Yeah, the combination of barrel twist, bullet length, bearing surface, BC, bullet center mass, bullet shape, velocity, etc., etc., almost seem like black magic, but we all know it's just a matter of actual physics. Still, the debate goes on as to how much of a percentage each element plays to make the whole performance envelope. Most would agree that all of them are important, and it's pretty much guaranteed that the percentage factor changes a bit amongst them as variables amongst those elements change.
This is probably why most everyone has to develop their personal loads for their rifle in steps to find the most ideal combination...nothing unusual in that. That Berger formula is probably one of the best I've seen, but if it was exact/perfect, there would be no need of as much load development. But then, that's what makes reloading a fun challenge.
I recently experienced an interesting issue when shooting some 220g subs from a 16", 1:10 twist AR15 barrel. I bought this barrel for shooting 110g supers which it does with excellent results. Using the Berger formula, I need an 1:8.5 twist for true bullet stabilization. I shot over a dozen shots at 100 yards with this 220g sub ammo. MOA was between 3-5 MOA with a red dot and amazingly no keyholing. I know that's not the do-all-end-all of a barrel twist vs. bullet type laboratory study, but it did show me how "shifty" some of these elements can be.