A ball generally has two or many preferred spin axes, those axes around which the core is stable when rotating. One is always through the pin. On symmetric balls, the ball is also stable 90* from the pin in most any direction (i.e., a spot on the equator when the pin is at the north pole). But it won't necessarily want to migrate to an axis like that because the preference isn't that strong.
On an asymmetric ball, the secondary PSA is at the MB mark, and since it is significant, the ball will try to satisfy both the PSA at the pin and at the MB mark. From what I understand from some of Track's drill sheets is that after the pin has migrated into a stable position, the MB will do the same.
The CG does no such thing. It is not at a preferred spin axis (unless it's a 0" or 6 3/4" pin length and then it's just coincidence), so it doesn't really influence the way the ball turns over as it rolls. I suppose that a high enough top weight could have that effect, but I think that a ball with that kind of top would be difficult to drill legally.
The MB mark isn't always 6 3/4" from the pin, there have been a few balls with longer or shorter pin-to-MB distances. Some early Ebonite BOMB marks were 5" from the pin, and Visionary's Executioner had the mark 11" (I think) from the pin. Admittedly, I don't know how they spun on a deTerminator, so the true PSA may have been 6 3/4" from the pin through the BOMB or neck. I think 6 3/4" is the strongest place to have a PSA other than the pin (i.e., the asymmetry isn't as strong on those balls with other distances) because being 90* from the pin, the PSA is the same distance from both the pin and the anti-pin (180* from the pin), not closer to one or the other.
SH