Zero shot time series forecasting requires clear evidence reporting

Tracing the Evidence Behind Zero-Shot Time-Series Forecasting: A Source-First Taxonomy and Audit Framework

Machine Learning

Summary

Forecasting future values in time-series data without updating models specifically for each new task is called zero-shot forecasting. The authors highlight that such forecasting can draw on very different sources of knowledge, like frozen language models, pretrained time-series models, or external memory retrieval, even though all claim to work without updates. They propose a system to clearly classify and report where the forecasting methods get their information from, so that results are more understandable and comparable. This framework aims to make forecasting benchmarks fairer by showing what prior knowledge and resources models actually use.

What this means in practice

  • For forecasting engineers: Design clearer benchmarking systems that require forecast models to state their data sources and context assumptions explicitly.
  • For data platform architects: Build forecasting infrastructure that enforces transparency about model evidence use and resource budgets for fair model comparisons.

A position paper. It proposes an approach and reports no results.

Authors

Delun Kong, Wanyun Ling, Chenxi Liu, Ziyue Li

Abstract

Zero-shot time-series forecasting (TSF) is often described as forecasting without target-specific parameter updates, but that training-status condition does not specify what evidence the system may use. A frozen language model prompted with serialized values, a time-series model pretrained on broad forecasting corpora, and a retrieval-augmented forecaster may all satisfy the no-update condition while drawing on different transferable evidence. This paper argues that zero-shot TSF should therefore be governed as an evidence-access claim. We propose a source-first taxonomy that separates three primary evidence sources---frozen LLM prior reuse, parametric time-series pretraining, and retrieval-augmented external memory---from the architectures that implement them. After the source is identified, four additional audit questions remain: task interface, forecast object and scoring, prediction-time context, and resource budget. The resulting agenda is to make zero-shot leaderboards auditable by reporting evidence boundaries and interface assumptions alongside scores, so that benchmark progress reflects transferable forecasting capability rather than undisclosed changes in context, memory, or budget.
View PDFOpen arXiv