Event signature transfer improves forecasting accuracy without retraining
Event Signature Transfer: Model-Agnostic Forecast Scenario Construction from Historical Events
Artificial IntelligenceMachine Learning
Summary
Sometimes forecasters know an important event is coming but not how it will unfold. The authors offer a way to turn past event patterns into forecast scenarios without having to retrain or change existing models. Their method works with any forecast that produces quantiles and adjusts past event signals onto current forecasts while keeping the forecast’s internal structure. Tests on real and synthetic data show this approach can significantly reduce forecasting errors compared to other methods. However, it creates possible event scenarios rather than predicting their likelihood.
What this means in practice
- •For demand planners: Construct realistic demand shock scenarios from past events without retraining forecasting models to improve planning.
- •For energy grid operators: Generate explicit disruption scenarios from historical events to anticipate grid load changes without modifying forecasting systems.
Authors
Karthik Sridhar, Aaditya Jain, Murari Mandal, Saurabh Deshpande
Abstract
Forecasters often know an event is imminent but not the shape, size, or timing of its effect. We introduce Event Signature Transfer (EST), a training-free, model-agnostic operator that turns a completed past event into an explicit forecast scenario. EST removes a source event's own trend and seasonality, then scales and retimes the remaining event signature onto a native forecast, preserving the forecast's linked structure and reducing to it exactly at zero strength. Because it reads only output quantiles, EST applies to any quantile forecaster, with no training, no model internals, at transfer time. Across twelve real episodes and ten synthetic scenarios on Chronos-2, TimesFM-2.5 and Toto-2.0, manually configured EST reduces real-episode WQL by 21.7-90\% in-sample. On Chronos-2, it leads eleven of twelve matched comparisons against covariate conditioning, activation editing and raw replay. The operator builds a scenario; it does not estimate its likelihood.