The Stochastic Calculus of Finance (Itô's Lemma) — Wiki | SOPHIE

Overview

Itô's Lemma is the mathematical bridge between the smooth, deterministic world of Newtonian calculus and the jagged, unpredictable reality of financial markets. It serves as the fundamental theorem of stochastic calculus, enabling the pricing of derivatives in continuous time.

The Problem with Ordinary Calculus

Standard calculus relies on local linearity (the assumption that if you zoom in enough, curves look like straight lines). However, financial asset paths (modeled as random walks) are continuous everywhere but differentiable nowhere. Because they are fractally rough, they have infinite total variation, rendering ordinary derivatives meaningless.

The Wiener Process & Quadratic Variation

To perform calculus, we rely on the properties of Brownian Motion (the Wiener process, $W_t$ ):

Independent Increments: The process is Markovian.
Gaussian Increments: Variance scales linearly with time ( $W_t - W_s \sim \mathcal{N}(0, t-s)$ ).
Quadratic Variation: Over a small interval $dt$ , the squared random increment converges deterministically to time itself: $(dW_t)^2 \rightarrow dt$

Because $(dW)^2$ is of order $dt$ (rather than vanishing like $(dx)^2$ in ordinary calculus), second-order terms must be retained in Taylor expansions.

Deriving Itô's Lemma

When expanding a function $f(x, t)$ to the second order where $x$ follows a diffusion process $dx = a dt + b dW$ , the quadratic term $(dx)^2$ simplifies to $b^2 dt$ . Substituting this back yields the fundamental Itô formula: $df = \left( \frac{\partial f}{\partial t} + a \frac{\partial f}{\partial x} + \frac{1}{2} b^2 \frac{\partial^2 f}{\partial x^2} \right) dt + b \frac{\partial f}{\partial x} dW$

The Convexity Correction

The term $\frac{1}{2} b^2 \frac{\partial^2 f}{\partial x^2}$ is the convexity correction. If a payoff profile is convex (curved upward), volatility creates a positive expected drift. This is the mathematical manifestation of Jensen's Inequality ( $E[f(x)] \geq f(E[x])$ ).

Case Study: Geometric Brownian Motion (GBM)

Applying Itô's Lemma to the log of a stock price ( $f(S) = \ln(S)$ ) where $dS = \mu S dt + \sigma S dW$ :

1st Derivative: $1/S$
2nd Derivative: $-1/S^2$ The resulting log-normal dynamics show that the drift is reduced by the volatility drag: $d(\ln S) = \left( \mu - \frac{1}{2}\sigma^2 \right) dt + \sigma dW$

The Black-Scholes PDE

The "killer app" of Itô's Lemma. By creating a risk-free portfolio (Delta hedging), the stochastic terms cancel out, leaving a deterministic partial differential equation: $\frac{\partial V}{\partial t} + rS \frac{\partial V}{\partial S} + \frac{1}{2}\sigma^2 S^2 \frac{\partial^2 V}{\partial S^2} = rV$ (Theta + Risk-Free Drift + Convexity Gains = Risk-Free Return)

This equation fundamentally proves that the option price depends only on volatility ( $\sigma$ ), not on the asset's expected physical return ( $\mu$ ).

Advanced Extensions

Multidimensional Itô: For baskets/spread options, cross-gamma ( $\frac{\partial^2 V}{\partial S_1 \partial S_2}$ ) captures correlation dynamics.
Girsanov Theorem: Allows for the change of measure from Physical ( $\mathbb{P}$ ) to Risk-Neutral ( $\mathbb{Q}$ ) by altering the drift of the Brownian motion.
Martingale Representation Theorem: Guarantees that any martingale can be written as an Itô integral, ensuring market completeness and the existence of a perfect hedge.