When the Uncertainty Is Bigger Than the Shock: Scenario Modelling for English Local Elections

A scenario model for the 2026 English local elections reveals that the model's calibrated uncertainty is significantly larger than the impact of any tested scenario. The strongest scenario, a +4pp challenger surge, moved the central estimate by only 13% of the median uncertainty band derived from historical forecast errors. The methodology involves using backtest residuals as an empirical distribution to bootstrap uncertainty, providing a data-driven way to understand potential forecast noise. This outcome emphasizes the need to compare an effect's size against its uncertainty to avoid false precision, as the modeled shocks were all contained within the historical noise.