Is it normal if I'm running 11.500 big blinds below EV in all-in pots over 700.000 hands? What are the chances for this to occur? My standard deviation BB/100 is about 100.
That is a relatively small sample size. I wouldn't worry about it too much if you're thinking that you may have a leak of some sort. I'd say you need tens of thousands of hands before that all-in number starts representing the real EV value. Hands rarely get all in, that's why the sample size needs to be so large.
EDIT: There is a correlation. Leak is present.