature(object = "missing", pars = "numeric", distribution = "character", model = "character", submodel = "ANY", vexdata = "ANY")
: Calculates and returns the long run unconditional variance of the GARCH fit given a named parameter vector as returned by the fit, a distribution model name and a GARCH model name with a submodel included if the model is of the nested type such as fGARCH and any external regressor data.- uncmean
signature(object = "uGARCHfit")
: Calculates and returns the unconditional mean of the conditional mean equation (constant, ARMAX, arch-in-mean).- persistence
signature(object = "uGARCHfit", pars = "missing", distribution = "missing", model = "missing")
: Calculates and returns the persistence of the GARCH fit model given auGARCHfit
object.- persistence
signature(object = "missing", pars = "numeric", distribution = "character", model = "character")
: Calculates and returns the persistence of the GARCH fit model given a named parameter vector as returned by the fit, a distribution model name and a GARCH model name with a submodel included if the model is of the nested type such as fGARCH.- halflife
signature(object = "uGARCHfit", pars = "missing", distribution = "missing", model = "missing")
: Calculates and returns the halflife of the GARCH fit variance given auGARCHfit
object.- halflife
signature(object = "missing", pars = "numeric", distribution = "character", model = "character")
: Calculates and returns the halflife of the GARCH fit variance given a named parameter vector as returned by the fit, a distribution model name and a GARCH model name with a submodel included if the model is of the nested type such as fGARCH.- convergence
signature(object = "uGARCHfit")
: Returns the solver convergence code for the fitted object (zero denotes convergence).- quantile
signature(x = "uGARCHfit")
: Calculates and returns, given a vector of probabilities (additional argument “probs”), the conditional quantiles of the fitted object (x).- pit
signature(object = "uGARCHfit")
: Calculates and returns the conditional probability integral transform given the data and estimated density.- reduce
signature(object = "uGARCHfit")
: Zeros parameters (fixing to zero in rugarch is equivalent to eliminating them in estimation) with p-values (optional argument “pvalue”) greater than 0.1 (default), and re-estimates the model. Additional arguments are passed tougarchfit
.An additional option “use.robust” (default TRUE) asks whether to use the robust calculated p-values.- plot
signature(x = "uGARCHfit", y = "missing")
: Fit plots.- show
signature(object = "uGARCHfit")
: Fit summary.
Note
Methods for coef, likelihood, fitted, sigma and residuals provide extractor
functions for those values.
Method for show gives detailed summary of GARCH fit with various tests.
Method for plot provides for interactive choice of plots, option of choosing
a particular plot (option “which” equal to a valid plot number) or a
grand plot including all subplots on one page (option “which”=“all”).
The infocriteria method calculates and returns the information criteria
(AIC, BIC etc) of the GARCH fit.
The nyblom method calculates and returns the Hansen-Nyblom joint and individual
coefficient stability test statistic and critical values.
The gof methods calculates and returns the adjusted goodness of fit statistic
and p-values for the fitted distribution. The groups parameter is a numeric
vector of grouped bin sizes for the test. See the references in the package
introduction for the original paper by Vlaar and Palm explaining the test.
The signbias methods calculates and returns the sign bias test of Engle and Ng
(see the references in the package introduction).
Methods for calculating and extracting persistence, unconditional variance and
half-life of the GARCH shocks exist and take either the GARCH fit object as a
single value otherwise you may provide a named parameter vector (see
uGARCHspec
section for parameter names of the various GARCH
models), a distribution name and the GARCH model (with submodel argument for the
fGARCH model).
Unconditional mean and variance of the model may be extracted by means of the
uncmean and uncvariance methods. The uncvariance may take either a fit object or
a named parameter list, distribution and GARCH model name. The uncmean will only
take a fit object due to the complexity of the calculation requiring much more
information than the uncoditional variance.
The news impact method returns a list with the calculated values (zx, zy) and
the expression (xexpr, yexpr) which can be used to illustrate the plot.
Author(s)
Alexios Ghalanos
See Also
Classes uGARCHforecast
, uGARCHsim
and
uGARCHspec
.
Examples
## Not run:
# Basic GARCH(1,1) Spec
data(dmbp)
spec = ugarchspec()
fit = ugarchfit(data = dmbp[,1], spec = spec)
fit
# object fit:
slotNames(fit)
# sublist fit@fit
names(fit@fit)
coef(fit)
infocriteria(fit)
likelihood(fit)
nyblom(fit)
signbias(fit)
head(sigma(fit))
head(residuals(fit))
head(fitted(fit))
gof(fit,c(20,30,40,50))
uncmean(fit)
uncvariance(fit)
#plot(fit,which="all")
# news impact example
spec = ugarchspec(variance.model=list(model="apARCH"))
fit = ugarchfit(data = dmbp[,1], spec = spec)
# note that newsimpact does not require the residuals (z) as it
# will discover the relevant range to plot against by using the min/max
# of the fitted residuals.
ni=newsimpact(z = NULL, fit)
#plot(ni$zx, ni$zy, ylab=ni$yexpr, xlab=ni$xexpr, type="l", main = "News Impact Curve")
## End(Not run)