| ARFIMAroll-class {rugarch} | R Documentation |
Class for the ARFIMA rolling forecast.
roll:Object of class "vector"
forecast:Object of class "vector"
model:Object of class "vector"
Class "ARFIMA", directly.
Class "rGARCH", by class "ARFIMA", distance 2.
signature(object = "ARFIMAroll"):
extracts and converts the forecast object contained in the roll object to
one of ARFIMAforecast given the refit number supplied
by additional argument ‘refit’ (defaults to 1).
signature(x = "ARFIMAroll"): extracts various
values from object (see note).
signature(object = "ARFIMAroll"):
Forecast performance measures.
signature(object = "ARFIMAroll"): roll backtest reports
(see note).
The as.data.frame extractor method allows the extraction of a variety of
values from the object. Additional arguments are:
which indicates the type of value to return. Valid values are “coefs”
returning the parameter coefficients for all refits, “density” for the
parametric density, “coefmat” for the parameter coefficients with their
respective standard errors and t- and p- values, “LLH” for the likelihood
across the refits, and “VaR” for the Value At Risk measure if it was
requested in the roll function call.
n.ahead for the n.ahead forecast horizon to return if which was
used with arguments “density” or ‘VaR’.
refit indicates which refit window to return the “coefmat”
if that is chosen. If “series” is chosen under via the which
argument, then the forecast series is returned for a particular refit, else
when “all” is used it returns the complete forecasted series across all
refits.
The report method takes the following additional arguments:
type for the report type. Valid values are “VaR” for the Value at
Risk report based on the unconditional and conditional coverage tests for VaR
exceedances (discussed below) and “fpm” for forecast performance measures.
n.ahead for the rolling n.ahead forecasts (defaults to 1).
VaR.alpha for the Value at Risk backtest report, this is the tail
probability and defaults to 0.01.
conf.level the confidence level upon which the conditional coverage
hypothesis test will be based on (defaults to 0.95).
Kupiec's unconditional coverage test looks at whether the amount of expected
versus actual exceedances given the tail probability of VaR actually occur as
predicted, while the conditional coverage test of Christoffersen is a joint test
of the unconditional coverage and the independence of the exceedances. Both the
joint and the separate unconditional test are reported since it is always
possible that the joint test passes while failing either the independence or
unconditional coverage test.
The “fpm” does not take any additional arguments, but instead returns the
forecast performance measures for all “n.ahead” values.
Alexios Ghalanos