Margin Recency
Margin regression with exponential recency weights on newer games.
| Status | implemented | Short Name | MargR |
|---|---|---|---|
| Rating Style | margin | Where Shown | teams rankings, team pages, model performance |
Model Card
Same linear margin model as Margin, but trained with exponentially decayed sample weights so recent games matter more.
Validation on this page uses stored season-level full-fit rows plus strict weekly rolling holdouts. Rolling ranks are based on average weekly log loss in the selected league/season. Current scope: All Games. All scored games in the selected league and season. AP Poll is excluded here.
Selected League / Season
Use the selected scope to inspect this model against peers and across rolling weekly holdouts.
| Full Fit AUC | - · 0 games |
|---|---|
| 7-Day Holdout Log Loss | 0.643 (#12) · 14 games |
| Rolling Avg Log Loss | 0.614 (#3) · 26 windows |
| Window Wins | 1 · 66.8% rolling accuracy |
Full-fit AUC is only available where the older stored-rating batch produced that model for the selected scope. Recency and ensemble families can still have valid 7-day and rolling holdout results even when the full-fit columns are blank.
Rolling Holdout Curves
Selected model plus comparison peers for NBA 2025-2026. The focus line is Margin Recency. Click legend items to hide/show peers.
Weekly strict holdout log loss. Lower is better. Showing 6 models across 26 windows.
Recent Windows
| Holdout | Log Loss | Rank | Winner |
|---|---|---|---|
| Apr 15 - Apr 19 | 0.660 | #13/16 | Adjusted Context Blend (0.566) |
| Apr 8 - Apr 14 | 0.544 | #7/16 | Elo (0.522) |
| Apr 1 - Apr 7 | 0.517 | #9/16 | Elo (0.479) |
| Mar 25 - Mar 31 | 0.535 | #9/16 | Dynamic Bradley-Terry (0.510) |
| Mar 18 - Mar 24 | 0.517 | #6/16 | Elo (0.475) |
| Mar 11 - Mar 17 | 0.569 | #7/16 | Adjusted Context Blend (0.560) |
| Mar 4 - Mar 10 | 0.607 | #7/16 | Adjusted Efficiency (0.590) |
| Feb 25 - Mar 3 | 0.515 | #7/16 | Log Adjusted (0.495) |
Selected Scope Peer Table
Comparison snapshot for NBA 2025-2026. Rows are ordered by rolling log-loss rank; full-fit AUC may be blank for models that do not have legacy full-season rows stored for this scope.
| Model | Full AUC | AUC Rank | 7d LogLoss | 7d Rank | Rolling LL | Rolling Rank | Windows |
|---|---|---|---|---|---|---|---|
| Recency Ensemble | - | - | 0.616 | #8 | 0.610 | #1 | 26 |
| Core Ensemble | - | - | 0.613 | #6 | 0.611 | #2 | 26 |
| Margin Recency | - | - | 0.643 | #12 | 0.614 | #3 | 26 |
| Points Off/Def Recency | - | - | 0.641 | #11 | 0.614 | #4 | 26 |
| Bradley-Terry Recency | - | - | 0.592 | #4 | 0.618 | #5 | 26 |
| Dynamic Bradley-Terry | - | - | 0.590 | #3 | 0.618 | #6 | 26 |
| Points Off/Def | 0.755 | #3 | 0.629 | #9 | 0.619 | #7 | 26 |
| Margin | 0.746 | #8 | 0.629 | #10 | 0.619 | #8 | 26 |
League × Season Summary
Cross-scope validation history for Margin Recency. Rolling ranks are by average weekly log loss.
| League | Season | Full AUC | AUC Rank | 7d LL | 7d Rank | Rolling LL | Rolling Rank | Windows | Wins |
|---|---|---|---|---|---|---|---|---|---|
| WNBA | 2025 | - | - | 0.609 | #8 | 0.636 | #9 | 21 | 1 |
| NCAAWD2 | 2025-2026 | - | - | 0.487 | #4 | 0.532 | #3 | 14 | 3 |
| NCAAW | 2025-2026 | - | - | 0.539 | #1 | 0.515 | #3 | 22 | 8 |
| NCAAMD2 | 2025-2026 | - | - | 0.536 | #5 | 0.543 | #4 | 14 | 2 |
| NCAAM | 2025-2026 | - | - | 0.641 | #8 | 0.554 | #4 | 22 | 3 |
| NBA | 2025-2026 | - | - | 0.643 | #12 | 0.614 | #3 | 26 | 1 |
| GLEAGUE | 2025-2026 | - | - | 0.787 | #9 | 0.671 | #7 | 22 | 2 |
Output Coverage
Stored site-output rows for this model.
| League | Season | Rows |
|---|---|---|
| WNBA | 2025 | 17 |
| NCAAWD3 | 2025-2026 | 730 |
| NCAAWD2 | 2025-2026 | 697 |
| NCAAW | 2025-2026 | 592 |
| NCAAMD3 | 2025-2026 | 781 |
| NCAAMD2 | 2025-2026 | 603 |
| NCAAM | 2025-2026 | 704 |
| NBA | 2025-2026 | 37 |
| GLEAGUE | 2025-2026 | 31 |