Saved in:
Bibliographic Details
Main Authors: Potapczynski, Andres, Selvam, Ravi Kiran, Konstantinova, Tatiana, Ramasubramanian, Shankar, Wolff, Malcolm, Olivares, Kin G., Ma, Ruijun, Cao, Mengfei, Mahoney, Michael W., Wilson, Andrew Gordon, Oreshkin, Boris N., Efimov, Dmitry
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.15802
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912969285173248
author Potapczynski, Andres
Selvam, Ravi Kiran
Konstantinova, Tatiana
Ramasubramanian, Shankar
Wolff, Malcolm
Olivares, Kin G.
Ma, Ruijun
Cao, Mengfei
Mahoney, Michael W.
Wilson, Andrew Gordon
Oreshkin, Boris N.
Efimov, Dmitry
author_facet Potapczynski, Andres
Selvam, Ravi Kiran
Konstantinova, Tatiana
Ramasubramanian, Shankar
Wolff, Malcolm
Olivares, Kin G.
Ma, Ruijun
Cao, Mengfei
Mahoney, Michael W.
Wilson, Andrew Gordon
Oreshkin, Boris N.
Efimov, Dmitry
contents In many time series forecasting settings, the target time series is accompanied by exogenous covariates, such as promotions and prices in retail demand; temperature in energy load; calendar and holiday indicators for traffic or sales; and grid load or fuel costs in electricity pricing. Ignoring these exogenous signals can substantially degrade forecasting accuracy, particularly when they drive spikes, discontinuities, or regime and phase changes in the target series. Most current time series foundation models (e.g., Chronos, Sundial, TimesFM, TimeMoE, TimeLLM, and LagLlama) ignore exogenous covariates and make forecasts solely from the numerical time series history, thereby limiting their performance. In this paper, we develop ApolloPFN, a prior-data fitted network (PFN) that is time-aware (unlike prior PFNs) and that natively incorporates exogenous covariates (unlike prior univariate forecasters). Our design introduces two major advances: (i) a synthetic data generation procedure tailored to resolve the failure modes that arise when tabular (non-temporal) PFNs are applied to time series; and (ii) time-aware architectural modifications that embed inductive biases needed to exploit the time series context. We demonstrate that ApolloPFN achieves state-of-the-art results across benchmarks, such as M5 and electric price forecasting, that contain exogenous information.
format Preprint
id arxiv_https___arxiv_org_abs_2603_15802
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Time-Aware Prior Fitted Networks for Zero-Shot Forecasting with Exogenous Variables
Potapczynski, Andres
Selvam, Ravi Kiran
Konstantinova, Tatiana
Ramasubramanian, Shankar
Wolff, Malcolm
Olivares, Kin G.
Ma, Ruijun
Cao, Mengfei
Mahoney, Michael W.
Wilson, Andrew Gordon
Oreshkin, Boris N.
Efimov, Dmitry
Machine Learning
In many time series forecasting settings, the target time series is accompanied by exogenous covariates, such as promotions and prices in retail demand; temperature in energy load; calendar and holiday indicators for traffic or sales; and grid load or fuel costs in electricity pricing. Ignoring these exogenous signals can substantially degrade forecasting accuracy, particularly when they drive spikes, discontinuities, or regime and phase changes in the target series. Most current time series foundation models (e.g., Chronos, Sundial, TimesFM, TimeMoE, TimeLLM, and LagLlama) ignore exogenous covariates and make forecasts solely from the numerical time series history, thereby limiting their performance. In this paper, we develop ApolloPFN, a prior-data fitted network (PFN) that is time-aware (unlike prior PFNs) and that natively incorporates exogenous covariates (unlike prior univariate forecasters). Our design introduces two major advances: (i) a synthetic data generation procedure tailored to resolve the failure modes that arise when tabular (non-temporal) PFNs are applied to time series; and (ii) time-aware architectural modifications that embed inductive biases needed to exploit the time series context. We demonstrate that ApolloPFN achieves state-of-the-art results across benchmarks, such as M5 and electric price forecasting, that contain exogenous information.
title Time-Aware Prior Fitted Networks for Zero-Shot Forecasting with Exogenous Variables
topic Machine Learning
url https://arxiv.org/abs/2603.15802