How can a superdeterministic theory ever match Quantum Mechanic's predictions? Superdeterminism says that we can observe violations from Bell's constraint for local hidden variable theories, if we assume the measurement choices are correlated, which they must be in a deterministic theory where all measurement choices are pre-decided at the Big Bang. It is correct that Bell's constraint gets violated if we assume correlations between measurement choices. But this does not automatically mean that the new hidden variable theory predicts the correlations exactly as predicted by quantum mechanics.
If we consider that all measurement choices are correlated since the Big Bang, then those correlations would effectively be randomised. So, wouldn't it be impossible for the hidden variable theory to predict the exact correlations as predicted by Quantum Mechanics?
This is because Quantum Mechanics only considers the Bell test in isolation, in order to predict the experimentally observed correlations. If two experimenters do two different Bells experiments, Quantum Mechanics predicts the same correlation both times. But a superdeterministic theory's predictions would depend not just on the experiment, but on everything that happened since the Big Bang.
 A: 
But this does not automatically mean that the new hidden variable theory predicts the correlations exactly as predicted by quantum mechanics.

No one is saying that. You can imagine an infinite number of superdeterminist theories that do not reproduce the predictions of quantum mechanics. But the idea is that if you choose the theory and the initial values of its hidden variables just right, then it will.

If we consider that all measurement choices are correlated since the Big Bang, then those correlations would effectively be randomised.

No. A deterministic theory does not contain things like "effectively random". They might look random to you because you do not know the underlying hidden variables, but they're not. That's the whole point of hidden variable theories.
The argument that you're implicitly making - that it sounds extremely far-fetched for this hidden variable theory to be set up in such a way that the world looks perfectly random to us and as if experimenters' choices of what to measure was unrelated to the experiment but actually isn't - is indeed the main intuitive objection most people have to superdeterminism.
Superdeterminism, in this sense, essentially posits that the world is controlled by a Cartesian demon that has set up the world in exactly such a way that our perceptions lead us to conclude that the world works one way (no local hidden variables) but in actuality works another way (extremely hidden local variables, so to speak). (Bonus: If you find it simpler to think that one person can be deceived like that rather than all humans, just go full Decartes and switch to solipsism so that the demon only needs to deceive you.) There are other epistemological issues with this concept, such as essentially saying that the worst case of the problem of induction is true and all the scientific laws that rely on the idea of the choice of measurement being independent of the thing being measured we have are just accidentally true and tomorrow the superdeterminist theory could behave differently, but this does not mean superdeterminism is inconsistent - just philosophically rather uncomfortable for many.
