When attempting to predict a players true talent level, you can't cherry pick someone's best streak or worst streak to determine their true talent level. You have to find a large enough sample to avoid over counting both hot and cold streaks. Even a full season is really a small sample size. So we have to attempt to parcel out data to extrapolate future results.
The ACC or second half shooting numbers for Starling are getting referenced a lot, and for two probable reasons.
One is recency bias. His most recent large sample he shot better, so that's what he'll do going forward. Not necessarily true, but more valuable than cherry picking his first 16 games before his hot streak and ignoring the second half...
The other more relevant reason is that he clearly had something wrong with him the first half. His shot just looked funny and he seemed resistant to shooting from 3 point range. Then, either his shoulder injury healed, his confidence improved, or they repaired his hitch in practice. Perhaps some combination. But whatever it was, he shot better and looked better shooting after about half the season.
From that point in the season Starling shot more often, and he shot better. His 3pt attempts went from 2.8 to 5.7 while his success rate went from .244 to .363.
This, to me, looks more like something actually changed to make him better in those last 16 games and not merely recency bias or cherry picking a hot streak.