There seems to be a bit of a consensus on the bottom, but not on the top (highest variance)
I acknowledge what you are saying with LHE vs NLHE, but what this assumes is that a player would try to play the 2 games in broadly the same way. Which they wouldnt. Or wouldnt after time, when they realised it was wrong to do so.
If a good NLHE player (with a $4k BR) sits down at a $200 buy-in every day, and a good LHE player (with a $4K BR) sits down at a $200 buy-in every day, then over time the limit players graph (std dev) will be a lot more steady IMO. I think you generally talked yourself around to this anyway by the end of that paragraph....i think
With Stud you have got to argue it from the informational point of view. In theory you can get to see 35 out of 49 cards. In most hands you get alot of solid information to work with, which when combined with information from betting patterns has got to mean this is not near the top. Indeed in HORSE I acknowledge it is a pain in the butt, and is in fact my second least favourite rotation.
With PLO, I would argue that the small margins on hands are actually good. A lot of hands are chopped between good players, where slightly worse players might miss out on the splits over time. For me this is abit of a mid-tabler.