Forecasting agents improve over time using feedback without retraining
Self-Evolving Time-Series Forecasting Agents with Episodic Memory and Online Policy Learning
Machine Learning
Summary
Forecasting future events based on past data is a common challenge, but most forecasting tools don't learn from their past mistakes. The authors introduced a system called FASE that remembers previous forecasts and learns from the actual outcomes to improve future predictions. This approach allows forecasting tools to get better over time without changing their internal settings. Tests showed that FASE performs better than other methods, especially as it receives more feedback.
What this means in practice
- •For financial analysts: Improve online time-series forecasts in stock or market data by incorporating feedback from realized outcomes without retraining models.
- •For energy grid operators: Provide adaptive power demand forecasts that evolve with real-time feedback to better manage supply and distribution.
Authors
Junyi Wang, Yilin Wang, Wen Wu, Chao Zhang
Abstract
LLM-based agents are increasingly used for time-series forecasting because they can organise contextual information, perform multi-step analysis, and guide the sequence of actions required to complete forecasting tasks. Most existing agents focus only on the current forecasting instance. However, in real-world deployments, forecasting commonly operates online, with new forecasts issued from the currently available history as the forecast origin advances and the ground-truth targets of earlier instances progressively become available. These targets provide feedback on the actions taken in earlier instances, yet existing agents generally do not preserve or utilise this information to adapt their subsequent actions. To address this limitation, we introduce FASE, a Feedback-Aware Self-Evolving forecasting agent that converts such feedback into task-specific experience for subsequent forecasting instances. FASE combines episodic memory, which retrieves relevant completed instances, with online policy learning, which summarises the feedback accumulated across instances into ranking guidance. The proposed framework is evaluated on 29 dataset configurations selected from the GIFT-Eval benchmark. Across these 29 configurations, FASE attains the strongest aggregate point forecasting performance among the evaluated methods and reduces the normalised MAE by 9.1% relative to the best individual foundation model baseline. The results further indicate that the cumulative advantage of FASE increases as delayed feedback accumulates. Together, these findings demonstrate that FASE can continually self-evolve through feedback from completed forecasting instances without updating the parameters of the LLM.