A violation of a Bell inequality means that, for the apparatus being considered, no local hidden variable can reproduce the observed correlations (so, roughly speaking, that you cannot reproduce the results with only classical physics).
This is not always true: many predictions of quantum mechanics are easily reproduced with classical theories. Nor you should expect otherwise, given that most systems we interact with in everyday life behave "classically", even though they are really "quantum mechanical" underneath.
The point of Bell's theorem is that quantum mechanics, in some circumstances, gives outcomes that cannot be reproduced via a local hidden variable theory.
There is therefore nothing strange in a graph like the one you show, and you should expect that kind of behaviour to be the norm rather than the exception.