Multidimensional Itô-Taylor Expansions and Lévy Areas

When pricing path-dependent derivatives in multi-factor models (e.g. a basket option under two correlated GBMs, or a Heston model combining a stock SDE and a variance SDE), Monte Carlo simulations require discretising the system of SDEs. The Euler-Maruyama scheme uses only the first-order stochastic Taylor term. The Milstein scheme adds the second-order terms, achieving a higher convergence rate — but in dimension $d > 1$ , the second-order terms involve the Lévy area: an iterated stochastic integral

A^{ij}_{st} = \int_s^t \int_s^u dW^i_r \, dW^j_u - \int_s^t \int_s^u dW^j_r \, dW^i_u

that cannot be computed from the BM increments

\Delta W^i = W^i_t - W^i_s

and

\Delta W^j = W^j_t - W^j_s

alone.

This creates a fundamental obstacle: in 1D, the Itô correction $(dW)^2 = dt$ is deterministic, so second-order terms reduce to $\Delta W^2 \approx \Delta t$ . In $d > 1$ , the off-diagonal second-order terms introduce random correction terms (the Lévy areas) that require either simulation (expensive) or bounding (introducing error). Understanding this is essential for anyone implementing or assessing multi-factor Monte Carlo engines.

Learning objectives. By the end of this note you should be able to:

Derive the first- and second-order stochastic Taylor expansion of an Itô process via iterated application of Itô's formula.
Define the iterated stochastic integral $I_{ij,st} = \int_s^t \int_s^r dW^i_u \, dW^j_r$ and the Lévy area $A^{ij}_{st}$ .
Explain why $I_{ij,st}$ for $i \ne j$ is not determined by $\Delta W^i$ and $\Delta W^j$ in dimension $d > 1$ .
State the Euler-Maruyama and Milstein schemes and compare their strong convergence orders in 1D and $d > 1$ .
Identify when Lévy areas matter in practical Monte Carlo.

In this lesson

The Stochastic Taylor Expansion
The Lévy Area
Numerical Implications
When Lévy Areas Matter (and When They Don't)

Before you start

The full lesson, with worked examples and exercise solutions, is for members.

Read the full lesson

New to SelfMadeQuant.com? See membership options or explore the curriculum.