The gist of nonlocality à la Bell is that there are conditional probability distributions $p(ab|xy)$, where $a,b$ and $x,y$ denote possible measurement outcomes and possible choice of measurements, respectively, which do not admit a "local realistic explanation", that is, a decomposition of the form
$$p(ab|xy) = \sum_\lambda p(\lambda) p_\lambda(a|x)p_\lambda(b|y).\tag 1$$
Some quantum systems produce probability distributions which behave like this (i.e. cannot be decomposed like (1)), hence quantum mechanics not being local realistic etc.
One major assumption in (1) is that you still assume that the choices of measurement bases, $x,y$, are not correlated. This means that you assume that Alice and Bob "choose freely" how to interact and observe the systems they are given. "Superdeterminism" is what you get lifting this assumption. If you assume that Alice and Bob's choices of measurements were also determined beforehand (e.g. maybe they organised before moving apart and decided on what they'll measure), then the problem becomes trivial: any possible observation can be explained deterministically.
Let me stress that this is not even really about quantum mechanics. I'm saying that any possible observation (in the two-party scenario we consider here), quantum mechanical or not, can be explained deterministically with a "superdeterministic hidden variable".
The trivial "proof" of this is that, if you consider the measurement choices $x,y$ as also determined beforehand, you effectively remove the locality aspect from the equation.
If the measurement choices are also correlated/chosen beforehand, then the fact that Alice and Bob are spatially separated is effectively vacuus. You just have a joint probability distribution $p(abxy)$, which you can always describe using some deterministic "classical" theory.
Of course, the downside of such a "superdeterministic explanation" is that it involves some sort of "cosmic conspiracy theory". Sure, it might just so happen that the weird correlations observed by Alice and Bob are actually due to the way they chose to measure their apparatuses being determined a priori together with some corresponding deterministic outcome, but how would that actually happen? If you were to devise a theory explaining that, it would be yes fully classical, but also likely quite weird and convoluted.