Read the game log before you rank the player.
A season average is a summary. A game log shows the workload, opponent context and missingness that make the summary worth trusting.
The NCAA player box archive keeps game-level releases from 2010–11 through 2025–26, with source player, team and contest IDs beside every usable row. The 2010–11 edition is a sparse historical source release and is labeled as such in the archive. The archive also retains older player-season summaries. It is a source record, not an identity bridge to ESPN, and a row does not establish a current roster spot or eligibility.
Start with minutes and games
Points per game can rise because a player received more possessions, played longer rotations or faced a particular schedule. Read minutes, games and starts when the source reports them, then open the individual dates. A strong rate across four appearances is a different scouting lead from the same rate across thirty games. The rankings board lets you set minimum games, minutes and denominator-specific volume before sorting.
Use the denominator that created the rate
True shooting uses field-goal and free-throw attempts. Effective field-goal percentage gives a made three its extra value. Turnover rate uses recorded offensive possessions, while three-point and free-throw accuracy use their own attempt totals. A missing denominator remains unavailable; it is never silently converted to zero. Open the retained source fields on a row when a rate needs explanation.
The source also records possession, transition and assisted/unassisted splits. Those fields describe what the publisher captured for that player-game row. They are useful context for workload and role, but they do not isolate scheme, matchup quality or player credit without lineup and film evidence.
Check the archive audit
Each game-level season reports missing IDs or matchup labels, date problems, impossible made/attempt totals, negative possessions, out-of-range minutes and zero-minute rows that carry production. The NCAA release does not include venue or home/away fields, so location questions belong to the schedule and team-box archives. This separation keeps a player row from pretending to contain context it never carried.
Connect production to recruiting carefully
A game log can verify prior workload for a source ID. It cannot prove that an announcement is a transfer, that a roster listing means availability or that the same player will receive the same role. Pair the log with the dated recruiting evidence board, the NCAA roster archiveand the program's film questions. If those layers disagree, keep the disagreement visible for staff review.
The current Silvermine model forecasts 1,579 2026–27 games from team efficiency and pace. Player logs and recruiting evidence help a coach decide what to investigate; they are not silently inserted into that primary forecast.
Data attribution: SportsDataverse, CC BY 4.0. See the source receipts and model notebook.