GRBNU
A cross-league projection of what a player adds per 100 possessions. NBA, G League, EuroLeague and EuroCup players are rated on one NBA scale, so a two-way guard in Austin and a starter in Madrid can be read against an NBA rotation.
One scale for four leagues
0 is an average NBA player. A +3 player adds three points per 100 possessions over that average; a −3 player gives three back. Offense and defense are rated separately and add up to the total, and on both sides positive helps his team.
A G League or EuroLeague rating means the same thing as an NBA one: what we expect him to add in NBA minutes. It does not say whether he will get those minutes. Playing time and roster spots are separate questions.
Nikola JokicBest 2026–27 projection in the NBA, Denver Nuggets
Jonas ValanciunasBest 2026–27 projection in the EuroLeague, Zalgiris Kaunas
Average NBA playerWeighted by minutes
Aaron HolidayBest 2026–27 projection in the EuroCup, Maxima Roma
Orlando RobinsonBest 2026–27 projection among G League players
The current ratings
Updated November 12, 2026. Every season from 2018–19 through the 2026–27 projection was rated again.
-
Four more leagues as evidence
Spain’s Liga ACB, Italy’s Lega Basket, Turkey’s Super League and the Basketball Champions League now count as evidence for players who reach the EuroLeague, EuroCup or NBA. EuroLeague and EuroCup values come from one model fit across all six European leagues.
-
League strength measured each season
Each league’s level is set from measured results across 27 competitions in the season before, instead of fixed anchors, and 2017–18 now counts as a full prior season. In backtests, league-level error fell by about a third and NBA accuracy was unchanged.
-
Fewer outliers, no stale projections
Ratings above +1 for players outside the NBA that season fell from 10 to 7 since the last update, and none are above +2. Players who have not played in a covered league for years no longer get a 2026–27 projection.
-
Schedule for all eight leagues
Every 2026–27 game in the NBA, G League, EuroLeague, EuroCup, ACB, LBA, BSL and Champions League, with a projection from season-start ratings, an updated projection from game-by-game ratings, and the result once played. The home map now shows the ACB, LBA, BSL and Champions League clubs too.
How a rating is made
Each season is projected before it starts, from what was known at the time.
Box-score prior
A statistical model reads the player’s stats and blends guard, wing and big versions of itself, for offense and for defense, by how much he plays each role.
Other leagues
G League, EuroLeague and EuroCup seasons are translated to the NBA scale and added as evidence.
NBA plus-minus
Earlier NBA lineup results (RAPM) are carried forward and fade with time. Garbage time is dropped and shooting luck is partly removed.
Short histories
Rookies and low-minute players lean on players of similar age, role and draft slot until their own minutes take over.
Tested forward
Every past season was projected with only earlier data, then checked against team results and against players who changed teams.
How well it predicts
Every result here comes from ratings made before each season, using only what was known at the time.
My view, from my research: GRBNU is the only metric built to scale with relative success across leagues. A rating means the same thing whether a player is in Denver, Austin or Madrid.
Outside the NBA, it is the most accurate predictive metric I have found in every league. In the NBA it sits with the best all-in-one metrics, and the gaps between the top few are within statistical noise. I am open to challengers: if another metric does better on the same kind of test, I want to see it.
NBA: an independent test
An independent comparison by @Starro____ scored about twenty public metrics the same way, walk-forward from 2019–20: each preseason rating, weighted by the minutes players actually played, predicting how teams did.
| Metric | Team net error | Game R² |
|---|---|---|
| xPM (VORP.app) | 2.94 | .177 |
| GRBNU | 3.04 | .182 |
| EPM projection | 3.07 | .172 |
| LOKI | 3.08 | .170 |
| DARKO | 3.36 | – |
| LEBRON | 3.39 | – |
| xRAPM | 3.48 | – |
| BPM | 3.50 | – |
| Last season’s RAPM | 3.53 | – |
| Every player set to 0 | 4.65 | – |
Lower team error is better; higher game R² is better. GRBNU had the best game R² of every metric tested.
Outside the NBA
No other public all-in-one metric has been tested this way outside the NBA, so the bar is the obvious alternative: last season’s team results. Rating every player before the season and adding up each roster predicted club margins with less error than last season’s margins did.
| League | Player ratings | Last season’s margin | Less error |
|---|---|---|---|
| ACB (Spain) | 4.59 | 5.43 | 15% |
| LBA (Italy) | 4.36 | 5.11 | 15% |
| BSL (Türkiye) | 5.98 | 7.29 | 18% |
| Basketball Champions League | 8.03 | 9.07 | 11% |
Error in team margin per 100 possessions, seasons 2019–20 to 2025–26, one setting for all four leagues. These leagues are research inputs entering the next build. EuroLeague, EuroCup and G League ratings are chosen and checked the same way: against club results in their own league and against how players do after they move to the NBA.
During the season
Player pages update GRBNU after every game from that game’s lineup results, starting from the preseason rating. In every season from 2018–19 to 2025–26, the updated ratings predicted the rest of the season’s game margins better than the preseason ratings did, by 1–5%.
True stats
What a player’s numbers say about the player, not just about the games he played.
- What they are
- Box scores describe what happened. True stats estimate the player underneath: the rate to expect from him if he plays, in every category, from shooting by zone and shot type to playmaking, turnovers, rebounding and defense. Counting stats are per 70 of his own possessions; percentages are estimated the same way.
- How they work
- Each stat starts from his own record, with recent seasons counting most. That record is blended with what similar players usually do: same role, size, age and league. The more evidence he has in a stat, the more his own numbers decide. Ten three-point attempts say little, three hundred say a lot. G League seasons are translated to the NBA scale and counted too.
- Why it matters
- A hot month, a small sample or a new role moves the box score much more than it moves the player’s ability. True stats separate the two, which is what makes them useful for projecting what comes next.
- Reading them
- Each season’s true stats are the estimate entering that season, made only with earlier data. Compare them across players and seasons, but do not add them up across seasons. Opponent-impact columns show the adjusted change in what opponents do per 70 of their possessions with him on the floor. True stats cover NBA players today.
Which leagues count
- Core4
- The NBA, G League, EuroLeague and EuroCup. Core4 seasons feed GRBNU, and Core4 players get full profiles.
- NC4
- Every other league. NC4 ratings are a separate research estimate for players before they reach Core4. After that, NC4 seasons stay in the career as history, not as a second rating.
Reading the site
- Player card
- The top of every profile, painted in his team’s colors: size, birth, draft, agent, latest contract and a scoreboard with his current GRBNU.
- Trajectory
- His whole career on one chart. Each colored band is a team stint, labeled with the club’s logo or abbreviation; dashed lines are projections, and thin lines show how his GRBNU moved after every game of a season. The lanes underneath mark moves T (trade, signing, waiver and so on), injuries and cap hit. Tap a season for details.
- Season table
- The same seasons as rows: team, league, GRBNU, rank, minutes, salary, coach and injuries.
- Injury figure
- Each documented injury is pinned where it happened, with surgeries and long absences in red. No pins means nothing was documented.
- Career map
- Birthplace, school and every club on one map. Play traces the route in order.
- Team page
- One roster table with every player stat, sortable by any column, and one ratings table covering every season of the club.
- Leaderboards
- Players by season, career peak or salary, for one league or any mix of leagues (tap a league, tap more to combine), with GRBNU against pay; NBA team projections since 2018–19 and how they did; cross-league club ratings; how the leagues compare over time; NC4 research. Every view exports to CSV, and the link keeps your settings.
Glossary
- GRBNU
- Projected points per 100 possessions above an average NBA player. Offense plus defense.
- Projection
- The rating entering a season. 2026–27 values are this season’s preseason projections.
- Player index
- A roster’s GRBNU averaged by minutes played, or by projected minutes for 2026–27.
- Cross-league net
- A club’s results in all its competitions, adjusted for league strength and put on the NBA scale.
- Cap %
- Cap hit divided by that season’s NBA salary cap.
- Archetype
- Offensive and defensive archetypes. They describe how a player plays and do not feed GRBNU.
- GRBNU vs. pay
- His GRBNU minus the usual GRBNU of NBA players paid the same share of the cap that season.