Using Random Variables to Predict Experimental Outcomes

TL;DR

This paper demonstrates that certain experiments, modeled as Bernoulli trials with success probability over 0.5, can be predicted more accurately than their success probability suggests, revealing surprising predictive capabilities.

Contribution

It introduces a novel insight that some Bernoulli trials allow for prediction with accuracy exceeding their success probability, challenging conventional assumptions.

Findings

Predictive accuracy surpasses success probability p in certain Bernoulli trials

Experiments with p > 0.5 can be predicted with probability > p

Highlights potential for improved outcome prediction in probabilistic experiments

Abstract

We shall show in this paper that there are experiments which are Bernoulli trials with success probability p > 0.5, and which have the curious feature that it is possible to correctly predict the outcome with probability > p.