It seems that players who are consistently winning, based on their AllinEV, are generally running below ev and players who are consistently losing, based on the AllinEV, are generally running above ev.

This is most clear in one of the last N50 CAP videos where the coach has a database of 1 Milion hands on one limit on one site. In the video series he takes the stats of a number of losing players and, although it is not the point of the video, we notice how they, as a group, are running above ev. We also see his graph and he is running below ev. I have tried the same on my data base. I took a number of losing players and created an alias. This alias runs above ev.

IMHO there are 3 possible reasons:

a) it is just a coincidence that winning players are running below ev and losing players are running above;

b) there is some inherent flaw in the AllinEV calculation. We know that this is true to some extent in multi way pots;

c) all the mad men saying that the poker rooms are helping the fish are not so mad.