NIL Model-Implied Valueβ
What this is and is not
This section estimates each player's retrospective contribution to team results from reconstructed on-court lineups. Separately, it publishes a speculative best guess of each player's part of the school's own revenue-sharing budget for men's basketball in 2025–26, with an 80% range (amendment A10). Neither is observed pay, a contract, an offer or a market price.
- No player-level pay data and no commercial NIL valuation is used or shown. Compensation figures are speculative, unvalidated and exclude third-party and collective deals, non-cash benefits and bonuses. Assumptions, anchors and checks
- There is no league-wide ranking of players by value or by dollars. Players are reached from team pages and rosters are ordered by possession share.
- Every point estimate carries a diagnostic resampling range and an evidence-quality label. Ranges are not calibrated intervals.
- The frozen program-level revenue allocation is still withheld unless the revenue gates pass and the wins-to-revenue elasticity prior agrees with the cross-section. Its published status and reason: suppressed: Elasticity cross-check failed: the wins-to-revenue elasticity prior and the cross-section implied elasticity differ by more than the declared factor, and no gate validates the prior. The speculative compensation estimate is a separate model that does not use it.
Method summary
Model nil-value-v3. Offense and defense effects are fit by ridge regression on accepted lineup stints. Players under 300 on-court possessions share a pooled term and receive no individual value. Ranges come from 200 whole-game resampling refits. Definitions, sources and exclusions are on Methodology.
Promotion gates
Thresholds were declared before any model output. A failed gate stays visible and does not remove the numbers shown on team and player pages; they remain experimental. The post-hoc disclosure is on Methodology.
| Gate | Requirement | Threshold | Observed | Scope | Result |
|---|---|---|---|---|---|
| L1 | accepted_game_share | >= 0.6 | 0.693 | 2023 | Pass |
| L1 | accepted_game_share | >= 0.6 | 0.721 | 2024 | Pass |
| L1 | accepted_game_share | >= 0.6 | 0.748 | 2025 | Pass |
| L3 | bound_possession_share | >= 0.85 | 1.000 | 2023 | Pass |
| L3 | bound_possession_share | >= 0.85 | 1.000 | 2024 | Pass |
| L3 | bound_possession_share | >= 0.85 | 1.000 | 2025 | Pass |
| L4 | minutes_within_2_share | >= 0.95 | 1.000 | 2023 | Pass |
| L4 | minutes_within_2_share | >= 0.95 | 1.000 | 2024 | Pass |
| L4 | minutes_within_2_share | >= 0.95 | 1.000 | 2025 | Pass |
| L2 | conference_accepted_share (minimum over conferences) | >= 0.4 | 0.385 | 2023 | Fail |
| L2 | conference_accepted_share (minimum over conferences) | >= 0.4 | 0.289 | 2024 | Fail |
| L2 | conference_accepted_share (minimum over conferences) | >= 0.4 | 0.360 | 2025 | Fail |
| P1 | rmse_gain_vs_box | >= 0.02 | 0.077 | 2023 | Pass |
| P1 | rmse_gain_vs_box | >= 0.02 | 0.066 | 2024 | Pass |
| P1 | rmse_gain_vs_box | >= 0.02 | 0.052 | 2025 | Pass |
| P1 | pooled school-cluster bootstrap lower bound of gain vs box | > 0 | 0.044 | pooled | Pass |
| P2 | seasons with incremental gain vs prior rating >= minimum (of 3) | >= 2 | 3.000 | pooled | Pass |
| P2 | pooled bootstrap lower bound of incremental gain | > 0 | 0.031 | pooled | Pass |
| P3 | interval_coverage_80 within [min, max] | within 0.7 | 0.785 | 2023 | Pass |
| P3 | interval_coverage_80 within [min, max] | within 0.7 | 0.778 | 2024 | Pass |
| P3 | interval_coverage_80 within [min, max] | within 0.7 | 0.810 | 2025 | Pass |
| P4 | worst slice RMSE of the model relative to box baseline | <= 0.05 | 0.053 | pooled | Fail |
| R1 | crosswalk coverage (min of school-data rows and registry teams) | >= 0.9 | 0.981 | primary | Pass |
| R2 | conference-blocked CV rmse gain; coefficient positive; interval lower bound above 0; sensitivity same sign | >= 0.02 | 0.257 | primary | Pass |
| R3 | implied per-win elasticity versus prior (report only) | report 0 | 0.095 | primary | report-only |
| R4 | temporal hold-out gain on multi-year EADA | >= 0.02 | — | not-run | Not run |
| S1 | rest effect points per rest-day unit (evaluation-pooled) | >= 0.3 | 0.857 | evaluation | Pass |
Held-out evaluation with known season rosters
RMSE of team-season net efficiency, lower is better. Lower and smaller is better; "gain" columns are known-roster comparisons, not forecasts made before the season.
| Season | Use | Team-seasons | Model | Box baseline | Prior rating | Raw prior | Constant | Model + prior | Gain vs box | Gain vs prior (known rosters) | 80% change coverage (survivors only) | Players |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2023 | evaluation | 337 | 7.184 | 7.786 | 6.759 | 6.942 | 8.800 | 6.539 | 0.077 | 0.032 | 0.785 | 1,833 |
| 2024 | evaluation | 333 | 7.086 | 7.589 | 7.150 | 7.245 | 9.204 | 6.828 | 0.066 | 0.045 | 0.778 | 1,818 |
| 2025 | evaluation | 343 | 6.996 | 7.376 | 6.947 | 7.224 | 9.075 | 6.491 | 0.052 | 0.066 | 0.810 | 1,805 |
| 2026 | descriptive | 364 | 7.613 | 8.099 | 7.441 | 7.770 | 9.739 | 7.030 | 0.060 | 0.055 | 0.764 | 1,808 |
Dated context figures
School-wide revenue-sharing figures follow. They are reported context only. They are not split by sport and not compared with any individual. The speculative compensation model uses the national total and the cap as budget anchors, with their sources, on the methodology page.
| Figure | Value | As of | Note | Source |
|---|---|---|---|---|
| House revenue-sharing cap, 2025-26 | 20,500,000 USD | 2025-26 | School-wide cap, no basketball split. Settlement agreement itself not retrieved. | Source ↗ |
| House revenue-sharing cap, 2026-27 (about) | 21,580,000 USD | 2026-27 | About; school-wide. 20.5M x 1.04 = 21.32M shows the step is not a flat 4 percent, so nothing is extrapolated. | Source ↗ |
| Cap share of revenue, wording conflict | CSC states 22 percent; NCAA president letter states 22.5 percent | 2025-06 | Unresolved conflict between two NCAA-side sources; both kept. | Source ↗ |
| NIL Go cleared deals, all sports | 46,478 deals worth 582,490,000 USD | 2026-08-31 | National, all sports, all Division I. Reported as $582.49M. Context only; never used to price or calibrate. | Source ↗ |
The cap share is reported as 22% by the College Sports Commission and 22.5% in an NCAA president letter. This conflict is unresolved.