Position: Beyond Model-Centric Prediction -- Agentic Time Series Forecasting

TL;DR

This paper argues that traditional model-centric time series forecasting is insufficient for adaptive, multi-turn decision contexts. It introduces agentic time series forecasting (ATSF), a framework that treats forecasting as an iterative decision-making process built from perception, planning, action, reflection, and memory, enabling tool interaction and experiential learning. It outlines three implementation paradigms—workflow-based design, agentic reinforcement learning, and the hybrid AgentFlow—and discusses opportunities such as system-wide tool evolution, heterogeneous learning integration, and human–agent collaboration, alongside challenges in memory design, tooling standards, multi-agent coordination, and deployment safety. By reframing forecasting as an agentic workflow, ATSF aims to align forecasting systems more closely with real-world decision processes and open avenues for flexible, interpretable, and decision-aware forecasting in dynamic environments.

Abstract

Time series forecasting has traditionally been formulated as a model-centric, static, and single-pass prediction problem that maps historical observations to future values. While this paradigm has driven substantial progress, it proves insufficient in adaptive and multi-turn settings where forecasting requires informative feature extraction, reasoning-driven inference, iterative refinement, and continual adaptation over time. In this paper, we argue for agentic time series forecasting (ATSF), which reframes forecasting as an agentic process composed of perception, planning, action, reflection, and memory. Rather than focusing solely on predictive models, ATSF emphasizes organizing forecasting as an agentic workflow that can interact with tools, incorporate feedback from outcomes, and evolve through experience accumulation. We outline three representative implementation paradigms -- workflow-based design, agentic reinforcement learning, and a hybrid agentic workflow paradigm -- and discuss the opportunities and challenges that arise when shifting from model-centric prediction to agentic forecasting. Together, this position aims to establish agentic forecasting as a foundation for future research at the intersection of time series forecasting.

Figures (2)

• Figure 1: Illustration of the core components and workflow of agentic time series forecasting.
• Figure 2: Illustration of the representative used tool for agentic time series forecasting.