Time series forecasting proves unreasonably difficult as sophisticated AI models, foundation models, and LLMs fail to consistently beat simple statistical baselines like Seasonal-Naive and MSTL across benchmark datasets; complex models trend in wrong directions except on stable seasonal data.

The Unreasonable Difficulty of Time Series Forecasting

Preview I’ve been thinking recently about what makes time series forecasting problems so difficult compared to other sequence learning tasks or IID Machine Learning problems. First some motivating baselines illustrating the difficulty of the forecasting problem. I ran a series of models statistical (Naive, AutoARIMA, Theta, MSTL, Seasonal-Naive) linear/transformer neural (DLinear, NLinear, PatchTST) a gradient-boosted tree (LightGBM) zero-shot foundation models (Chronos, TimesFM, TTM) “AI” LLM p...