A capped negative lineup-context correction addresses the clearest box-score contradiction.
Adopted Jul 21, 2026
Live model · 2025–26
Live standardHow lineup context changes the live ranking.
The live hussEyquation blends the box-score estimate with multi-season lineup evidence, then limits any adjustment to ±0.75 impact points. The full calculation remains visible here for inspection.
Inspect all 395 players60% box-score prior40% multi-season RAPM±0.75 maximum correction395 qualified players
This bounded model now determines the public 2025–26 Impact Rate. The frozen 2026–27 prospective test remains active as an independent validation check.
Return to official rankings →The formula will remain unchanged throughout the 2026–27 prospective test. Weekly snapshots are append-only; player disagreements can trigger an audit, but not an in-season formula adjustment.
Before each update, the release system preserves its source inputs and checks team coverage, player identities, minutes, model version, and unusual rank movement. Weekly movement will appear only after two real snapshots exist.
Prospective test registeredThe 76-season audit tests historical defense, alternative scoring rules, era balance, and simulation-based rank ranges. The live historical pages now show a small roster-event defense adjustment experimentally, with its failed gates disclosed.
Read the Career Ranking Audit →The live standard gives full credit at 30+ MPG and gradually reduces the above-replacement rate for shorter nightly roles. The unadjusted impact view remains available.
Inspect the adjustment →The live model preserves a recognizable top tier while changing disputed cases selectively.
The lineup signal can refine the box estimate, but it cannot overwhelm it.
Official bounded-context ranking
Full 2025–26 qualified cohort · live Impact Rate.
| Movement | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| 1 | 100.0 | Nikola Jokić | DEN | 65 | 2,265 | #1 | — | +0.30 | +6.70 |
| 2 | 99.6 | Shai Gilgeous-Alexander | OKC | 68 | 2,259 | #3 | ↑1 | +0.75 | +7.46 |
| 3 | 98.6 | Kawhi Leonard | LAC | 65 | 2,085 | #6 | ↑3 | +0.58 | +5.63 |
| 4 | 98.3 | Giannis Antetokounmpo | MIL | 36 | 1,039 | #2 | ↓2 | +0.17 | +5.18 |
| 5 | 98.2 | Victor Wembanyama | SAS | 64 | 1,866 | #4 | ↓1 | +0.42 | +5.45 |
| 6 | 98.2 | Luka Dončić | LAL | 64 | 2,289 | #5 | ↓1 | +0.02 | +3.71 |
| 7 | 97.8 | Stephen Curry | GSW | 43 | 1,329 | #7 | — | +0.09 | +3.37 |
| 8 | 97.5 | Jimmy Butler | GSW | 38 | 1,182 | #8 | — | +0.20 | +3.49 |
| 9 | 97.1 | Donovan Mitchell | CLE | 70 | 2,342 | #11 | ↑2 | +0.50 | +4.18 |
| 10 | 96.8 | Tyrese Maxey | PHI | 70 | 2,661 | #10 | — | +0.02 | +2.60 |
| 11 | 96.4 | Cade Cunningham | DET | 64 | 2,172 | #13 | ↑2 | +0.26 | +3.26 |
| 12 | 96.0 | Joel Embiid | PHI | 38 | 1,201 | #9 | ↓3 | -0.08 | +2.32 |
| 13 | 95.1 | Jayson Tatum | BOS | 16 | 522 | #26 | ↑13 | +0.24 | +2.84 |
| 14 | 95.0 | Jamal Murray | DEN | 75 | 2,652 | #24 | ↑10 | +0.23 | +2.80 |
| 15 | 94.6 | Lauri Markkanen | UTA | 42 | 1,443 | #30 | ↑15 | +0.38 | +3.21 |
| 16 | 94.5 | Chet Holmgren | OKC | 69 | 1,997 | #49 | ↑33 | +0.75 | +4.95 |
| 17 | 94.2 | Kevin Durant | HOU | 78 | 2,840 | #17 | — | +0.10 | +2.46 |
| 18 | 94.1 | LaMelo Ball | CHO | 72 | 2,017 | #27 | ↑9 | +0.30 | +3.06 |
| 19 | 93.9 | Derrick White | BOS | 77 | 2,625 | #65 | ↑46 | +0.75 | +4.69 |
| 20 | 93.2 | Alperen Şengün | HOU | 72 | 2,398 | #25 | ↑5 | +0.10 | +2.32 |
| 21 | 92.8 | Michael Porter Jr. | BRK | 52 | 1,689 | #15 | ↓6 | -0.13 | +1.66 |
| 22 | 92.1 | Scottie Barnes | TOR | 80 | 2,681 | #19 | ↓3 | -0.05 | +1.85 |
| 23 | 91.7 | Karl-Anthony Towns | NYK | 75 | 2,322 | #53 | ↑30 | +0.56 | +3.51 |
| 24 | 91.3 | Collin Gillespie | PHO | 80 | 2,282 | #51 | ↑27 | +0.56 | +3.52 |
| 25 | 91.0 | Anthony Edwards | MIN | 61 | 2,137 | #21 | ↓4 | -0.03 | +1.88 |
| 26 | 90.6 | Devin Booker | PHO | 64 | 2,146 | #41 | ↑15 | +0.42 | +3.09 |
| 27 | 89.9 | Jalen Johnson | ATL | 72 | 2,532 | #12 | ↓15 | -0.43 | +0.72 |
| 28 | 89.5 | Franz Wagner | ORL | 34 | 1,019 | #81 | ↑53 | +0.57 | +3.37 |
| 29 | 88.6 | Zion Williamson | NOP | 62 | 1,841 | #28 | ↓1 | -0.18 | +1.24 |
| 30 | 88.1 | Deni Avdija | POR | 66 | 2,199 | #20 | ↓10 | -0.27 | +0.99 |
| 31 | 87.8 | Kon Knueppel | CHO | 81 | 2,551 | #76 | ↑45 | +0.45 | +2.94 |
| 32 | 87.4 | Bam Adebayo | MIA | 73 | 2,365 | #85 | ↑53 | +0.52 | +3.14 |
| 33 | 87.2 | Jalen Duren | DET | 70 | 1,976 | #22 | ↓11 | -0.24 | +1.06 |
| 34 | 87.2 | Jarrett Allen | CLE | 56 | 1,519 | #46 | ↑12 | +0.27 | +2.47 |
| 35 | 87.1 | Isaiah Hartenstein | OKC | 47 | 1,137 | #45 | ↑10 | +0.48 | +3.28 |
| 36 | 86.9 | Neemias Queta | BOS | 76 | 1,926 | #80 | ↑44 | +0.67 | +3.75 |
| 37 | 86.7 | Amen Thompson | HOU | 79 | 2,953 | #61 | ↑24 | +0.32 | +2.56 |
| 38 | 86.0 | OG Anunoby | NYK | 67 | 2,224 | #118 | ↑80 | +0.74 | +3.69 |
| 39 | 85.6 | Evan Mobley | CLE | 65 | 2,074 | #47 | ↑8 | +0.18 | +2.13 |
| 40 | 85.3 | Jalen Brunson | NYK | 74 | 2,590 | #74 | ↑34 | +0.38 | +2.68 |
| 41 | 84.9 | Paul George | PHI | 37 | 1,135 | #57 | ↑16 | +0.21 | +2.15 |
| 42 | 84.5 | Josh Giddey | CHI | 54 | 1,731 | #29 | ↓13 | -0.29 | +0.79 |
| 43 | 83.7 | Robert Williams | POR | 59 | 1,008 | #16 | ↓27 | -0.04 | +1.95 |
| 44 | 83.2 | Al Horford | GSW | 45 | 969 | #38 | ↓6 | +0.15 | +2.16 |
| 45 | 83.1 | James Harden | LAC / CLE | 70 | 2,438 | #32 | ↓13 | -0.22 | +0.90 |
| 46 | 83.1 | Jrue Holiday | POR | 53 | 1,560 | #88 | ↑42 | +0.34 | +2.44 |
| 47 | 82.7 | Dyson Daniels | ATL | 76 | 2,520 | #87 | ↑40 | +0.33 | +2.40 |
| 48 | 82.4 | Pascal Siakam | IND | 62 | 2,057 | #84 | ↑36 | +0.29 | +2.30 |
| 49 | 81.7 | Austin Reaves | LAL | 51 | 1,762 | #67 | ↑18 | +0.15 | +1.90 |
| 50 | 81.5 | Sandro Mamukelashvili | TOR | 80 | 1,751 | #34 | ↓16 | -0.01 | +1.63 |
Showing 1–50 of 395
Page 1 / 8
Methodology
A correction, not a replacement.
The live model uses lineup evidence to correct important blind spots in Raw BPM without letting noisy single-season RAPM take over.
Start with the box prior
The stabilized Raw BPM estimate remains the majority signal and preserves the model's transparent box-score foundation.
Add lineup context
Possession-weighted RAPM from three seasons supplies evidence that controls for the other nine players on the court.
Cap the correction
The lineup adjustment cannot exceed ±0.75 internal impact points per 100 possessions, limiting collinearity and single-season noise.
Keep testing it
The frozen 2025–26 inputs will still be judged against completed 2026–27 RAPM. The predeclared gates remain visible as an independent audit of the live decision.
Model boundary: RAPM is an observational estimate with meaningful uncertainty, not ground truth. The formula was selected on population-level validation; player names were interpretation checks, not tuning targets. Prospective evaluation continues after adoption.