On predictive causality in the statistical analysis of a series of events