To See Far, Look Close: Evolutionary Forecasting for Long-term Time Series
This work addresses a computational bottleneck for researchers and practitioners in time series forecasting by enabling a single model to handle multiple horizons without re-training, representing a paradigm shift rather than an incremental improvement.
The paper tackles the inefficiency of Direct Forecasting in long-term time series forecasting, where re-training is needed for different horizons, and shows that their Evolutionary Forecasting paradigm, using models trained on short horizons, outperforms direct long-horizon training with robust asymptotic stability in benchmarks.
The prevailing Direct Forecasting (DF) paradigm dominates Long-term Time Series Forecasting (LTSF) by forcing models to predict the entire future horizon in a single forward pass. While efficient, this rigid coupling of output and evaluation horizons necessitates computationally prohibitive re-training for every target horizon. In this work, we uncover a counter-intuitive optimization anomaly: models trained on short horizons-when coupled with our proposed Evolutionary Forecasting (EF) paradigm-significantly outperform those trained directly on long horizons. We attribute this success to the mitigation of a fundamental optimization pathology inherent in DF, where conflicting gradients from distant futures cripple the learning of local dynamics. We establish EF as a unified generative framework, proving that DF is merely a degenerate special case of EF. Extensive experiments demonstrate that a singular EF model surpasses task-specific DF ensembles across standard benchmarks and exhibits robust asymptotic stability in extreme extrapolation. This work propels a paradigm shift in LTSF: moving from passive Static Mapping to autonomous Evolutionary Reasoning.