Cast-R1: Learning Tool-Augmented Sequential Decision Policies for Time Series Forecasting DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06
Xiaoyu Tao, Mingyue Cheng, Chuang Jiang, Tian Gao, Huanjian Zhang, Yaguo Liu · 2026
Read Cast-R1: Learning Tool-Augmented Sequential Decision Policies for Time Series Forecasting DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 by Xiaoyu Tao, Mingyue Cheng, Chuang Jiang, Tian Gao, Huanjian Zhang, Yaguo Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-13802.txt.
Description
Recent time series forecast approaches use language models to run forecast tools on their own and reason about time patterns from evidence, but most rely on big models in training-free workflows. It remains open whether smaller models can learn those interaction and reasoning skills. Cast-R1, an 8-billion-parameter forecast model, is trained through path distillation and multi-turn reinforcement learning with structured memory. A forecast agent harness with analysis and prediction tools generates multi-turn interaction paths using a teacher model for fine-tuning.
Group relative policy optimization across forecast episodes then refines the model, with memory keeping evidence and middle forecasts. Rewards for accuracy and time structure guide tool selection, model routing, and forecast refinement, while a difficulty-aware curriculum combines sample filtering and step ordering. Experiments across data sets show Cast-R1's effectiveness. Xiaoyu Tao leads the work with six co-authors.
Catalog record
| CREATOR | Xiaoyu Tao, Mingyue Cheng, Chuang Jiang, Tian Gao, Huanjian Zhang, Yaguo Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 71 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-13802