You can't project with the precision you seem to want, but there's been evidence for too long that minor league stats are indeed useful as a predictive tool. The key is to look at the right ones - ERA is a comparatively crude choice. Tim Melville's performance at Rochester in 2017 suggests to me that he might be marginally better in the majors than he showed yesterday, but not good enough to be allowed the additional innings necessary to demonstrate that very modest potential. His OPS-against in AAA would be about league average, after you factor in an unusually low BABIP - and league average pitching is not going to cut the mustard when brought up to the majors. IMO he would struggle to keep his ERA in the majors under 6, if he were foolishly allowed to get beyond SSS. I'd be very curious to hear someone in the Twins analytic squad say what better measures they use that made them see anything more in Melville than that. Maybe there's nothing, and the emergency start was driven by the other factors being discussed here.