Report the Floor: A Training-Free Conformal Interval Is a Mandatory Baseline for Probabilistic Time-Series Forecasting
arXiv:2606.09473v1 Announce Type: cross Abstract: Probabilistic forecasters are increasingly learned, yet the baselines they are compared against are often weak or omitted. We show that the simplest possible conformal interval - a last-value point forecast wrapped in a finite-sample split-conformal residual quantile, with no parameters and no training - is a far stronger baseline than its near-total absence from recent learned-forecasting and conformal-time-series comparisons would suggest. In one-step-ahead online forecasting across 2,217 real series from nine public sources (Monash, LOTSA, t
This research highlights a critical methodological flaw in current leading AI research for time-series forecasting, suggesting many reported advancements might be overstated.
A strategic reader should care because proper baseline evaluation is fundamental to assessing the true progress and reliability of AI models, particularly in critical applications like predictive maintenance, financial modeling, or supply chain optimization.
The focus should now shift towards more rigorous benchmarking against robust, simple baselines, potentially slowing down the perceived pace of generative AI advancements in certain time-series forecasting domains until methodological standards improve.
- · AI researchers promoting robust statistical methods
- · Industries relying on reliable time-series forecasts
- · Over-hyped probabilistic forecasting models
- · Early adopters of unverified AI forecasting techniques
Increased scrutiny on reported performance metrics for new probabilistic forecasting models.
A potential re-evaluation of 'state-of-the-art' models in time-series forecasting, leading to a more grounded understanding of their true capabilities.
Resource reallocation within AI research towards developing more fundamentally robust and truly superior forecasting methods, rather than incremental improvements over weak baselines.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG