On the Distance from Calibration in Sequential Prediction

Mingda Qiao, Letian Zheng

We study a sequential binary prediction setting where the forecaster is evaluated in terms of the calibration distance, which is defined as the $L_1$ distance between the predicted…