← back to finance 2

Monte Carlo Methods

MIT OCW 18.S096 + 15.450 (CC BY-NC-SA 4.0)

Monte Carlo methods price derivatives by simulating thousands of random paths for the underlying asset, then averaging the discounted payoffs. The law of large numbers guarantees convergence; variance reduction makes it practical.

Time Price K S₀ Sᵀ

Simulating random paths

Under geometric Brownian motion, a stock price evolves as S(t+dt) = S(t) * exp((mu - sigma^2/2)*dt + sigma*sqrt(dt)*Z), where Z is a standard normal draw. Each simulation traces one possible future. Run enough of them and you map the distribution.

Scheme

Pricing European options by simulation

A European call pays max(S_T - K, 0) at expiry. Simulate many terminal prices under the risk-neutral measure (replace mu with r), compute each payoff, discount back, and average. With enough paths, the Monte Carlo estimate converges to the Black-Scholes price.

Scheme

Variance reduction techniques

Antithetic variates: for every normal draw Z, also use -Z. The two paths are negatively correlated, so their average has lower variance than two independent draws. This cuts the standard error roughly in half for free.

Scheme

Path-dependent options

An Asian option pays based on the average price along the path, not just the terminal price. A barrier option knocks in or out if the price crosses a threshold. Both require simulating the full path, not just the endpoint.

Scheme

Convergence and standard error

Monte Carlo converges at rate 1/sqrt(N) regardless of dimension. Double the paths, cut the error by sqrt(2). The standard error of the estimate is the sample standard deviation divided by sqrt(N). This tells you how many paths you need for a given precision.

Scheme
Neighbors