/**
* Calculates the delta of a call option under a Black-Scholes model
*
* <p>The method also handles cases where the forward and/or option strike is negative and some
* limit cases where the forward or the option strike is zero. In the case forward = option strike
* = 0 the method returns 1.0.
*
* @param initialStockValue The initial value of the underlying, i.e., the spot.
* @param riskFreeRate The risk free rate of the bank account numerarie.
* @param volatility The Black-Scholes volatility.
* @param optionMaturity The option maturity T.
* @param optionStrike The option strike.
* @return The delta of the option
*/
public static RandomVariableInterface blackScholesOptionDelta(
RandomVariableInterface initialStockValue,
RandomVariableInterface riskFreeRate,
RandomVariableInterface volatility,
double optionMaturity,
RandomVariableInterface optionStrike) {
if (optionMaturity < 0) {
return initialStockValue.mult(0.0);
} else {
// Calculate delta
RandomVariableInterface dPlus =
initialStockValue
.div(optionStrike)
.log()
.add(volatility.squared().mult(0.5).add(riskFreeRate).mult(optionMaturity))
.div(volatility)
.div(Math.sqrt(optionMaturity));
UnivariateFunction cummulativeNormal =
new UnivariateFunction() {
public double value(double x) {
return NormalDistribution.cumulativeDistribution(x);
}
};
RandomVariableInterface delta = dPlus.apply(cummulativeNormal);
return delta;
}
}
/**
* Calculates the Black-Scholes option value of a call, i.e., the payoff max(S(T)-K,0) P, where S
* follows a log-normal process with constant log-volatility.
*
* <p>The model specific quantities are considered to be random variable, i.e., the function may
* calculate an per-path valuation in a single call.
*
* @param forward The forward of the underlying.
* @param volatility The Black-Scholes volatility.
* @param optionMaturity The option maturity T.
* @param optionStrike The option strike. If the option strike is ≤ 0.0 the method returns the
* value of the forward contract paying S(T)-K in T.
* @param payoffUnit The payoff unit (e.g., the discount factor)
* @return Returns the value of a European call option under the Black-Scholes model.
*/
public static RandomVariableInterface blackScholesGeneralizedOptionValue(
RandomVariableInterface forward,
RandomVariableInterface volatility,
double optionMaturity,
double optionStrike,
RandomVariableInterface payoffUnit) {
if (optionMaturity < 0) {
return forward.mult(0.0);
} else {
RandomVariableInterface dPlus =
forward
.div(optionStrike)
.log()
.add(volatility.squared().mult(0.5 * optionMaturity))
.div(volatility)
.div(Math.sqrt(optionMaturity));
RandomVariableInterface dMinus = dPlus.sub(volatility.mult(Math.sqrt(optionMaturity)));
UnivariateFunction cumulativeNormal =
new UnivariateFunction() {
public double value(double x) {
return NormalDistribution.cumulativeDistribution(x);
}
};
RandomVariableInterface valueAnalytic =
dPlus
.apply(cumulativeNormal)
.mult(forward)
.sub(dMinus.apply(cumulativeNormal).mult(optionStrike))
.mult(payoffUnit);
return valueAnalytic;
}
}
/* (non-Javadoc)
* @see net.finmath.stochastic.RandomVariableInterface#equals(net.finmath.montecarlo.RandomVariable)
*/
@Override
public boolean equals(RandomVariableInterface randomVariable) {
if (this.time != randomVariable.getFiltrationTime()) return false;
if (this.isDeterministic() && randomVariable.isDeterministic()) {
return this.valueIfNonStochastic == randomVariable.get(0);
}
if (this.isDeterministic() != randomVariable.isDeterministic()) return false;
for (int i = 0; i < realizations.length; i++)
if (realizations[i] != randomVariable.get(i)) return false;
return true;
}
/**
* Calculates the option value of a call, i.e., the payoff max(S(T)-K,0) P, where S follows a
* normal process with constant volatility, i.e., a Bachelier model.
*
* @param forward The forward of the underlying.
* @param volatility The Bachelier volatility.
* @param optionMaturity The option maturity T.
* @param optionStrike The option strike.
* @param payoffUnit The payoff unit (e.g., the discount factor)
* @return Returns the value of a European call option under the Bachelier model.
*/
public static RandomVariableInterface bachelierOptionValue(
RandomVariableInterface forward,
RandomVariableInterface volatility,
double optionMaturity,
double optionStrike,
RandomVariableInterface payoffUnit) {
if (optionMaturity < 0) {
return forward.mult(0.0);
} else {
RandomVariableInterface integratedVolatility = volatility.mult(Math.sqrt(optionMaturity));
RandomVariableInterface dPlus = forward.sub(optionStrike).div(integratedVolatility);
UnivariateFunction cummulativeNormal =
new UnivariateFunction() {
public double value(double x) {
return NormalDistribution.cumulativeDistribution(x);
}
};
UnivariateFunction densityNormal =
new UnivariateFunction() {
public double value(double x) {
return NormalDistribution.density(x);
}
};
RandomVariableInterface valueAnalytic =
dPlus
.apply(cummulativeNormal)
.mult(forward.sub(optionStrike))
.add(dPlus.apply(densityNormal).mult(integratedVolatility))
.mult(payoffUnit);
return valueAnalytic;
}
}
public RandomVariableInterface barrier(
RandomVariableInterface trigger,
RandomVariableInterface valueIfTriggerNonNegative,
double valueIfTriggerNegative) {
return this.barrier(
trigger,
valueIfTriggerNonNegative,
new RandomVariable(valueIfTriggerNonNegative.getFiltrationTime(), valueIfTriggerNegative));
}
/* (non-Javadoc)
* @see net.finmath.stochastic.RandomVariableInterface#subRatio(net.finmath.stochastic.RandomVariableInterface, net.finmath.stochastic.RandomVariableInterface)
*/
public RandomVariableInterface subRatio(
RandomVariableInterface numerator, RandomVariableInterface denominator) {
// Set time of this random variable to maximum of time with respect to which measurability is
// known.
double newTime =
Math.max(Math.max(time, numerator.getFiltrationTime()), denominator.getFiltrationTime());
if (isDeterministic() && numerator.isDeterministic() && denominator.isDeterministic()) {
double newValueIfNonStochastic =
valueIfNonStochastic - (numerator.get(0) / denominator.get(0));
return new RandomVariable(newTime, newValueIfNonStochastic);
} else {
double[] newRealizations =
new double[Math.max(Math.max(size(), numerator.size()), denominator.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = get(i) - numerator.get(i) / denominator.get(i);
return new RandomVariable(newTime, newRealizations);
}
}
/**
* This static method calculated the gamma of a call option under a Black-Scholes model
*
* @param initialStockValue The initial value of the underlying, i.e., the spot.
* @param riskFreeRate The risk free rate of the bank account numerarie.
* @param volatility The Black-Scholes volatility.
* @param optionMaturity The option maturity T.
* @param optionStrike The option strike.
* @return The gamma of the option
*/
public static RandomVariableInterface blackScholesOptionGamma(
RandomVariableInterface initialStockValue,
RandomVariableInterface riskFreeRate,
RandomVariableInterface volatility,
double optionMaturity,
double optionStrike) {
if (optionStrike <= 0.0 || optionMaturity <= 0.0) {
// The Black-Scholes model does not consider it being an option
return initialStockValue.mult(0.0);
} else {
// Calculate gamma
RandomVariableInterface dPlus =
initialStockValue
.div(optionStrike)
.log()
.add(volatility.squared().mult(0.5).add(riskFreeRate).mult(optionMaturity))
.div(volatility)
.div(Math.sqrt(optionMaturity));
RandomVariableInterface gamma =
dPlus
.squared()
.mult(-0.5)
.exp()
.div(
initialStockValue
.mult(volatility)
.mult(Math.sqrt(2.0 * Math.PI * optionMaturity)));
return gamma;
}
}
@Override
public double getAverage(RandomVariableInterface probabilities) {
if (isDeterministic()) return valueIfNonStochastic;
if (size() == 0) return Double.NaN;
/*
* Kahan summation on (realizations[i] * probabilities.get(i))
*/
double sum = 0.0;
double error = 0.0; // Running error compensation
for (int i = 0; i < realizations.length; i++) {
double value = realizations[i] * probabilities.get(i) - error; // Error corrected value
double newSum = sum + value; // New sum
error = (newSum - sum) - value; // New numerical error
sum = newSum;
}
return sum / realizations.length;
}
/* (non-Javadoc)
* @see net.finmath.stochastic.RandomVariableInterface#floor(net.finmath.stochastic.RandomVariableInterface)
*/
public RandomVariableInterface floor(RandomVariableInterface randomVariable) {
// Set time of this random variable to maximum of time with respect to which measurability is
// known.
double newTime = Math.max(time, randomVariable.getFiltrationTime());
if (isDeterministic() && randomVariable.isDeterministic()) {
double newValueIfNonStochastic = FastMath.max(valueIfNonStochastic, randomVariable.get(0));
return new RandomVariable(newTime, newValueIfNonStochastic);
} else if (isDeterministic()) return randomVariable.floor(this);
else {
double[] newRealizations = new double[Math.max(size(), randomVariable.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = FastMath.max(realizations[i], randomVariable.get(i));
return new RandomVariable(newTime, newRealizations);
}
}
@Override
public double getVariance(RandomVariableInterface probabilities) {
if (isDeterministic()) return 0.0;
if (size() == 0) return Double.NaN;
double average = getAverage(probabilities);
/*
* Kahan summation on (realizations[i] - average)^2 * probabilities.get(i)
*/
double sum = 0.0;
double errorOfSum = 0.0;
for (int i = 0; i < realizations.length; i++) {
double value =
(realizations[i] - average) * (realizations[i] - average) * probabilities.get(i)
- errorOfSum;
double newSum = sum + value;
errorOfSum = (newSum - sum) - value;
sum = newSum;
}
return sum;
}
/* (non-Javadoc)
* @see net.finmath.stochastic.RandomVariableInterface#addProduct(net.finmath.stochastic.RandomVariableInterface, double)
*/
public RandomVariableInterface addProduct(RandomVariableInterface factor1, double factor2) {
// Set time of this random variable to maximum of time with respect to which measurability is
// known.
double newTime = Math.max(time, factor1.getFiltrationTime());
if (isDeterministic() && factor1.isDeterministic()) {
double newValueIfNonStochastic = valueIfNonStochastic + (factor1.get(0) * factor2);
return new RandomVariable(newTime, newValueIfNonStochastic);
} else if (isDeterministic() && !factor1.isDeterministic()) {
double[] factor1Realizations = factor1.getRealizations();
double[] newRealizations = new double[Math.max(size(), factor1.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = valueIfNonStochastic + factor1Realizations[i] * factor2;
return new RandomVariable(newTime, newRealizations);
} else if (!isDeterministic() && factor1.isDeterministic()) {
double factor1Value = factor1.get(0);
double[] newRealizations = new double[Math.max(size(), factor1.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = realizations[i] + factor1Value * factor2;
return new RandomVariable(newTime, newRealizations);
} else {
double[] factor1Realizations = factor1.getRealizations();
double[] newRealizations = new double[Math.max(size(), factor1.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = realizations[i] + factor1Realizations[i] * factor2;
return new RandomVariable(newTime, newRealizations);
}
}
/* (non-Javadoc)
* @see net.finmath.stochastic.RandomVariableInterface#barrier(net.finmath.stochastic.RandomVariableInterface, net.finmath.stochastic.RandomVariableInterface, net.finmath.stochastic.RandomVariableInterface)
*/
public RandomVariableInterface barrier(
RandomVariableInterface trigger,
RandomVariableInterface valueIfTriggerNonNegative,
RandomVariableInterface valueIfTriggerNegative) {
// Set time of this random variable to maximum of time with respect to which measurability is
// known.
double newTime = Math.max(time, trigger.getFiltrationTime());
newTime = Math.max(newTime, valueIfTriggerNonNegative.getFiltrationTime());
newTime = Math.max(newTime, valueIfTriggerNegative.getFiltrationTime());
if (isDeterministic()
&& trigger.isDeterministic()
&& valueIfTriggerNonNegative.isDeterministic()
&& valueIfTriggerNegative.isDeterministic()) {
double newValueIfNonStochastic =
trigger.get(0) >= 0 ? valueIfTriggerNonNegative.get(0) : valueIfTriggerNegative.get(0);
return new RandomVariable(newTime, newValueIfNonStochastic);
} else {
int numberOfPaths =
Math.max(
Math.max(trigger.size(), valueIfTriggerNonNegative.size()),
valueIfTriggerNegative.size());
double[] newRealizations = new double[numberOfPaths];
for (int i = 0; i < newRealizations.length; i++) {
newRealizations[i] =
trigger.get(i) >= 0.0
? valueIfTriggerNonNegative.get(i)
: valueIfTriggerNegative.get(i);
}
return new RandomVariable(newTime, newRealizations);
}
}
/* (non-Javadoc)
* @see net.finmath.stochastic.RandomVariableInterface#discount(net.finmath.stochastic.RandomVariableInterface, double)
*/
public RandomVariableInterface discount(RandomVariableInterface rate, double periodLength) {
// Set time of this random variable to maximum of time with respect to which measurability is
// known.
double newTime = Math.max(time, rate.getFiltrationTime());
if (isDeterministic() && rate.isDeterministic()) {
double newValueIfNonStochastic = valueIfNonStochastic / (1 + rate.get(0) * periodLength);
return new RandomVariable(newTime, newValueIfNonStochastic);
} else if (isDeterministic() && !rate.isDeterministic()) {
double[] rateRealizations = rate.getRealizations();
double[] newRealizations = new double[Math.max(size(), rate.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = valueIfNonStochastic / (1.0 + rateRealizations[i] * periodLength);
return new RandomVariable(newTime, newRealizations);
} else if (!isDeterministic() && rate.isDeterministic()) {
double rateValue = rate.get(0);
double[] newRealizations = new double[Math.max(size(), rate.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = realizations[i] / (1.0 + rateValue * periodLength);
return new RandomVariable(newTime, newRealizations);
} else {
double[] rateRealizations = rate.getRealizations();
double[] newRealizations = new double[Math.max(size(), rate.size())];
for (int i = 0; i < newRealizations.length; i++)
newRealizations[i] = realizations[i] / (1.0 + rateRealizations[i] * periodLength);
return new RandomVariable(newTime, newRealizations);
}
}
/**
* Create a random variable from a given other implementation of <code>RandomVariableInterface
* </code>.
*
* @param value Object implementing <code>RandomVariableInterface</code>.
*/
public RandomVariable(RandomVariableInterface value) {
super();
this.time = value.getFiltrationTime();
this.realizations = value.isDeterministic() ? null : value.getRealizations();
this.valueIfNonStochastic = value.isDeterministic() ? value.get(0) : Double.NaN;
}
@Override
public RandomVariableInterface applyStateSpaceTransform(
int componentIndex, RandomVariableInterface randomVariable) {
return randomVariable.exp();
}
