This comment in which gwern describes a 'hope function' - how, given a diffuse prediction of an event with known overall confidence, the confidence should change as specific opportunities for the event fail to occur.