Class DisplacedLognormalGARCH

  • All Implemented Interfaces:
    HistoricalSimulationModel

    public class DisplacedLognormalGARCH
    extends Object
    implements HistoricalSimulationModel
    Displaced log-normal process with GARCH(1,1) volatility. This class estimate the process \[ \mathrm{d} \log(X + a) = \frac{\sigma}{b + a} \mathrm{d}W(t) \] where \( a > -min(X(t_{i}) \) and thus \( X+a > 0 \) and \( b = 1 - -min(X(t_{i}) \) \) and \( \sigma \) is given by a GARCH(1,1) process. The choice of b ensures that b+a ≥ 1. For a=0 we have a log-normal process with volatility σ/(b + a). For a=infinity we have a normal process with volatility σ.
    Version:
    1.0
    Author:
    Christian Fries
    • Constructor Detail

      • DisplacedLognormalGARCH

        public DisplacedLognormalGARCH​(double[] values)
      • DisplacedLognormalGARCH

        public DisplacedLognormalGARCH​(double[] values,
                                       double lowerBoundDisplacement)
      • DisplacedLognormalGARCH

        public DisplacedLognormalGARCH​(double[] values,
                                       int windowIndexStart,
                                       int windowIndexEnd)
      • DisplacedLognormalGARCH

        public DisplacedLognormalGARCH​(double[] values,
                                       double lowerBoundDisplacement,
                                       int windowIndexStart,
                                       int windowIndexEnd)
    • Method Detail

      • getCloneWithWindow

        public HistoricalSimulationModel getCloneWithWindow​(int windowIndexStart,
                                                            int windowIndexEnd)
        Description copied from interface: HistoricalSimulationModel
        Create a new model, using only a window of the times series.
        Specified by:
        getCloneWithWindow in interface HistoricalSimulationModel
        Parameters:
        windowIndexStart - Index of the first element to be part of the new time series.
        windowIndexEnd - Index of the last element to be part of the new time series.
        Returns:
        A new historical simulation using a different data window.
      • getCloneWithWindow

        public HistoricalSimulationModel getCloneWithWindow​(double lowerBoundDisplacement,
                                                            int windowIndexStart,
                                                            int windowIndexEnd)
      • getLogLikelihoodForParameters

        public double getLogLikelihoodForParameters​(double omega,
                                                    double alpha,
                                                    double beta,
                                                    double displacement)
      • getLastResidualForParameters

        public double getLastResidualForParameters​(double omega,
                                                   double alpha,
                                                   double beta,
                                                   double displacement)
      • getSzenarios

        public double[] getSzenarios​(double omega,
                                     double alpha,
                                     double beta,
                                     double displacement)
      • getQuantilPredictionsForParameters

        public double[] getQuantilPredictionsForParameters​(double omega,
                                                           double alpha,
                                                           double beta,
                                                           double displacement,
                                                           double[] quantiles)