MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning

TL;DR

MemCast tackles time series forecasting by introducing an experience-driven, memory-augmented framework. It constructs a hierarchical external memory from training data, separating historical patterns, reasoning wisdom, and general laws to support experience-conditioned reasoning during inference. A dynamic confidence adaptation mechanism enables continual memory evolution while preserving train–test separation. Across diverse datasets, MemCast consistently outperforms strong baselines, especially on volatile series, demonstrating the value of explicit experience accumulation and memory-guided reasoning for robust TSF.

Abstract

Time series forecasting (TSF) plays a critical role in decision-making for many real-world applications. Recently, LLM-based forecasters have made promising advancements. Despite their effectiveness, existing methods often lack explicit experience accumulation and continual evolution. In this work, we propose MemCast, a learning-to-memory framework that reformulates TSF as an experience-conditioned reasoning task. Specifically, we learn experience from the training set and organize it into a hierarchical memory. This is achieved by summarizing prediction results into historical patterns, distilling inference trajectories into reasoning wisdom, and inducing extracted temporal features into general laws. Furthermore, during inference, we leverage historical patterns to guide the reasoning process and utilize reasoning wisdom to select better trajectories, while general laws serve as criteria for reflective iteration. Additionally, to enable continual evolution, we design a dynamic confidence adaptation strategy that updates the confidence of individual entries without leaking the test set distribution. Extensive experiments on multiple datasets demonstrate that MemCast consistently outperforms previous methods, validating the effectiveness of our approach. Our code is available at https://github.com/Xiaoyu-Tao/MemCast-TS.

Figures (8)

• Figure 1: Comparison of training-based, training-free, and memory-enhanced LLM forecasting approaches.
• Figure 2: Overview of MemCast as an LLM-driven time series forecasting framework that constructs hierarchical memory from the training set via experience accumulation for experience-conditioned reasoning on the testing set.
• Figure 3: Ablation on dynamic confidence adaptation. Dynamic confidence adaptation leads to consistently lower errors compared with the variant without adaptation.
• Figure 4: Exploration on aggregation strategies. The full model outperforms baselines, confirming that active selection via semantic consistency surpasses passive aggregation.
• Figure 5: A detailed case study on the oil temperature (OT) forecasting task, illustrating experience-conditioned reasoning via constructed memory to effectively handle out-of-distribution thermal shifts.