Testing the theory over the second half of the 2006-2007 season, it was found that when there are 6 games or more being played, and the calculated number differs by at least 1 goal or more from the published number, there was good correlation between the predicted over or under, and the actual result. If the differential was less than 1 goal, or less than 6 games were played, no action was recommended. For days in which these criteria were met, the method was predictive 10 times out of 11.