The chain rule of stochastic calculus. For X_t satisfying dX_t = μ_t,dt + σ_t,dW_t and f in C^1,2: df(t, X_t) = left(dfracpartial fpartial t + μ_t dfracpartial fpartial x + tfrac12σ_t^2 dfracpartial^2 fpartial x^2right)dt + σ_t dfracpartial fpartial x,dW_t. The extra tfrac12σ^2 partial_xx f term comes from (dW)^2 = dt (Brownian quadratic variation), absent from ordinary calculus. Applied to f = ln S under GBM yields the famous drift correction dln S = (μ - tfrac12σ^2)dt + σ,dW.