This is what runs in the staff room. Type a name, set position thresholds (Olen-style), filter by cohort. Add up to three players to compare side-by-side. Every number is real model output from BB Combined R/Y/G v1.
| Player | Ht | OReb% | DReb% | Stl% | Blk% | USG% | Risk | Trajectory | Combined | R/Y/G |
|---|
Up to 3 players. The math underneath is the same model — different inputs land at different signals.
Returning college players have prior-year college stats. The trajectory model uses those + the player's scheme context + behavioral signals to predict next-season improvement or decline. That's the model behind the table above. Validated on past cycles where the next-season outcome already played out.
HS recruits with no college stats yet use a different read: HS recruiting class quality (247 composite + national rank), HS scheme experience (asked in the survey), position archetype matching against historical comparables, and program-fit modeling. The output is a development band, not a single trajectory score. Same architecture — different inputs.
D2, NAIA, international players are on the Q3 2026 roadmap — pipelines for those data sources are being built. Today the search is D1-live; coverage expands as data lands.