Every college prospect ranked by projected career peak on the 2K scale, built from box scores, shot profile, age and recruiting.
Projected peak places a prospect's model score on the real distribution of career peaks, so the top prospects land where top careers actually land. The shaded band runs from the conservative expected peak to a ceiling that 1 in 10 similar historical prospects reached. Pick is the actual draft slot. Click a player for the full readout.
A few profiles the model tends to overrate are pulled toward where scouts had them (ESPN's pre-draft big board, else the draft slot): unathletic bigs; high-usage wings who rarely get to the rim and shoot poorly inside the arc; and bigs whose case rests on shooting touch or experience, who also take a small extra cut. Freshman bigs with elite numbers are left alone. The pull only moves a player down, never up. Across the 2009-21 backtest it improved 12 classes and hurt none.
A blend of ridge regression, an ordinal tier model and a two-stage model (chance of an NBA career times peak if he sticks), trained on 2008-2022 college prospects. The target is each player's career peak on the 2K scale: a hand-curated grade (with injury-free peaks for players like Markelle Fultz), filled in from NBA production and honors where there's no grade or a young career has outgrown it. Players who never reached the NBA count as 65; careers still in progress are projected from their first seasons.