CastFSR: A Fast--Slow--Reflect Agentic Reasoning Framework for Context-Aware Time Series Forecasting

2026-08-03

AI Summary

This work addresses the limitations of existing time series forecasting methods, which often lack explicit mechanisms to model contextual relevance, reason about its influence, and verify consistency with temporal and domain-specific constraints. To overcome these challenges, the authors propose CastFSR, a novel framework introducing a “fast–slow–reflect” three-stage proxy reasoning paradigm. In the fast stage, it generates data-driven initial forecasts; in the slow stage, it performs deep reasoning by integrating retrieved contextual evidence; and in the reflect stage, it iteratively refines predictions to ensure multidimensional consistency. CastFSR enables training-free inference, supports context-adaptive window selection, and can be efficiently deployed on small-scale large language models (LLMs). Extensive experiments demonstrate that CastFSR significantly outperforms state-of-the-art baselines across multiple public datasets, confirming its effectiveness and strong generalization capability.

Abstract

Time series forecasting is fundamental to decision-making in complex systems, where future dynamics are influenced not only by historical observations but also by evolving contextual features. Recent advances in large language models (LLMs) have extended forecasting beyond numerical extrapolation toward context-aware reasoning. However, existing approaches often lack explicit mechanisms to identify relevant contexts, reason about their impacts, and validate forecasts against temporal and domain constraints. In this work, we propose CastFSR, an agentic framework that formulates context-aware forecasting as a Fast--Slow--Reflect workflow. In fast thinking, CastFSR profiles observations and selects lightweight forecasters to construct a data-driven forecast prior. In slow deliberation, it retrieves contextual evidence, adaptively determines informative look-back windows, and reasons about how contexts reshape future dynamics. In reflection, it iteratively refines forecasts to ensure temporal, contextual, and domain consistency. CastFSR supports both training-free inference with off-the-shelf LLMs and efficient deployment through a two-stage SFT and reinforcement learning strategy that transfers its orchestration capability to compact LLMs. Extensive experiments on public datasets demonstrate that CastFSR consistently outperforms representative baselines. Our code is available at https://github.com/Xiaoyu-Tao/CastFSR.

Problem

• time series forecasting
• context-aware reasoning
• large language models
• temporal consistency
• contextual relevance

Innovation

• context-aware forecasting
• agentic reasoning
• fast-slow-reflect framework
• large language models
• time series forecasting