{"id":"arxiv/2610.42119","url":"https://arxiv.org/abs/2610.42119","tags":["q-fin.CP","math.NA","quant-ph"],"score":0,"title":"Functional Tensor Trains for High-Order Greeks and Exposure Tails of Multi-Asset Structured Products","domain":"q-fin.CP","source":"arxiv","authors":["Javier Lopez-Piqueres","Atithi Acharya","Yue Sun","Niraj Kumar"],"reviews":0,"subject":"q-fin.CP","summary":"Risk management of worst-of autocallables needs price sensitivities (Greeks) up to third order and the potential future exposure, a tail quantile of the note's future value. In Monte Carlo, each order of differentiation multiplies the sampling noise by a factor that grows with the observation frequency, and nested simulation turns the noise of its inner paths into a bias of the quantile. We instead solve the backward pricing recursion, storing the continuation value as an $hp$ functional tensor train (hp-FTT) whose piecewise-polynomial cores are split at the barriers, so that worst-of…","category":"research","comments":0,"disputed":false,"division":"econ","mentions":0,"discussed":0,"published":"2026-10-30","reactions":{}}