A fixed-time stable limit theorem for non-normalized functionals on irregular time grids

Abstract

Let an Itô semimartingale be observed at times satisfying $τ_{i+1}^n-τ_i^n=(nθ_{τ_i^n})^{-1}$, where $θ$ is adapted and càdlàg, and $θ$ and its reciprocal are bounded on compact intervals. Under the usual hypothesis (H) on the semimartingale, we prove fixed-time stable convergence for test functions whose Hessian is $o(\norm{x})$ at zero. The limit is a sum of jump contributions with sampling scale $θ_{T-}^{-1}$ and separate pre-jump and post-jump volatilities. The proof establishes joint stable convergence of the local Poisson positions and Brownian increments on the endogenous grid, and controls infinitely many small jumps by a uniform Itô estimate. Three examples identify the left-limit sampling scale and show why positivity and càdlàg regularity alone cannot replace local control of the reciprocal sampling intensity.