Unfortunately, the sim probably won't have reliable data for AOA beyond critical values so training in that regime would be impractical.
This point comes up again and again in various posts, in an attempt to explain why stalls are not trained in the simulator. But I keep asking myself, is it really a valid reason?
Yes, I understand that no one has ever test flown the big jets in stall regimes and the exact numbers are not available. But we know what should happen, we know the direction where thing should be going when the pilot just keeps pulling the stick, and we can replicate this behaviour in a simulator. Yes, it will drop at a somewhat incorrect speed, some nasty effect may be overlooked, some extreme situations like going down at 60 knots with AOA 40 degrees will not be covered. But that is not the point - the idea is to provide the pilot with a training/evaluation tool that responds to clearly incorrect inputs (stick back) in a generally realistic manner (airspeed drops, cabin shakes, controls don't work, aircraft goes down) and requires basic corrective measures (stick forward) to fly again.
Wouldn't such a crude tool be better than having no tool at all, relying on mantras that "this aircraft cannot stall", on "computer protections", and a manual stall recovery procedure that is never trained in realistic conditions?