Does Scaling Law Apply in Time Series Forecasting?

2025-05-15

AI Summary

This work challenges the prevailing assumption that larger models are inherently superior for time-series forecasting, questioning the universality of scaling laws in this domain. To address this, we propose Alinear—a hyper-lightweight model that abandons attention mechanisms entirely. Alinear introduces two novel components: (i) horizon-aware temporal adaptive decomposition, which dynamically disentangles trend and seasonal components according to forecast horizon, and (ii) progressive frequency attenuation, which selectively suppresses high-frequency noise while preserving predictive signals. These are integrated with efficient linear modeling for scalable, multi-horizon forecasting. Furthermore, we design a parameter-aware evaluation framework that quantifies the differential contributions of trend and seasonal components to overall prediction accuracy. Evaluated on seven benchmark datasets, Alinear achieves state-of-the-art performance across short- to ultra-long-term horizons—despite using less than 1% of the parameters of leading large models—demonstrating both the feasibility and superiority of compact, principled modeling in time-series forecasting.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Large Vision Models
• Natural Language Processing: (Large) Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Rapid expansion of model size has emerged as a key challenge in time series forecasting. From early Transformer with tens of megabytes to recent architectures like TimesNet with thousands of megabytes, performance gains have often come at the cost of exponentially increasing parameter counts. But is this scaling truly necessary? To question the applicability of the scaling law in time series forecasting, we propose Alinear, an ultra-lightweight forecasting model that achieves competitive performance using only k-level parameters. We introduce a horizon-aware adaptive decomposition mechanism that dynamically rebalances component emphasis across different forecast lengths, alongside a progressive frequency attenuation strategy that achieves stable prediction in various forecasting horizons without incurring the computational overhead of attention mechanisms. Extensive experiments on seven benchmark datasets demonstrate that Alinear consistently outperforms large-scale models while using less than 1% of their parameters, maintaining strong accuracy across both short and ultra-long forecasting horizons. Moreover, to more fairly evaluate model efficiency, we propose a new parameter-aware evaluation metric that highlights the superiority of ALinear under constrained model budgets. Our analysis reveals that the relative importance of trend and seasonal components varies depending on data characteristics rather than following a fixed pattern, validating the necessity of our adaptive design. This work challenges the prevailing belief that larger models are inherently better and suggests a paradigm shift toward more efficient time series modeling.

Problem

• Challenges scaling law necessity in time series forecasting
• Proposes ultra-lightweight model with adaptive decomposition for efficiency
• Questions performance gains from exponentially increasing model sizes

Innovation

• Ultra-lightweight model with k-level parameters
• Horizon-aware adaptive decomposition mechanism
• Progressive frequency attenuation strategy