Conformal prediction silently breaks under drift - and how to make it hold (opens in new tab)

Conformal prediction is the easiest way to put a calibrated uncertainty band around any model: wrap a point predictor, and you get intervals with a finite-sample coverage guarantee — no distributional assumptions. It's deservedly popular. There's a catch that bites in production: that guarantee is marginal and it assumes exchangeability. The moment your data drifts — almost any time series, any online-serving setting — exchangeability is gone, and split-conformal silently stops delivering the...