01Game planner
Read the next matchup
Move from the schedule to a practical game question, then test the four factors behind the forecast.
Basketball reading room / Field guideStart with the question
How to use the desk
Learn the game behind the numbers.
A scoreboard tells you what happened. This guide shows how to move from a number to a useful basketball question, then points you to the evidence behind the answer.
01 / Pick your route
Four ways to learn the desk.
02Player researcher
Build a player file
Start with a complete source row, qualify the workload, and open the exact player evidence behind a ranking.
03Recruiting analyst
Turn a roster row into a question
Separate what a school published from what the stats record, then make the next fit check explicit.
04Methods reader
Audit the claim
Understand how the estimate was tested, what it can see, and whether the underlying source clock is current.
Your study desk
Pick up where you left off.
Mark a lesson complete as you work through the four routes. The checklist stays on this device and only organizes your reading; it does not change a forecast, ranking or recruiting record.
0 of 8lessons checked
Desk directory
Open a specific room.
Research path
Find a player
Rates, workload, shooting and game evidence across the archive.
Open find a player →Research path
Compare programs
Turn ratings and four factors into a venue-aware matchup question.
Open compare programs →Research path
Read the next slate
See every published 2026–27 forecast, interval and model timestamp.
Open read the next slate →Research path
Open NCAA leaders
Browse source-native Division I, II and III national leaderboards.
Open open ncaa leaders →Research path
Inspect player impact
Review NCAA-derived RAPM with offensive and defensive samples.
Open inspect player impact →Research path
Read within-team RAPM
Compare source-published player impact relative to teammates across 17 seasons.
Open read within-team rapm →Research path
Study team box history
Compare NCAA-derived efficiency, tempo and Four Factor profiles across 17 seasons.
Open study team box history →Research path
Compare published models
Read attributed team ratings and Box Plus/Minus beside Silvermine's independent model.
Open compare published models →Research path
Study lineups
Use possession thresholds and lineup net ratings to ask which combinations actually worked.
Open study lineups →Research path
Rank player production
Apply game and minute thresholds before comparing scoring, playmaking or shooting efficiency.
Open rank player production →Research path
Compare historical seasons
Set a season window, keep source identities visible and apply workload thresholds before comparing player production.
Open compare historical seasons →Research path
Read shot profiles
Compare shot volume, zone conversion and recorded distance within the NCAA source identity namespace.
Open read shot profiles →Research path
Read possession style
Learn how team-season possession context is summarized without assigning individual credit.
Open read possession style →Research path
Compare market evidence
Inspect retained lines, capture clocks and the rules that keep unverified odds out of the forecast record.
Open compare market evidence →02 / Identity and availability
Recruiting starts with the source row.
Recruiting path
Study recruiting
Separate announced additions, roster observations and prior production.
Open study recruiting →Recruiting path
Build a role shortlist
Choose a program need and rank source-listed candidates by transparent prior-production percentiles.
Open build a role shortlist →Recruiting path
Rank roster workload
Sort source-listed 2026–27 players by prior workload, rates or publisher Box BPM, then open the evidence.
Open rank roster workload →Recruiting path
Browse player profiles
Start with source identity, position and roster context before reading production or recruiting evidence.
Open browse player profiles →Recruiting path
Search roster intel
Use class, position, size, hometown and high-school fields as recruiting context, then verify any transaction with a dated statement.
Open search roster intel →Recruiting path
Trace high-school pipelines
Aggregate source roster rows by high-school label, then open the underlying players before drawing a recruiting conclusion.
Open trace high-school pipelines →Recruiting path
Open a player card
Connect source-native production, shot profile, roster context, impact and recent game evidence.
Open open a player card →Recruiting path
Build a recruiting brief
Move from an exact source row to dated school evidence, prior production and a question for film.
Open build a recruiting brief →Recruiting path
Verify availability
Read injury, redshirt and tournament-availability signals with the right source, date and evidence boundary.
Open verify availability →03 / Metric dictionary
Translate the stat into a decision.
30 of 30 definitions · choose a row to focus the lesson.
Silvermine calculations are independent estimates. Publisher metrics retain their source labels and source identities. A missing value is evidence that the required field or qualifying sample was unavailable.
Try it / Four Factors
Turn a box score into a question.
Enter nonnegative team counts. The calculator stays in this browser; it does not change Silvermine ratings, forecasts or saved research.
Missing inputs keep the affected rate unavailable instead of treating a blank as zero. Lower turnover rate is better; higher eFG%, ORB% and FT rate can be useful only after checking volume and opponent context.
Try it / Model mechanics
Turn efficiency into a score.
This classroom worksheet averages each offense with the opponent's defense, applies an expected pace, and adds a home-court term. It mirrors the intuition behind an efficiency matchup while staying separate from Silvermine's fitted forecast.
Formula: home efficiency = (home Adj O + away Adj D) / 2 + home-court term; away efficiency = (away Adj O + home Adj D) / 2; score = pace × efficiency / 100. Lower Adj D is better. Missing inputs keep the whole exercise unavailable, and the result does not change ratings, forecasts or the ledger.
Interactive market lesson
Translate a price into a probability.
A model probability and a market price answer different questions. Use this worksheet to see the raw implied probability, the two-way no-vig estimate and the gap between them before comparing a forecast with a quote.
Margin interval exercise
See how uncertainty changes a spread probability.
This is arithmetic for learning and auditability. It assumes a two-way market, ignores pushes and market limits, and does not establish that a model is calibrated or that a quote is current. The cover exercise uses a normal approximation to the symmetric nominal 80% interval and should be read with that limitation. Silvermine keeps missing or unlicensed market observations unavailable.
Interactive recruiting lesson
Follow the minutes.
Enter the prior season's team minutes and the exact-ID minutes represented by the next roster snapshot. The worksheet shows why returning share, incoming workload and unrepresented minutes answer different questions.
The roster lab uses the same arithmetic on source-listed IDs. A represented minute is historical workload evidence; it does not establish a transfer, eligibility, availability or a future role. Incoming minutes can exceed the prior denominator in a hand-entered exercise, so review the source rows before drawing a continuity conclusion.
Coach's checkpoint
Can you read the evidence?
Five quick questions reinforce the habits that keep a scouting note useful: qualify the sample, read the uncertainty and preserve each source boundary.
Question 1 of 5
A team has a lower Adj D than its opponent. What does that tell you first?
04 / Read a forecast
Probability is a starting point.
- Check the timestamp. Confirm the model version, training cutoff and scheduled start before treating the estimate as a pregame view.
- Read the margin and interval together. A narrow projected gap can still carry a wide uncertainty band; the range describes model error in the held-out calibration sample.
- Ask what the model cannot see. The baseline uses historical efficiency and venue. Injuries, eligibility and lineup confirmation belong in the staff’s next evidence check.
05 / Read recruiting evidence
Keep announcements and production separate.
A school announcement is evidence that a program published an addition or availability update. A prior box-score line describes recorded production at a previous program. Neither one alone proves eligibility, a current roster spot or a future role.
The recruiting board keeps source links, publication dates and later availability statements beside the historical statistics. Start there, then check the player archive and program dossier.
Recruiting evidence ladder
Use each source for the question it can answer.
Good recruiting research is cumulative. Start with an exact identity, add dated reporting, measure recorded work, and leave eligibility and future role to the sources that can actually establish them.
| Evidence layer | It can establish | It cannot establish by itself |
|---|---|---|
| Exact source row | Which publisher ID, season and program label the archive recorded. | A unique cross-source person, a current roster spot or a transfer reason. |
| Dated school statement | What a program publicly announced and when it published the statement. | NCAA eligibility, playing-time guarantees or an unreported transaction. |
| Prior production | Recorded games, minutes, rates and source-native value at the prior program. | Future health, role, availability or production at a new program. |
| NCAA rule or availability source | The governing context or an official availability record when the source publishes one. | A conclusion inferred from a roster listing or a news headline alone. |
| Fit and film question | A transparent shortlist and the next basketball question to investigate. | A recruiting grade, a forecast input or proof that the role will work. |
Silvermine keeps these layers separate in the recruiting board and player cards. A missing layer stays unavailable; it is never replaced with a guess from a name match or a roster absence.
Recruiting vocabulary
Know what the roster signal says.
The board uses precise labels so a coach can move from an observation to the next source check without turning an absence or a name match into a transaction claim.
Source term
Roster observation
Meaning. A player/program row supplied by the source roster snapshot for a particular season.
Boundary. It is evidence that the source listed the row. It does not establish eligibility, enrollment, a scholarship or a guaranteed roster spot.
Source term
Same program
Meaning. The exact source player ID appears for the same program in the preceding source season.
Boundary. It describes observed continuity in the archive. It does not prove a return decision or future availability.
Source term
Different program
Meaning. The exact source player ID was recorded at another program in the preceding source season.
Boundary. It flags a research handoff. It does not establish a transfer date, reason, destination announcement or eligibility.
Source term
New to dataset
Meaning. No matching prior source player ID was found in the preceding archive edition.
Boundary. It does not mean freshman, newcomer, transfer or first-year college player; the source may simply be incomplete.
Source term
Prior workload
Meaning. Recorded games and minutes from a source player/team season, with rates shown only when denominators are present.
Boundary. It is context for role and film review. It is not a projection of future minutes or a promise of production.
Source term
Source identity
Meaning. The publisher’s stable player, team and season keys used to connect rows inside one source namespace.
Boundary. A key in one release is not automatically the same person as a key in another release; unverified name joins remain separate.
06 / Build a recruiting brief
Move from a name to a defensible question.
Step 01 / Find the source row
Start with the roster observation.
Search the source-listed 2026–27 roster and record the season, program ID and exact player label. A listed row is an observation, not an eligibility ruling.
Open the roster board →Step 02 / Read the announcement
Keep the publisher’s date attached.
Open the school statement and any later availability update. The timeline tells you what was reported and when; it does not fill in an absent transaction record.
Open dated evidence →Step 03 / Measure prior work
Use the complete college sample.
Check games, minutes, efficiency and the game log at the prior program. Keep prep and international additions separate when no college box-score profile exists.
Open the player archive →Step 04 / Hand off to the model
Ask what still needs confirmation.
Put the evidence beside the forecast and write the next check: availability, role, health, lineup fit or opponent matchup. The primary forecast stays reproducible until a reviewed model release changes it.
Read the model boundary →A missing roster observation is not evidence of departure, and prior minutes are not a promise of future minutes. Preserve both statements when you share a recruiting brief.