Method

Transfer Portal Rankings

Each player is scored with no destination information. Players with a qualifying 2025-26 season use a ridge regression trained on transfers and stayers, plus a gradient-boosted correction. Players below the minutes cutoff use a model that shrinks toward typical outcomes. The projection is next season's BartTorvik BPM if he plays (5+ games, 10%+ of team minutes).

Tested on 2020-26 transfers, the average miss is 2.10 BPM. The projected range is that model's historical 10th to 90th percentile of misses. Market is the chance he lands at a high-major, mid-major or low-major program, and on 1,437 real 2026-27 moves it scored AUC 0.89. A player with a 50% or better high-major chance counts as High-Major; anyone else counts as whichever of Mid-Major or Low-Major is more likely. Positions for UAA players are assumed from their stats.

High School Rankings

The board starts from my 17U EYBL and 3SSB model (5+ GP, 16+ MPG, 200+ MIN). The Model Score is built on BPM, with shooting, rebounding, defense and creation weighted by what went with team winning across 101 EYBL and 3SSB teams. Level uses the 247 Composite: top 100 is high-major.

UAA publishes no minutes, offensive rebounds or fouls. UAA BPM is calculated with a method rebuilt from 3SSB BPM, and tested on 3SSB it tracked real BPM at r 0.87. UAA scores carry a circuit adjustment of two steps below EYBL.

Data

BartTorvik, 247Sports, circuitstats.com, underarmournext.com and 17U EYBL and 3SSB box scores.