ExtremeConformal: Reliable Extreme Conformal Prediction Intervals for High-Impact Events

A new framework for extreme conformal prediction that provides informative prediction intervals at the high-confidence levels for which classical conformal methods fail. In applications with potentially high-impact events, a very high level of confidence is often required for predictions. If that level is too large relative to the amount of data used for calibration, classical conformal methods provide infinitely wide, thus, uninformative prediction intervals. Our extreme conformal procedure bridges extreme value statistics and conformal prediction to provide reliable and informative prediction intervals with high-confidence coverage, which can be constructed using any black-box extreme quantile regression method. A weighted version of the approach can account for nonstationary data. The methodology was introduced in Pasche, Lam, and Engelke (2025).

Links

source code: released soon on https://github.com/opasche/ExtremeConformal

Related article:: https://arxiv.org/abs/2505.08578 (PDF) (In press for publication in the special issue ‘Bridging Heavy Tails and Artificial Intelligence’ of Extremes.)