S4 class for the marginal settings

Specify which marginal models (individual_spec & default_specs) are fitted and how often they are refit as well as how big the training data set is. Remember that the forecasting is done in a rolling window fashion and the arguments (train and refit size) will have to match with the arguments of the also to be specified vine_settings.

Usage

marginal_settings(
  train_size,
  refit_size,
  individual_spec = list(),
  default_spec = default_garch_spec()
)

# S4 method for marginal_settings
show(object)

Arguments

train_size: equivalent to the slot definition below
refit_size: equivalent to the slot definition below
individual_spec: equivalent to the slot definition below
default_spec: equivalent to the slot definition below
object: An object of class marginal_settings

Value

Object of class marginal_settings

Details

For specifying the list for individual_spec or the argument default_spec the function default_garch_spec() might come in handy.

Functions

marginal_settings(): Class constructor taking the arguments specified in the slots below

Slots

train_size: Positive count specifying the training data size.
refit_size: Positive count specifying size of the forecasting window.
individual_spec: A named list. Specify ARMA-GARCH models for individual assets by naming the list entry as the asset and providing a rugarch::ugarchspec object.
default_spec: rugarch::ugarchspec object specifying the default marginal model (used if the marginal model is not specified through individual_spec)

Examples

# the most basic initialization
marginal_settings(train_size = 100, refit_size = 10)
#> An object of class <marginal_settings>
#> train_size: 100 
#> refit_size: 10 
#> No custom specifications.
# some individualism
marginal_settings(
  train_size = 100, refit_size = 10,
  individual_spec = list("GOOG" = default_garch_spec(ar = 3)),
  default_spec = default_garch_spec(dist = "norm")
)
#> An object of class <marginal_settings>
#> train_size: 100 
#> refit_size: 10 
#> Custom specifications were given for assets:
#>  GOOG