1. Offensive blend
Standardized box-score offense + 0.6 × standardized play-type value + 0.5 × standardized playmaking creation.
NBAI / ORIGINAL RESEARCH
Beyond the box score.
There’s more to a player than a scoring average.
Compare the ways players contribute, then take a look at how we measure them.
Loading historical research…
THE PLAYER VALUE EXPLORER
Find the contribution.
| Rank | Player | Team | WAR v4 | Off / 100 | Def / 100 | Minutes / game |
|---|
SIDE BY SIDE
Different routes to value.
Total value depends on playing time; contribution rates describe the model's offense and defense components. Differences are descriptive, not causal estimates or future win probabilities.
OPEN THE FORMULA
What goes into WAR v4?
2. Defensive blend
Standardized box-score defense + 0.6 × rim protection + 0.6 × deflections + 0.25 × charges + 0.25 × defensive box-outs. Each added term is standardized within season.
3. Season value
The blends are rescaled, added and centered against the season's minutes-weighted average, then adjusted for replacement level and playing time.
Exact conversion and limitations
Raw WAR = (BPM4 + 2) × [minutes / (5 × 48 × games played)] × (games played / 82) × 2.7. A shared scale factor anchors the mean season sum of raw WAR to 490. The displayed value is rounded to one decimal.
The offensive and defensive blends are mapped back to the box components' mean and standard deviation across the implementation's full input population. Missing auxiliary components are filled with zero before standardization. These choices affect interpretation: the result is a descriptive metric, not a causal impact estimate or an out-of-sample probability model.
Only the exported top 60 players with 500+ minutes appear here. Leaders and comparisons refer to this displayed population. Offense and defense are component rates; WAR also includes minutes and the replacement baseline. Small differences may disappear when rounded.
Historical team labels describe the source season and do not confirm current rosters. This page is not evidence that a betting strategy is profitable.
Source snapshot and provenance
Formula: scripts/129_war_v4.py. Export: scripts/export_web.py. This page reuses the saved website metrics; it does not rerun a model.
Download metrics and source hashes · Validation and historical evidence