Recruit the role before you recruit the name.

A recruiting board should answer a role question before it ranks a player. Start with the source-listed roster, identify the workload that is already represented, and then compare candidates against a clear priority.

The recruiting fit board uses the 5,459 source-listed player rows in the current 2026–27 roster release. It groups source position labels into guard, wing and big roles, then lets you choose creation, shooting, rebounding, defensive events or workload as the priority.

Start with the role room

The selected program’s role cards show listed players and the prior minutes attached to exact source IDs. Returning and incoming minutes stay separate, and minutes attached to ambiguous source rows remain in an explicit unclassified bucket. They describe what the release represents; they do not prove that a player left, transferred, is eligible or will play the same role next season.

Read the fit score as a sorting aid

Candidates are filtered outside the selected program and require a prior production record plus the chosen minimum workload. The board combines a 70% percentile for the selected skill priority with a 30% percentile for prior minutes. Creation blends assists and points; shooting blends true shooting and effective field goal percentage; defense blends steals and blocks. Missing source fields remain missing rather than receiving a guessed value.

Percentiles make the units comparable, but they do not make the score a fitted transfer model. A high workload can reflect a different role, and a strong shooting percentile can come from a small sample. Open the source roster, the prior game log and the dated announcement before making a recruiting conclusion.

Keep the evidence boundary visible

The recruiting file currently contains selected school announcements, while the roster release provides the broader source frame. The fit board joins those layers only through exact source IDs when a player link exists. It does not infer a portal transaction from a different program label or turn a roster observation into an availability claim.

Once a shortlist is built, use the player profile browser, dated recruiting file and program dossier together. The model forecasts 1,579 games for 2026–27, but recruiting evidence remains a staff verification layer outside the primary forecast.

Data attribution: SportsDataverse, CC BY 4.0. See the source receipts and model notebook.