Dynamic Regime-Aware Conformal Calibration for Reliable Economic Forecast Intervals under Multiple Distribution Shifts
Dynamic Regime-Aware Conformal Prediction (DRACP) addresses prediction interval reliability when economic data shifts due to covariate shift, concept drift, and latent regimes. It combines density-ratio weighting, kernel methods, and probabilistic regime awareness with a self-tuning controller. The method prioritizes consistent coverage (0.890 vs 0.90 nominal) over narrow intervals for dependable forecasting.