Read the matchup before you read the score.
A college basketball forecast is more useful when it tells you which possessions deserve attention. Start with shooting, turnovers, rebounding and free throws, then use pace and roster evidence to plan the questions you will ask in film.
Silvermine rates 366 programs from paired 2025–26 box scores and publishes 1,579 2026–27 forecasts in this edition. The model estimates efficiency and tempo from historical team performance. It does not copy a proprietary KenPom rating or claim to know a player's current health or role.
Four questions for every possession
Effective field-goal percentage gives made threes their extra value; a team that creates efficient shots can survive an ordinary free-throw night. Turnovers per possession measure how often an offense gives the ball away before it can create a shot. Offensive rebounding rate asks whether a missed shot becomes another possession. Free-throw rate describes pressure at the rim and the value of getting to the line.
Read offense and defense together. A high offensive rebounding rate can be muted by an opponent that ends possessions with defensive rebounds. A low turnover rate matters less if the team cannot generate efficient attempts. Each program dossier shows the numerator, denominator and number of games behind the displayed split.
Translate the forecast into a plan
The published 80% margin range is a reminder that the score is a distribution, not a promise. Open the game-plan workbench, choose the floor, and compare the matchup factors. The roster panel places source-listed 2026–27 movement beside the historical rotation. A returning-player match or different-program observation is a question for staff review, not proof of availability or a new model input.
What the model has not seen
A box-score model cannot identify an injury announced after its cutoff, a changed role, a redshirt decision or a late transfer unless a later source edition records it. The recruiting board preserves dated school statements and prior production with explicit coverage limits. It does not turn a signing into an eligibility ruling. Keep those facts beside the forecast and report uncertainty when you write the preview.
For the formulas, independent holdout and source receipts, read the basketball model notebook. For the player-level archive, use the scouting board to set your own production priorities and export the evidence behind a rank.
Data attribution: SportsDataverse, CC BY 4.0. See the source receipts and model notebook.