Zero-Shot Time Series Forecasting

Overview

Zero-shot time series forecasting studies whether models pretrained on diverse real or synthetic sequences can predict new datasets without updating their parameters. Approaches include transformer and recurrent foundation models, synthetic-data training, covariate-aware prediction, and inference-time methods such as retrieval, context selection, and forecast correction. Benchmarks such as GIFT-Eval and fev-bench test performance across domains, while applications include energy, traffic, health, and finance. Results are often competitive when data are scarce or patterns transfer, but performance can lag specialized models when covariates, domain shifts, extreme events, or calibration matter; when and how to adapt a pretrained model remains an open question.

Latest papers

CardsList

96 papers14 papers in the last four weeks, up 27% on the four weeks before. 0.1% of all new papers.

  1. Revisiting the Generic Transformer: Deconstructing a Strong Baseline for Time Series Foundation Models

    Feb 6, 2026Yunshi Wen, Wesley M. Gifford, Chandra Reddy +3Time Series TransformersTime-Series Foundation Models

  2. Time Series Foundation Models for Process Model Forecasting

    Dec 8, 2025Yongbo Yu, Jari Peeperkorn, Johannes De Smedt +1Zero-Shot Time Series ForecastingProcess Mining

  3. Re(Visiting) Time Series Foundation Models in Finance

    Nov 23, 2025Eghbal Rahimikia, Hao Ni, Weiguan WangTime Series ForecastingTime-Series Foundation Models

  4. TempoPFN: Synthetic Pre-training of Linear RNNs for Zero-shot Time Series Forecasting

    Oct 29, 2025Vladyslav Moroshan, Julien Siems, Arber Zela +2Zero-Shot Time Series ForecastingTime-Series Foundation Models

  5. Kairos: Toward Adaptive and Parameter-Efficient Time Series Foundation Models

    Sep 30, 2025Kun Feng, Shaocheng Lan, Yuchen Fang +6Rotary Positional EmbeddingsZero-Shot Time Series Forecasting

  6. ChronoSteer: Bridging Large Language Model and Time Series Foundation Model via Synthetic Cross-Modal Alignment Dataset

    May 15, 2025Chengsen Wang, Qi Qi, Zhongwen Rao +2Cross-Modal LearningMultivariate Time Series Forecasting