Monte Carlo Simulation for Derivative Pricing › The Risk-Neutral Valuation Framework

Before we can simulate anything, we need to establish the theoretical foundation that justifies our approach. The risk-neutral valuation framework provides this foundation by connecting derivative prices to expectations under a specific probability measure. Recall from our discussion of no-arbitrage pricing that the fair value of a derivative equals the expected discounted payoff under the risk-neutral measure $\mathbb{Q}$:

$$V_0 = e^{-rT} \mathbb{E}^{\mathbb{Q}}[\text{Payoff}(S_T)]$$

where:

• $V_0$: fair value of the derivative at time 0
• $\text{Payoff}(\cdot)$: function defining the derivative's payout
• $r$: risk-free interest rate
• $T$: time to maturity
• $\mathbb{E}^{\mathbb{Q}}$: expectation operator under the risk-neutral measure
• $S_T$: asset price at time $T$

This formula captures a key result from financial economics. In a world without arbitrage opportunities, every derivative has a unique fair price that can be computed as an expectation. The catch is that this expectation must be taken under the risk-neutral measure, not the real-world probability measure that governs actual asset returns. Under the risk-neutral measure, all assets earn the risk-free rate on average, which simplifies pricing enormously because we no longer need to estimate expected returns or risk premiums.

Under the risk-neutral measure, the underlying asset grows at the risk-free rate $r$ rather than its true expected return $\mu$. For geometric Brownian motion, the risk-neutral dynamics are expressed through the following stochastic differential equation:

$$dS_t = rS_t \, dt + \sigma S_t \, dW_t^{\mathbb{Q}}$$

where:

• $dS_t$: change in asset price over a small time interval
• $S_t$: current asset price
• $r$: risk-free interest rate
• $\sigma$: volatility of the asset
• $dt$: small time increment
• $dW_t^{\mathbb{Q}}$: increment of a Brownian motion under the risk-neutral measure

This equation tells us how the asset price evolves over infinitesimal time intervals. The equation states that the price change is composed of two distinct components. The first component, $r S_t \, dt$, represents the deterministic drift, capturing the expected growth at the risk-free rate over the small time interval. The second component, $\sigma S_t \, dW_t^{\mathbb{Q}}$, represents the stochastic diffusion, introducing a random shock whose magnitude is proportional to both the current price and the volatility parameter. The Brownian increment $dW_t^{\mathbb{Q}}$ behaves like a normally distributed random variable with mean zero and variance $dt$, creating the randomness that makes option pricing interesting.

To price options using Monte Carlo, we generate samples from the distribution of $S_T$ (or the entire path $\{S_t\}_{0 \leq t \leq T}$) under the risk-neutral measure $\mathbb{Q}$ and approximate the expectation with a sample average:

$$V_0 \approx e^{-rT} \cdot \frac{1}{N} \sum_{i=1}^{N} \text{Payoff}(S_T^{(i)})$$

where:

• $V_0$: fair value of the derivative
• $r$: risk-free interest rate
• $T$: time to maturity
• $N$: number of simulations
• $S_T^{(i)}$: $i$-th simulated asset price at time $T$
• $\text{Payoff}(\cdot)$: function defining the derivative's payout

This approximation replaces the abstract mathematical expectation with a concrete computational procedure. We generate $N$ independent random draws of the terminal asset price, compute the payoff for each realization, average these payoffs, and discount the result back to the present. By the law of large numbers, this estimator converges to the true price as $N \to \infty$. The strength of this approach is its generality: regardless of how complex the payoff function might be, if we can simulate the underlying price dynamics and evaluate the payoff, we can estimate the option's value.