Seems to me, Glen, that you and Shan each take "percentage of hits" to be a meaningful number. I'm no mathematician, but percentage of hits does not in any way measure the likelihood of a prediction method having any merit (I am assuming that that is the question of overriding interest in most of our discussions on this board). Percentage of hits can easily be arranged by any predictor to have any number between one and zero, without any significance. If Shan were to reduce his magnitudes, OR expand the geographical area, OR lengthen the time frame, for instance, he would automatically raise his percentage of hits.

Unfortunately, a statistical analysis to determine if a prediction method actually has merit is considerably more difficult than determining percentage of hits. Roger has apparently done that analysis (previous post), and has determined that Shan's predictions have not yet beens shown to be any better than chance.

Mike Williams in Arroyo Grande, CA USA