Saved in:
Bibliographic Details
Main Authors: Perez-Diaz, Alvaro, Loach, James C., Toutoungi, Danielle E., Middleton, Lee
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.19345
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914107739865088
author Perez-Diaz, Alvaro
Loach, James C.
Toutoungi, Danielle E.
Middleton, Lee
author_facet Perez-Diaz, Alvaro
Loach, James C.
Toutoungi, Danielle E.
Middleton, Lee
contents Time-series foundation models (TSFMs) achieve strong forecast accuracy, yet accuracy alone does not determine practical value. The form of a forecast -- point, quantile, parametric, or trajectory ensemble -- fundamentally constrains which operational tasks it can support. We survey recent TSFMs and find that two-thirds produce only point or parametric forecasts, while many operational tasks require trajectory ensembles that preserve temporal dependence. We establish when forecast types can be converted and when they cannot: trajectory ensembles convert to simpler forms via marginalization without additional assumptions, but the reverse requires imposing temporal dependence through copulas or conformal methods. We prove that marginals cannot determine path-dependent event probabilities -- infinitely many joint distributions share identical marginals but yield different answers to operational questions. We map six fundamental forecasting tasks to minimal sufficient forecast types and provide a task-aligned evaluation framework. Our analysis clarifies when forecast type, not accuracy, differentiates practical utility.
format Preprint
id arxiv_https___arxiv_org_abs_2510_19345
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Foundation Model Forecasts: Form and Function
Perez-Diaz, Alvaro
Loach, James C.
Toutoungi, Danielle E.
Middleton, Lee
Machine Learning
Artificial Intelligence
Time-series foundation models (TSFMs) achieve strong forecast accuracy, yet accuracy alone does not determine practical value. The form of a forecast -- point, quantile, parametric, or trajectory ensemble -- fundamentally constrains which operational tasks it can support. We survey recent TSFMs and find that two-thirds produce only point or parametric forecasts, while many operational tasks require trajectory ensembles that preserve temporal dependence. We establish when forecast types can be converted and when they cannot: trajectory ensembles convert to simpler forms via marginalization without additional assumptions, but the reverse requires imposing temporal dependence through copulas or conformal methods. We prove that marginals cannot determine path-dependent event probabilities -- infinitely many joint distributions share identical marginals but yield different answers to operational questions. We map six fundamental forecasting tasks to minimal sufficient forecast types and provide a task-aligned evaluation framework. Our analysis clarifies when forecast type, not accuracy, differentiates practical utility.
title Foundation Model Forecasts: Form and Function
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2510.19345