TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecasting

TL;DR

TimeBridge introduces a novel framework that effectively handles non-stationarity in long-term multivariate time series forecasting by combining attention mechanisms to model both short-term dependencies and long-term cointegration, achieving state-of-the-art results.

Contribution

The paper presents TimeBridge, a new method that explicitly addresses non-stationarity's dual impact on short-term and long-term modeling in time series forecasting.

Findings

Achieves state-of-the-art performance in short-term and long-term forecasting.

Excels in financial forecasting on CSI 500 and S&P 500 indices.

Demonstrates robustness and effectiveness across diverse datasets.

Abstract

Non-stationarity poses significant challenges for multivariate time series forecasting due to the inherent short-term fluctuations and long-term trends that can lead to spurious regressions or obscure essential long-term relationships. Most existing methods either eliminate or retain non-stationarity without adequately addressing its distinct impacts on short-term and long-term modeling. Eliminating non-stationarity is essential for avoiding spurious regressions and capturing local dependencies in short-term modeling, while preserving it is crucial for revealing long-term cointegration across variates. In this paper, we propose TimeBridge, a novel framework designed to bridge the gap between non-stationarity and dependency modeling in long-term time series forecasting. By segmenting input series into smaller patches, TimeBridge applies Integrated Attention to mitigate short-term non-stationarity and capture stable dependencies within each variate, while Cointegrated Attention preserves non-stationarity to model long-term cointegration across variates. Extensive experiments show that TimeBridge consistently achieves state-of-the-art performance in both short-term and long-term forecasting. Additionally, TimeBridge demonstrates exceptional performance in financial forecasting on the CSI 500 and S&P 500 indices, further validating its robustness and effectiveness. Code is available at https://github.com/Hank0626/TimeBridge.