A project that might be interesting would be look at player performance by game or ideally by stint on court and divide the results into 3 or 5 levels and then look at the distributions for players, year to year and coach to coach. Who is achieving more peak performances?
When projecting team wins these trends matter, maybe especially for guys changing coaches or still learning to optimize themselves. Yearly stats are "big data" but trying to look at and understand the quality of lineup management and player utilization within them could help in projecting where the average is going to go next. I am not aware of any of the leading team win projection models looking at player optimization / coaching impact. It is implicit in the average data for most players who stay with the same coach but even in these cases it might be helpful to try to ascertain which player-coach pairs are getting better or worse. Player-coach pairings may not all be going in same direction and don't matter the same from player to player or year to year. Player minute projections could be affected within positions or roles significantly, depending on how likely, quickly and intensely a coach is to respond to changes in performance results.