Class MonteCarloSimulator

java.lang.Object

com.quantfinlib.simulation.MonteCarloSimulator

public final class MonteCarloSimulator
extends Object

Monte Carlo simulation of geometric Brownian motion, single- and multi-asset, in parallel; results are deterministic for a given seed regardless of thread scheduling (each scenario has its own SplittableRandom stream derived from the seed).

• simulate(double, double, double, int, int) and simulatePortfolio(double, double[], double[], double[][], int, int) return terminal values for risk analytics (SimulationResult). Both sample the exact lognormal terminal distribution, so there is no time-discretisation error.
• simulatePaths(double[], double[], double[], double[][], double, int, int, boolean, MonteCarloSimulator.PathPayoff) generates correlated multi-asset GBM paths on a uniform grid with exact log-Euler steps and hands each path to a caller payoff (barriers, Asians, autocalls, baskets), with optional antithetic variates (Glasserman, Monte Carlo Methods in Financial Engineering, 2003, §4.2).

Model per asset i: dS_i/S_i = b_i dt + σ_i dW_i, d⟨W_i, W_j⟩ = ρ_ij dt. For pricing pass the risk-neutral drift b_i = r − q_i and discount the payoff yourself (or use price(double[], double[], double[], double[][], double, int, int, boolean, double, MonteCarloSimulator.PathPayoff)); for risk pass the real-world drift.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static final record	MonteCarloSimulator.Estimate	Monte Carlo estimate: mean, its standard error, and the number of independent samples.
static interface	MonteCarloSimulator.PathPayoff	Payoff of one simulated path.

Constructor Summary

Constructors

Constructor	Description
MonteCarloSimulator()
MonteCarloSimulator(long seed)

Method Summary

Modifier and Type	Method	Description
MonteCarloSimulator.Estimate	price(double[] spots, double[] drifts, double[] vols, double[][] correlation, double maturity, int steps, int paths, boolean antithetic, double discountRate, MonteCarloSimulator.PathPayoff payoff)	Discounted Monte Carlo price e^{−rT}·E[payoff] with its standard error.
SimulationResult	simulate(double initialValue, double annualReturn, double annualVol, int horizonDays, int simulations)	Single-asset GBM terminal values after horizonDays trading days (T = horizonDays/252): V_T = V_0·exp((μ − σ²/2)T + σ√T·Z).
double[]	simulatePaths(double[] spots, double[] drifts, double[] vols, double[][] correlation, double maturity, int steps, int paths, boolean antithetic, MonteCarloSimulator.PathPayoff payoff)	Simulates paths correlated GBM paths and returns the payoff of each.
SimulationResult	simulatePortfolio(double initialValue, double[] weights, double[] dailyMeanLogReturns, double[][] dailyCovariance, int horizonDays, int simulations)	Buy-and-hold multi-asset portfolio.

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

MonteCarloSimulator

public MonteCarloSimulator()

MonteCarloSimulator

public MonteCarloSimulator(long seed)

Method Details

simulate

public SimulationResult simulate(double initialValue,
 double annualReturn,
 double annualVol,
 int horizonDays,
 int simulations)

Single-asset GBM terminal values after horizonDays trading days (T = horizonDays/252): V_T = V_0·exp((μ − σ²/2)T + σ√T·Z).

Parameters:

initialValue - starting value, > 0

annualReturn - annualized drift μ (so E[V_T] = V_0·e^{μT})

annualVol - annualized volatility σ, ≥ 0

horizonDays - trading days to simulate, ≥ 1

simulations - number of scenarios, ≥ 1

simulatePortfolio

public SimulationResult simulatePortfolio(double initialValue,
 double[] weights,
 double[] dailyMeanLogReturns,
 double[][] dailyCovariance,
 int horizonDays,
 int simulations)

Buy-and-hold multi-asset portfolio. Daily log returns are jointly normal with mean dailyMeanLogReturns and covariance dailyCovariance, so the terminal log return of asset i over d days is N(d·m_i, d·Σ_ii) and the portfolio is worth V_0·Σ w_i·exp(X_i). Long-only weights keep the value positive; short weights can drive it below zero, and it is not clamped (there is no absorbing bankruptcy state in GBM).

Parameters:

weights - initial value weights (need not sum to 1)

dailyCovariance - PSD covariance of daily log returns

simulatePaths

public double[] simulatePaths(double[] spots,
 double[] drifts,
 double[] vols,
 double[][] correlation,
 double maturity,
 int steps,
 int paths,
 boolean antithetic,
 MonteCarloSimulator.PathPayoff payoff)

Simulates paths correlated GBM paths and returns the payoff of each. With antithetic, paths come in pairs driven by Z and −Z; entry 2p and 2p+1 are a pair (paths must then be even).

Parameters:

spots - S_i(0) > 0

drifts - b_i per year (r − q_i for pricing)

vols - σ_i ≥ 0 per year

correlation - PSD correlation matrix (unit diagonal)

maturity - T in years, > 0

steps - grid points after spot, ≥ 1 (Δt = T/steps)

price

public MonteCarloSimulator.Estimate price(double[] spots,
 double[] drifts,
 double[] vols,
 double[][] correlation,
 double maturity,
 int steps,
 int paths,
 boolean antithetic,
 double discountRate,
 MonteCarloSimulator.PathPayoff payoff)

Discounted Monte Carlo price e^{−rT}·E[payoff] with its standard error. Under antithetic sampling the error is computed on the pair averages, which are the independent samples.