Current stat: Signal 26,932 PLAYER-SEASONS · TUNE IT IN THE STAT LAB

Share link copied

The complete pro basketball record

Every era.
One signal.

Explore 80 seasons through one transparent rating system. Search any player, compare careers, build rosters, and pull apart offense, defense, and the details between them.

26,932player-seasons indexed

Rate

Type a name. Every column header sorts, and hovering any header or badge explains the stat. The amber column is your stat, live from the current Stat Lab weights.

Signal vs Signal

Two careers, one chart. Align by season for rivals, by career year for draft-class arguments, by age for trajectory arguments.

Align by From To Rate

Pick two players. The overlay chart, the metric face-off, and the aligned season table build themselves.

Roster Lab

Build a team out of any player-seasons in history, split 240 minutes a game, and the Lab grades it: the three scores, the style DNA, the closest real teams ever assembled, and a projected record from a model fit on every real roster since the merger.

Leaderboards

Season Stat Team League Qualified only Rookies only Age Class of Rows Order Rate

Playoff Signal

80 postseasons, graded only against that postseason. Playoff Signal uses the available Balanced Signal weights from playoff game logs. Playoff Green grades scoring, efficiency, creation, and ball security; Red grades steals, blocks, defensive rating, and rebounding share; Amber grades impact, boards, workload, and PIE. Missing historical stats are omitted and the remaining weights are renormalized—never treated as zero. Badges are scarce by design: top 5% is Elite, top 1% is Generational. The fair minutes floor scales with the size of each postseason (50 to 100 minutes).

Season Rank by Qualified only Rows

Play-Type Boards

Synergy tracking, 2016 to now. Rank the league at any action, both sides of the ball. PPP is points per possession; percentile is against the whole league at that action.

Side Play type Season Min poss/game Rows

Career Signal rankings

Career score is the minutes-weighted average of every season, so hanging on too long costs you. Value is total career production above league average: the longevity number. Peak is the single best year. Filters with a team selected rank only the seasons played for that franchise; the era filter counts only seasons inside the chosen era, so a player who straddled two eras is judged separately in each.

Team From To League Era Min career minutes Rank by Rows

Team Season Rate

Overall

Green Light · offense

Red Light · defense

Amber Light · glue

Hidden Gems: low-minute killers

Players under the qualification floor, scored with Signal's Durability weight forced to zero. Pure rate-stat quality: who produces when they actually get on the floor. Small minutes means small samples, so treat it like a scouting list, not gospel.

Season Team Min minutes League Rows Rate

Skills Board

Scouting tags. Each skill scores volume plus efficiency against qualified players in the same season and league, 50 is average, 10 is one standard deviation. This layer is separate from Signal and the Lights, whose formulas are untouched.

Season League Show

Defense

Tracking-era defense, 2014 to now. The Red Light Score blends steals, blocks, DBPM, rim protection, matchup suppression, assignment burden, and defensive hustle (deflections, contests, charges) into one era-adjusted score. From 2018, weights slide by position; matchup suppression grades the result, while Assignment Burden credits the difficulty and volume of the job. An absolute Red Light Score measures defensive value, which lives at the rim, so bigs lead it honestly. Switch Compare to Vs position to grade everyone against their own job. Hover any header for definitions.

View Compare Season Team Qualified only Rows

Summer League Signal — 2026

Box-score projection, not a full Signal. Updated through July 17 across the California Classic, Salt Lake City, and Las Vegas. The leaderboard requires 2+ games with no minutes floor; every player, including one-game appearances, still has a searchable card in All Player Cards. The All-Summer teams are positionless and require at least 85 total minutes. Scores are rebuilt from cumulative totals against the full summer pool after every slate.

All-Summer Signal Teams

Positionless · minimum 85 minutes · through July 17

View Event Team Stats Rank by Rows Search

LIGHTS: SCORING · CREATION · REBOUNDING · DISRUPTION | each 0–100, 50 = summer average. Tap a player for their card. SOURCES: NBA and ESPN box scores.

This is your own stat, separate from the canonical Signal. SIG is locked to the Balanced weights and never moves; these dials shape a score that belongs to you. Rename it, it saves to this browser, and it shows up on the Leaderboards under its own name.

Each dial weighs one era-adjusted component. Every component is measured against that season's league mean and spread (a z-score), which is also what neutralizes pace: 1962 and 2026 are each judged on their own curve.

Presets:

Team From To League Rows Order

Signal all-time top seasons

The Book · How The Ledger grades basketball

One score, receipts included

Every rating here starts the same way: take a thing a player does, measure it against everyone else who played that same season, and express it in standard deviations from the league mean. That single move, the z-score, is what lets 1962 and 2026 share a leaderboard. Wilt is graded against Wilt's league, Jokić against Jokić's. Pace, rule changes, and era inflation cancel out because everyone in a season faced the same conditions.

Scores read like this: 50 is a league-average player. Every 10 points is one standard deviation. The tiers below are percentile-based, so they always describe where a season or career actually ranks, never a fixed number that drifts out of date.

The tiers

Nine rungs, picked by the math

Each rung shows a real 2026 season that lands in that tier under the locked Signal weights. These examples are computed live from the data, not chosen by hand.

The overall score

Ten dials, one number

The overall rating blends ten era-adjusted components. Each card explains what feeds the dial and why it earns a place in the formula. The face on each card is the current league leader in that component, computed from this season's data. Weights shown are the locked Signal formula, the Balanced blend; the Stat Lab lets you build your own separate stat.

The traffic lights

Green Light and Red Light

The Green Light Score is offense: scoring volume (30), usage-scaled efficiency (30), creation (25), and ball security (15). Fixed weights, so offensive ranks never shift under you. Green means go, and this measures who should be going. As of v2.2 efficiency is scaled by usage, so carrying real load at good efficiency outranks finishing lobs at great efficiency; creation is points generated by assists, the passing that actually produces points, not raw assist counts; and turnovers are priced per creation possession so engines are not punished for the cost of running an offense.

The Red Light Score is defense, and from 2018 on it grades every position on the defense that position is asked to play. The base components are rim deterrence, blocks, matchup suppression, defensive hustle, steals, and defensive BPM. The weights slide by position, and position means the minutes a player actually logged at each spot from play-by-play lineups, not a roster label: a 55 percent small forward is graded 55 percent as one. A center carries rim deterrence at 27 and matchup at 16, while a point guard carries matchup suppression at 40 and rim at just 10, because a guard is never assigned the rim and a center is never asked to chase the ball. Matchup suppression grades the result against expected scoring. Assignment Burden separately grades the job: 90 percent opponent quality and 10 percent primary-matchup workload per 36, standardized within the same season, league, and position group. Its sample-shrunk, capped adjustment prevents a tiny or unusually hard assignment sample from hijacking the score. As of v2.2, defensive rebounding is removed from the score entirely and lives in Amber where glue belongs; box outs, the labor version of rebounding, are folded into the hustle blend instead. Positionally skewed components (rim, matchup, blocks, hustle) are z-scored inside the position; steals and defensive BPM stay league-wide so absolute rim value keeps its honest premium. Seasons without assignment tracking are unchanged. Before 2016 the score is a box-score blend, and pre-1974 defense is estimated from defensive win shares per minute, so Bill Russell is judged on the evidence his era recorded rather than scoring blank.

The badges

Earned, never given

Positive skill badges are reserved for the top 5 percent inside each season: Elite or Generational only. A 1998 badge and a 2026 badge were earned against the wearer's own qualified league pool, and they never move when you change the Stat Lab dials. Counts below are unique qualified players, not rows or appearances. Red flags use their own scarce bottom-tier tests.

Evidence over time

What the record keeps, and since when

Older seasons are not penalized for stats that were not yet tracked. When a component has no data for a season, it drops out and the remaining weights renormalize.

1947

Points, games, the box score begins

1952

Minutes tracked, rate stats possible

1974

Steals, blocks, BPM era opens

1978

Turnovers counted

1997

Play-by-play: lineup plus-minus, shot zones

2014

Cameras: rim protection, defended shots

2016

Hustle: deflections, screens, charges

2018

Matchup data: points allowed as primary defender

Fine print that matters

Qualification and honesty rules

Season stats qualify at 500 minutes, prorated for short seasons. Career badges and all-time ranks require 10,000 career minutes, roughly four real seasons, because a career average from 900 minutes is a guess wearing a suit. Skill badges are percentile-tested within each season and never move when you change the dials. Sources: Basketball Reference season tables, raw game logs back to 1946, FiveThirtyEight RAPTOR, and stats.nba.com tracking, hustle, and matchup data.

The formulas, in full

Every score, written out

Nothing here is hidden. Each score is 50 plus ten times the weighted average of its component z-scores. A component that has no data for a season drops out and the remaining weights renormalize, which is how a 1962 season and a 2026 season share one ledger.

Signal · the overall score

50 + 10 × (weighted mean of ten z-components)

Scoring 17 · Efficiency 12 · Playmaking 10 · Rebounding 8 · Defense 17 · Impact 16 · Ball Security 5 · Durability 4 · Lineup Impact 5 · Hustle 6 (Balanced blend, the locked default). In tracking seasons, the defensive input includes the same Assignment Burden adjustment used by Red Light; it is not counted twice. Lineup Impact is reliability-shrunk: multiplied by min(1, minutes / 1750), because small-sample on/off is noise wearing a stat's clothes. And the whole score is reliability-shrunk below 500 minutes: pulled toward 50 in proportion to how little a player was on the floor, so a three-minute cameo cannot post a fake all-time season on a lucky shooting night.

Playoff Signal · the postseason score

50 + 10 × (weighted mean of available playoff z-components)

Scoring 17 · True shooting 12 · Playmaking 10 · Rebounding 8 · Steals + blocks 17 · Plus-minus 16 · Ball security 5 · Total playoff minutes 4. Each player is graded only against that postseason. Untracked historical categories drop out and the remaining weights renormalize. The qualification floor scales from 50 to 100 minutes with postseason size; below-floor scores shrink toward 50, while every playoff appearance remains visible on the player card.

Green Light · offense

50 + 10 × (Scoring 30 + Efficiency 30 + Creation 25 + Ball Security 15)

Efficiency = (true shooting − league TS) × clamp(usage / league usage, 0.5, 1.6): value is volume times efficiency, not volume plus efficiency. Creation = points generated by assists (1997+), the passing that produces points, else assist rate. Ball Security = −turnovers / (FGA + 0.44×FTA + AST + TOV): turnovers priced per creation possession, inverted.

Red Light · defense (v2.3)

50 + 10 × (position-blended weighted mean of six z-components)

Guard weights: Matchup 40 · Hustle 18 · Steals 14 · DBPM 12 · Rim 10 · Blocks 6. Center weights: Rim 27 · Hustle 19 · DBPM 14 · Blocks 17 · Matchup 16 · Steals 7. Wings and forwards interpolate. Each player's table is blended by the minute share he logged at each position. Hustle = deflections + charges + contests + box outs. Matchup = suppression versus expected, inverted so stops score high. Assignment Burden then applies a position-relative 20-weight adjustment: 90 percent opponent quality and 10 percent primary-matchup possessions per 36, capped at ±5 Red Light points and shrunk until 1,500 tracked possessions. Defensive rebounding is NOT here; it lives in Amber. Pre-tracking seasons are unchanged.

Amber Light · glue

50 + 10 × (Impact 16 + Rebounding 8 + Hustle 6 + Lineup Impact 5 + Durability 4)

A standalone view of Signal's connective-tissue components, not added on top of Signal. This is where defensive rebounding value is credited, because ending possessions on the glass is glue, not shot suppression.

The trust gates · badges, not scores

A gate never moves the number; it caps the credibility chip when the direct evidence contradicts the accumulated score. Red Light: a Legit-or-better chip requires matchup suppression not be bottom-tier for the position (z above roughly −0.75), unless the player anchors the rim (2.5+ challenged rim shots a game with negative deterrence), in which case the rim vouches for him. Green Light: a Legit-or-better chip requires self-created efficiency (isolation and pick-and-roll ball-handler percentile) not be bottom-quartile, so a catch-and-shoot feaster grades as a scorer but not an engine. Capped chips read Mid* with the reason on hover.

Rank squares · the card corner

Each scope shows Signal, Green Light offense, Red Light defense, and Amber Light glue. All-Time uses a 10,000-minute career floor and displays the top 100. Era ranks use a 6,000-minute floor inside that era and display the top 50; a player who straddles a fault line can rank in both eras. Current-year ranks display the top 50 qualified players. Eras: Pioneer 1947-54, Classic 1955-79, Golden 1980-2004, Modern 2005-15, Pace & Space 2016-present, split at the shot clock, the three-point line, the hand-check ban, and the pace revolution.

All-Signal honors · positionless by design

Beginning in 1952, the first season with player-minute data, every NBA season uses the same qualified-player gate as the yearly boards: 500 minutes, prorated for shortened seasons. The locked canonical Signal selects five players each for First, Second, and Third Team, regardless of position. The top ten qualified rookies make All-Rookie First and Second Team. Red Light selects two five-player All-Defense teams. The five lowest qualified Signal seasons form the Bad Signal Team. Ties break by minutes, then player name. A historical rookie slot stays unfilled if the season does not contain enough qualified rookies; the minutes rule never bends just to fill a badge.

Glossary

The terms, ours and the game's

Court Signal terms

Signal: The overall rating. 50 is league average; every 10 points is one standard deviation above or below.
Playoff Signal: A separate postseason-only rating, graded against that playoff field from game-log evidence. It never mixes regular-season production into a playoff score.
Green Light / Red Light / Amber Light: The offense, defense, and glue sub-scores. Green means go (who should have the ball), Red is stopping power, Amber is the connective value that belongs to neither end.
Trust gate / Mid* chip: A credibility check that caps a badge, never the score, when the direct film contradicts the accumulated number. The asterisk means "the score is earned, but the claim is unproven."
Matchup suppression: Points a defender allows versus what the specific men he guarded normally score. The one stat that measures "does his man score on him."
Rim deterrence: How much a defender lowers opponents' shooting at the rim versus those shooters' normal rate. Negative is good: the shot got harder.
Creation: Points generated by a player's assists, not raw assist counts, so setting up a three or a foul shot is credited fully and a screen is not double-paid.
Usage-scaled efficiency: True shooting judged against the load carried, so a high-volume engine at good efficiency outranks a low-usage finisher at great efficiency.
Reliability shrinkage: Damping a noisy stat (on/off plus-minus) toward zero until the sample is large enough to trust, around 1,750 minutes.
Skill badge: A top-5-percent within-season test for a specific skill (Floor Spacer, Table Setter, Gravity, Lockdown). Positive badges are Elite or Generational only, earned against the wearer's own qualified league pool and immovable when you change the dials.
Manley EFF: Martin Manley's Basketball Efficiency, the per-game sum of the good minus the bad, credited by name throughout the app.

General basketball terms

z-score: How many standard deviations a value sits from its season's mean. The device that makes eras comparable.
True Shooting % (TS%): Scoring efficiency that counts twos, threes, and free throws in one number: points / (2 × (FGA + 0.44 × FTA)).
Usage rate (USG%): The share of a team's possessions a player finishes with a shot, foul drawn, or turnover while on the floor.
BPM / DBPM: Box Plus/Minus and its defensive half: box-score estimates of a player's points added per 100 possessions versus league average.
Win Shares / WS48: An estimate of wins a player produced; WS48 puts it on a per-48-minute basis.
Defensive Win Shares (DWS): The defensive portion of Win Shares, the best defensive estimate available before steals and blocks were tracked in 1974.
Deflection / Charge drawn / Contest: Hustle events: knocking a ball loose, taking an offensive foul, and challenging a shot. The labor of defense.
Box out: Sealing an opponent off the glass so a teammate can rebound. The effort behind a rebound, tracked separately from the rebound itself.
Screen assist: A screen that directly frees the teammate who then scores. The big man's invisible playmaking.
Play type / PPP: The action a possession runs (isolation, pick-and-roll, spot-up) and its Points Per Possession, the efficiency of that action.
On/off (plus-minus): How a team's margin changes with a player on versus off the floor. Powerful but noisy, heavily context-dependent.
Qualified season: Here, 500+ minutes (prorated for short seasons); the floor for appearing in season leaderboards and badges.

Inflating 80 seasons of basketball...