All Downloads are FREE. Search and download functionalities are using the official Maven repository.

net.finmath.marketdata.model.curves.DiscountCurveNelsonSiegelSvensson Maven / Gradle / Ivy

Go to download

finmath lib is a Mathematical Finance Library in Java. It provides algorithms and methodologies related to mathematical finance.

There is a newer version: 6.0.19
Show newest version
/*
 * (c) Copyright Christian P. Fries, Germany. Contact: [email protected].
 *
 * Created on 20.05.2005
 */
package net.finmath.marketdata.model.curves;

import java.io.Serializable;
import java.time.LocalDate;
import java.util.Arrays;

import net.finmath.marketdata.model.AnalyticModel;

/**
 * Implementation of a discount factor curve given by a Nelson-Siegel-Svensson (NSS) parameterization.
 * In the NSS parameterization the zero rate \( r(T) \) is given by
 *
 * \[ r(T) = \beta_0 + \beta_1 \frac{1-x_0}{T/\tau_0} + \beta_2 ( \frac{1-x_0}{T/\tau_0} - x_0) + \beta_3 ( \frac{1-x_1}{T/\tau_1} - x_1) \]
 *
 * where \( x_0 = \exp(-T/\tau_0) \) and \( x_1 = \exp(-T/\tau_1) \).
 *
 * The sub-family of curves with \( \beta_3 = 0 \) is called Nelson-Siegel parameterization.
 *
 * Note: This is a time-parameterized model. The finmath lib library uses an internal mapping from date to times \( t \).
 * This mapping does not necessarily need to correspond with the curves understanding for the parameter \( T \).
 * For that reason this class allows to re-scale the time parameter. Currently only a simple re-scaling factor is
 * supported.
 *
 * The parameter T used in the parameterization is given by T = timeScaling * t, where t is the maturity as an ACT/365
 * year fraction from the given reference date.
 *
 * @author Christian Fries
 * @version 1.0
 */
public class DiscountCurveNelsonSiegelSvensson extends AbstractCurve implements Serializable, DiscountCurve {

	private static final long serialVersionUID = 8024640795839972709L;

	private final double	timeScaling;
	private final double[]	parameter;

	/**
	 * Create a discount curve using a Nelson-Siegel-Svensson parametrization.
	 *
	 * @param name The name of the curve (the curve can be referenced under this name, if added to an AnalyticModelFromCuvesAndVols.
	 * @param referenceDate The reference date of this curve, i.e. the date associated with t=0.
	 * @param parameter The Nelson-Siegel-Svensson parameters in the order \( ( \beta_0, \beta_1, \beta_2, \beta_3, \tau_0, \tau_1 ) \).
	 * @param timeScaling The time parameter argument rescaling. See {@link #getDiscountFactor(AnalyticModel, double)}.
	 */
	public DiscountCurveNelsonSiegelSvensson(final String name, final LocalDate referenceDate, final double[] parameter, final double timeScaling) {
		super(name, referenceDate);
		this.timeScaling = timeScaling;

		this.parameter = parameter.clone();
	}

	@Override
	public double getDiscountFactor(final double maturity)
	{
		return getDiscountFactor(null, maturity);
	}

	/**
	 * Return the discount factor within a given model context for a given maturity.
	 * @param model The model used as a context (not required for this class).
	 * @param maturity The maturity in terms of ACT/365 daycount form this curve reference date. Note that this parameter might get rescaled to a different time parameter.
	 * @see net.finmath.marketdata.model.curves.DiscountCurve#getDiscountFactor(net.finmath.marketdata.model.AnalyticModel, double)
	 */
	@Override
	public double getDiscountFactor(final AnalyticModel model, double maturity)
	{
		// Change time scale
		maturity *= timeScaling;

		final double beta1	= parameter[0];
		final double beta2	= parameter[1];
		final double beta3	= parameter[2];
		final double beta4	= parameter[3];
		final double tau1		= parameter[4];
		final double tau2		= parameter[5];

		final double x1 = tau1 > 0 ? Math.exp(-maturity/tau1) : 0.0;
		final double x2 = tau2 > 0 ? Math.exp(-maturity/tau2) : 0.0;

		final double y1 = tau1 > 0 ? (maturity > 0.0 ? (1.0-x1)/maturity*tau1 : 1.0) : 0.0;
		final double y2 = tau2 > 0 ? (maturity > 0.0 ? (1.0-x2)/maturity*tau2 : 1.0) : 0.0;

		final double zeroRate = beta1 + beta2 * y1 + beta3 * (y1-x1) + beta4 * (y2-x2);

		return Math.exp(- zeroRate * maturity);
	}

	@Override
	public double getValue(final AnalyticModel model, final double time) {
		return getDiscountFactor(model, time);
	}

	/**
	 * Returns the zero rate for a given maturity, i.e., -ln(df(T)) / T where T is the given maturity and df(T) is
	 * the discount factor at time $T$.
	 *
	 * @param maturity The given maturity.
	 * @return The zero rate.
	 */
	public double getZeroRate(final double maturity)
	{
		if(maturity == 0) {
			return this.getZeroRate(1.0E-14);
		}

		return -Math.log(getDiscountFactor(null, maturity))/maturity;
	}

	/**
	 * Returns the zero rates for a given vector maturities.
	 *
	 * @param maturities The given maturities.
	 * @return The zero rates.
	 */
	public double[] getZeroRates(final double[] maturities)
	{
		final double[] values = new double[maturities.length];

		for(int i=0; i




© 2015 - 2024 Weber Informatics LLC | Privacy Policy