But really, the goal would be to create a shortlist of viable candidates who could plausibly make a production leap with increased usage. That production bump would need to correlate with actual impact on winning, since almost any player’s raw numbers will rise if you simply give them more usage.

Is it possible to extrapolate the bench data to build a predictive model that identifies bench players who could become valuable extension or free-agency bargains, specifically, players whose production is likely to rise with increased usage? Obv, “breakouts” of that magnitude are hard to predict.