Universally consistent predictive distributions

TL;DR

This paper introduces simple, universally consistent probability forecasting methods that are valid for small samples under the assumption of IID observations, ensuring reliable predictions across diverse scenarios.

Contribution

It presents new universally consistent forecasting procedures that maintain small-sample validity under IID assumptions, advancing predictive reliability.

Findings

Procedures are universally consistent for IID data.

Methods ensure small-sample validity.

Applicable across various prediction tasks.

Abstract

This paper describes simple universally consistent procedures of probability forecasting that satisfy a natural property of small-sample validity, under the assumption that the observations are produced independently in the IID fashion.