The eight steps — S / A / B / C / D / E / F / G — that every metric is graded on.
A grade comes from nothing but where a player ranks within the league that season. The top 2% get S, up to 8% A, up to 25% B, up to 50% C, up to 75% D, up to 92% E, up to 98% F, and the rest G.
What each metric measures is set out in that metric’s own explanation.
Related: Percentile / Qualified players / “—” (blank)
Where a player sits in the qualified pool counting from the top, on a 0–100 scale. The closer to 0, the higher.
It runs the opposite way to the percentile used in statistics. There the figure counts up from the bottom and higher is better; here it matches a standings table — how far down the list you are (the two add up to 100).
Even first place is not 0% — a player takes up their own share of the scale, so in a pool of 434 that is 1 ÷ 434, or 0.2%. Last place lands on exactly 100%.
Every grade on this site is assigned from this ranking.
Related: Grade / Qualified players
It is not below G. It means there were too few attempts to rank the player at all.
A metric measured as a rate swings in both directions on luck alone when attempts are few: a player who hit 3 of 3 would lead the league at 100%. To stop that, every metric sets a minimum number of attempts, and anyone short of it is left unranked rather than placed.
Which metric needs how many is listed on the How grades work page for each league. Metrics that measure volume itself — Scoring, Rebounding and the like — have no minimum.
This is why blanks are common early in a season. They fill in as attempts accumulate.
Related: Shrinkage / Qualified players / Grade
The age shown on player pages and in the lists is as of February 1 of that season — not opening night, and not the end of the season.
February 1 falls near the midpoint of a season, and it is the basis the other stats sites use. Age is a convention rather than something measured, so we matched the one that can be compared against elsewhere.
The value the league publishes is as of the end of the season, and printing that as-is would make anyone with a birthday in the second half look a year older. So it is recalculated here from the date of birth.
Related: Qualified players / Experience / Draft
The set of players being ranked against one another. It takes in everyone who reached a set amount of playing time that season. The exact minimum, and how many players clear it, are on the How grades work page for each league.
The minimum is scaled to the number of games in the season, so that a shortened season does not end up being the only one with a stricter bar. It scales against the games the player’s team played, not the player’s own games played — otherwise the bar would quietly drop for anyone who missed a lot of them.
There is no split by position. Where a player stands in the league as a whole is what comes out.
Related: Grade / Percentile / “—” (blank)
The number sitting on each axis of a rose chart you build in the Editor. It is how many attempts it takes to be ranked in that metric — shots taken, possessions used — and a player below it shows “—” on that axis.
Shooting percentages use the NBA’s official qualification as it stands (300 FG, 125 FT, 82 3P). Where there is no official qualification — by zone, by play type, on defense — the number is one this site set.
You can move it. Lower it and more players appear; raise it and fewer do — the count in the metric’s dropdown changes as you go, so you can read right there how many you added.
Take it low enough and a player with ten attempts tops the list. The official qualification exists precisely to prevent that.
Related: Qualified players / Shrinkage / “—” (blank)
A correction applied to every metric measured as a rate. The fewer the attempts, the harder the value is pulled toward the league average; the more there are, the closer it stays to what the player actually did.
A player who made 45 of 50 and one who made 450 of 500 both shot 90%, but how far the two figures can be trusted is not remotely the same. Treat the first like the second and the top of the list fills with the players who attempted least, not the best ones.
How hard the pull is differs by metric. How many attempts’ worth of league average gets mixed in is set for each metric from measured reliability (the values are on the How grades work page for each league).
Whatever that still cannot settle is left unranked (“—”).
Related: “—” (blank) / Percentile
A name for a tendency in how someone plays, such as “Slasher” or “Pull-Up Shooter”.
Like grades, every one of them comes from a formula — none are put on or taken off by hand.
Quality is the grades’ job; tags cover the way a player contributes. So a player with no tags at all is not a lesser player — no pattern happened to fit.
The order on a player page runs from the rarest tag first, not by rank. Nor does carrying more of them put a player above anyone else. How many players each tag covers is shown on that tag’s own page, along with the list for the season.
Related: Grade
Four of them: scoring mix, play-type mix, rebounding mix and defensive mix. Each splits where the points come from, and what the player is doing, into shares stacked to 100%.
They show proportions, not volume, so two identical bars can sit on very different contributions when the playing time differs. Volume is the grades’ job.
They describe a style. They do not rank one above another.
The circle under “Grades” on a player page. It shows every metric’s grade as the length of a petal — the longer, the higher.
A petal that runs straight through the center is a “—”, and it does not mean the same thing as G: there were too few attempts to rank it. Those are drawn with hatching.
Related: Grade / “—” (blank)
The career and single-season leaders, taken from every season on record (1946-47 onward).
Anything from before its record began is missing — steals and blocks begin in 1973-74, and three-pointers in 1979-80. Players who were already active before then have career totals that fall short by that much.
On the per-game side, seasons where a player did not reach 70% of that season’s schedule are left out (the same line as “Hide players with few minutes” in Players & Teams). The order is by value, and among equal values the order means nothing.
Related: Qualified players / Grade
The career leaders for every franchise, current and defunct, plus their best single seasons.
The most wins favors the 82-game era — shortened seasons (50 games in 1998-99, 66 in 2011-12, and 64 to 75 by team in 2019-20) cannot reach it. Win percentage does not depend on the number of games, so both are given side by side.
Related: Winning percentage / Player records (career and season)
Last season is where that player stood in the same metric a year ago.
“Rookie” means that year was their NBA debut. “—” means no rank was assigned in this metric last year — too few games, or below the minimum attempts. Neither means the player was poor.
The difference cannot be read as how much a player improved — the number of players ranked changes from season to season.
Related: Grade / Qualified players
That year’s picks, in order of round and selection.
The faded rows are players who never appeared in an NBA game (they have no player page). Career figures are the totals across every season on record.
Some picks show a round or a number of 0 — territorial picks, which had no round as a matter of the rules, and picks from the early seasons where no record survives.
Related: Player records (career and season)
One row per season, reading champions and the main awards down the page.
The MVP was created in 1955-56 and Defensive Player of the Year in 1982-83, so earlier seasons are blank.
Team abbreviations are the ones in use that season, and an abbreviation shown as an outline is a team name that no longer exists. An abbreviation with no color is a team that ended there.
Related: Season tables (standings and team stats) / MVP (Most Valuable Player)
Grades are assigned to players with a set amount of playing time that season. They exist only for the seasons that are graded, and the range is on the How grades work page for each league.
Per game and totals go back to 1946-47, and they are the NBA’s own record, unconnected to the grades.
Advanced starts in 1996-97 and does not exist for earlier seasons (the NBA does not publish the totals). There are no career figures for it either — rates cannot be added up across seasons.
2P is a derived column, FG minus 3P.
Related: Grade / Qualified players
The height listed on the roster. It arrives in feet and inches, and is converted to centimeters here.
It is not a fixed value. Listings used to be a mix of measurements taken in shoes and figures self-reported back in college. Before the 2019-20 season the NBA required every team to measure its players barefoot and submit the result, and the listings were updated across the board that year.
Counting it up: 243 of 400 players had their listing move between 2018-19 and 2019-20, shrinking by 2.4cm on average. Neighboring years move about a third as much (110 of 433 from 2021-22 to 2022-23). It is the kind of shift a single season will not show you, so comparing heights across years takes care.
Several tags — Inside Defender, Post Player and others — use height as a condition.
Related: Weight / Listed position / Tag
The weight listed on the roster. It arrives in pounds, and is converted to kilograms here.
It moves more than height does. Weight changes during a season, but the listing is only updated once a year, so it is not necessarily what a player weighed that year.
Related: Height
The spot listed on the roster. It is some combination of guard (G), forward (F) and center (C), up to three of them.
It is a separate thing from what a player actually does. Two players both listed “G” can be a ball-carrier and a spot-up shooter. That is why this site does not rank by position — what a player does is the tags’ job.
Listings can change from season to season, so the letters move as the years do.
Related: Tag / Qualified players / Height
How a player entered the league. It is given as which round and which overall pick, and anyone not selected is shown as undrafted.
Where a player went is an assessment made at that moment, a separate thing from what they went on to do. Every year, someone undrafted turns into a starter.
Related: College / Experience / Age
The country the NBA has on file. It is neither where the player was born nor whom they represent.
It comes apart in both directions in the actual data. Sabonis was born in the United States and is filed as Lithuania. Embiid was born in Cameroon and plays for the United States, and is filed as Cameroon. Clarkson is filed as the United States and plays for the Philippines.
The NBA holds one value per player, and that is what is shown.
The school attended before turning pro.
Players who came straight from a club overseas, or through a minor league, have none (the field is left blank). There are routes into the league that do not pass through college, so a blank on its own means nothing.
Which season this is for the player at that point. A first season is shown as Rookie.
It counts seasons on the books, which need not be seasons actually played. A year missed entirely still counts as one.
It can sit at odds with age. The younger a player entered, the more experience they carry at the same age.
The number of games the player appeared in. A single second on the floor counts as one, with no distinction for how long.
Games spent on the bench without getting in are not included.
Related: Minutes (MP) / Qualified players
Time spent standing on the floor. It is held as a total in minutes and usually given as an average per game.
Overtime is included. The game runs 48 minutes, so an average of 36 means a player was out there for three quarters of it.
Related: Games (G) / Possession (POSS)
A field goal is any shot other than a free throw.
| Abbr. | Meaning |
|---|---|
| FGM (FG) | Made |
| FGA | Attempted |
| FG% | Percentage (FGM ÷ FGA) |
It does not separate twos from threes. The more threes a player takes, the lower their FG% falls, so comparing shooters on FG% alone will mislead you. The metrics that weigh in the value of the shot are eFG% and TS%.
Related: 3P / 3PA / 3P% / eFG% (effective field goal) / TS% (true shooting) / PTS (points)
Three-pointers made, attempted and the percentage, for shots taken from beyond the arc. They are part of FG as well — threes sit inside the field-goal totals.
Each one is worth three points, so at the same percentage a three produces 1.5 times what a two does. 3P% of 35% is as efficient as 2P% of 52.5%.
Related: FG / FGA / FG% / TS% (true shooting)
Free throws made, attempted and the percentage, awarded after a foul. They are not part of FG — they are counted separately.
A player with a lot of FTA is either going at the rim often or has a way of forcing contact.
Related: FG / FGA / FG% / TS% (true shooting)
The number of missed shots a player came away with.
| Abbr. | Meaning |
|---|---|
| OREB | Taken on offense (offensive rebounds) |
| DREB | Taken on defense (defensive rebounds) |
| REB | The two together |
The count moves with the team around a player. A shot that goes in creates no chance at all, and a strong rebounder alongside takes some of the share away. To read it against the chances available, use REB%.
Related: REB% / OREB% / DREB% / Rebound chances / Box-outs
The number of passes that led directly to a teammate’s basket.
It is only credited once the receiver scores, so a good pass that ends in a miss produces no number. The other way round, an easy pass counts as long as the shot drops.
The record involves a judgment call — how directly is directly enough. To see what share of a team’s baskets a player was part of, use AST%.
Related: AST% (assist percentage) / TOV (turnovers)
The number of times a player took the ball from an opponent and won possession. It covers picking off a pass and stripping a dribble alike.
Playing to take the ball away raises the count, but a miss leaves the defense a man short. A high number does not by itself mean good defense.
Related: Deflections / BLK (blocks)
The number of times a player got to an opponent’s shot and stopped it. Getting a hand on it and changing its flight is not enough — the shot has to miss.
It is part of protecting the rim, but the shots never taken leave no number behind. Some players have few blocks precisely because nobody comes in any more.
Related: Contested shots / STL (steals)
Losing possession without getting a shot off — a bad pass, a lost dribble, an offensive foul.
This site’s Ball security metric measures them per 100 touches. Without dividing by playing time or by how often a player has the ball, anyone who barely touches it would rise to the top automatically.
Related: TOV% (turnover ratio) / Touches / Possession (POSS)
The number of personal fouls committed. Six in a game and the player is out.
Fewer is not always better. Defending without making contact keeps the count down, so players who put their body in tend to accumulate more.
Related: Charges drawn / FT / FTA / FT%
Total points scored — twos, threes and free throws all added together.
The longer a player is on the floor, the more there are. To compare with the time difference taken out, this site scales them to 100 possessions.
Related: Per 100 possessions / TS% (true shooting) / USG% (usage rate)
How many points the team won or lost by while that player was on the floor. Score ten and concede six during those minutes and it reads +4.
The four teammates alongside and the five opponents opposite go straight into it. A reserve on a strong team tends to read high, a starter on a weak one low.
Related: Minutes (MP) / PIE (player impact estimate) / NetRtg (net rating)
The number of times a player’s own shot was blocked by an opponent.
The more a player goes at the rim, the higher it climbs, so a high number does not mean a poor shooter.
Related: BLK (blocks) / FG / FGA / FG%
Of five categories — points, rebounds, assists, steals and blocks — reaching double figures in two is a double-double, and in three a triple-double.
It is a widely used marker, and one that gets counted as a record in its own right. But the count is confined to those five, so a game built on, say, points and threes made never shows up here, however good it was.
Related: PTS (points) / REB / OREB / DREB / AST (assists)
The number of seasons with an appearance in an NBA game. A single game counts as one, so it is not the same as years on a roster — years spent only on the books do not count.
It covers every season on record (1946-47 onward), and a year with a mid-season trade still counts as one.
The column carries the same name Basketball Reference uses, Yrs.
Related: The draft table / Qualified players
One stretch of holding the ball, from gaining it until it ends in a shot, a turnover or a foul drawn. Think of it as one turn with the ball.
Teams play at different tempos, so comparing raw point totals favors players on fast ones. Scaling to possessions cancels that out.
The POSS that appears in the stats tables is the total number of times the team attacked while that player was on the floor — not the number the player finished themselves (that share is USG%).
Related: Per 100 possessions / PPP (points per possession) / PACE / USG% (usage rate)
A volume figure scaled to 100 turns with the ball. It is the unit that keeps players on fast teams from gaining an advantage in points alone.
100 is the same unit as the NBA’s official stats and Basketball-Reference’s Per 100 Possessions, so the figures compare directly with numbers from elsewhere. They look larger than a per-game average because a rotation player’s game runs about 75 possessions.
Related: Possession (POSS) / PACE
Possessions per 48 minutes — how many times the team attacked while that player was on the floor. Around 100 is typical, and higher means a faster game.
The more a team runs, the more points, rebounds and assists everyone collects in a game. Before comparing totals or per-game averages, check that this is level first. Scaling volume to possessions, as this site does, is a way of cancelling the difference.
Playing fast does not mean playing well.
A team’s own PACE is how many times that team attacked per 48 minutes. A player’s value is the team’s pace while they were on the floor, so the two sit on the same scale.
Related: Possession (POSS) / Per 100 possessions / Team ratings
The average number of points a play produces each time it is run. At 1.00, one turn with the ball is worth one point.
Across the league it sits around 1.10. The going rate differs by play type — cuts and transition run high, isolation and post-ups low. It is meant for comparing within a type, not across them.
Related: Possession (POSS) / Transition
Shooting efficiency given by (FGM + 0.5 × 3PM) ÷ FGA.
A three is worth three points, so it is counted back in as 1.5 times a two. FG% counts a three and a two as one shot each, which reads unfairly low for anyone who shoots from outside. eFG% removes that distortion.
Free throws do not go into it. That is the difference from TS%, which is the one to use when the ability to draw fouls should count too.
Related: FG / FGA / FG% / TS% (true shooting) / 3P / 3PA / 3P%
Shooting efficiency given by points ÷ (2 × (FGA + 0.44 × FTA)).
FG% has two holes in it. It counts a three and a two as one shot each, so anyone who shoots a lot of threes reads unfairly low; and free throws sit outside the calculation, so the contribution of anyone who draws a lot of fouls disappears. TS% folds in both.
0.44 is the widely used coefficient for the share of free throws that actually used up a turn with the ball — and-ones and technicals are in there too, so one free throw does not equal one possession.
Related: eFG% (effective field goal) / USG% (usage rate) / PPP (points per possession)
Of the team’s possessions while that player is on the floor, the share the player ended themselves — with a shot, a turnover, or by drawing a foul.
It measures the size of a role, not quality. At 30%, a player is finishing one possession in three, which is lead-option territory; down in the teens, the role is about making others work.
It means something read alongside efficiency (TS%). A player high in both — shooting a great deal without the efficiency falling — is the most valuable type there is.
Related: TS% (true shooting) / Possession (POSS) / TOV% (turnover ratio)
Of the field goals a player’s teammates made while that player was on the floor, the share that came off their assist.
This site’s Playmaking metric is built on it.
For a team it refers to something else. A team’s AST% is the share of made shots that carried an assist, which describes not one player’s work but how well the ball is moving.
Related: AST (assists) / Possession (POSS) / AST/TO (assist to turnover ratio) / Team ratings
Of the rebound chances that arose while a player was on the floor, the share they came away with. OREB% is on offense, DREB% on defense, and REB% the two together.
Counting rebounds per game favors anyone who plays long minutes; a share cancels that out.
This site’s Rebounding metric is built on it.
Related: REB / OREB / DREB / Rebound chances / Per 100 possessions
How often possessions end in a turnover, as a share. The denominator differs for players and teams. The same “%” is a different thing in each, so an explanation of one does not carry to the other.
| Denominator | |
|---|---|
| Player | FGA + 0.44 × FTA + assists + turnovers |
| Team | Possessions (turns with the ball) |
For a player it is the share of the plays they were involved in — shooting, assisting or losing it — that ended in a loss. Assists are in the denominator, so at the same number of turnovers, a player whose passes are connecting reads lower. Which also means a player who barely touches the ball gets no credit here.
For a team it is simply how many of its turns with the ball were thrown away.
0.44 is the share of free throws that used up a turn with the ball, the same coefficient as in TS%.
Related: Assist ratio / TOV (turnovers) / AST/TO (assist to turnover ratio) / Touches / Team ratings
Assists ÷ (FGA + 0.44 × FTA + assists + turnovers). Of the plays a player was involved in — shooting, assisting or losing it — the share that ended in an assist.
The denominator is exactly the one TOV% uses, with only the numerator changed. Put the two side by side and you can read whether the plays a player touched ended by being handed off or by being dropped.
It is not the similarly named AST/TO. That one simply divides assists by turnovers, and the total number of plays involved never enters it.
Related: AST/TO (assist to turnover ratio) / TOV% (turnover ratio) / AST (assists) / AST% (assist percentage)
Assists ÷ turnovers. It is only a division, so it is untouched by playing time or by how often a player has the ball.
Easy to read, but cutting mistakes alone will raise it. A player who only makes the safe pass reads high, so without the volume of assists (AST%) beside it the role is easy to misread.
It is not assist ratio. The names and the words are alike, but that one puts the total number of plays involved in the denominator.
Related: Assist ratio / AST (assists) / TOV (turnovers) / TOV% (turnover ratio)
Points the team scored per 100 possessions while that player was on the floor.
It is the team’s result over those minutes, not an individual’s scoring. Teammates land close to one another, and most of the spread is explained by the difference between teams. A reserve on a strong team reads high, a starter on a weak one low.
A team’s own ORtg is the points that team scored per 100 possessions. A player’s value is the team’s result while they were on, so the two sit on the same scale — above the team’s figure means the team scores more with that player out there.
Some sites use the same name for a different metric that measures the points an individual produced. When the numbers do not agree, check which one is meant first.
Related: DRtg (defensive rating) / NetRtg (net rating) / Possession (POSS) / Team ratings
Points the opponent scored per 100 possessions while that player was on the floor. Lower is better.
Like ORtg it is a team result, so it is not an individual’s defense. How a team defends is largely decided by the team, and the four teammates’ work rides along with it. A fine defender on a poor team will read badly.
A team’s own DRtg is the points that team conceded per 100 possessions. It is what team pages show as defensive efficiency.
Defense on this site is built from the records that count as the player’s own actions.
Related: ORtg (offensive rating) / NetRtg (net rating) / Matchups / Team ratings
ORtg − DRtg. How many points per 100 possessions the team came out ahead by while that player was on the floor.
Think of it as +/− put on a per-100-possession footing. Removing the difference in minutes makes it easier to compare than +/−, but the four teammates and five opponents still ride along with it all the same.
A team’s own NetRtg is widely used as a measure of how good a team really is. It often reorders teams before their win-loss records do, which makes it good for spotting the teams that are better, or worse, than their record says. It is what team pages show as margin per 100 possessions.
Related: ORtg (offensive rating) / DRtg (defensive rating) / +/− (plus-minus) / Team ratings / Points scored and allowed per game
Short for Player Impact Estimate. It takes points, rebounds, assists, steals and blocks, subtracts the failures — misses, turnovers, fouls — and gives the result as a share of the combined total for both teams in that game. It is the league’s own all-in-one metric.
Do everything in a game single-handed and it reads 100%. In practice the league leaders sit around 20%.
Rounding to a single number loses what a player is actually good at. This site shows it for reference only.
Related: TS% (true shooting) / USG% (usage rate) / PER (player efficiency rating)
An all-in-one metric that adds and subtracts a player’s production per minute — points, rebounds, assists and the rest — into a single figure. It is adjusted so that the league average always comes out at 15.0, which makes it readable on the same scale across years.
It is not PIE, despite the similar name. PIE is the league’s own, and gives the share of everything that happened in a game which that player produced.
This column alone is not supplied by the NBA but computed by this site (the NBA does not publish PER). The formula is John Hollinger’s as published, checked against the figures in circulation. It needs turnovers, so it cannot be computed before 1977-78. It has nothing to do with the sixteen graded metrics on this site.
Related: PIE (player impact estimate) / Minutes (MP)
Only the passages late in a game where the result is still open. The cut is the last five minutes within five points, and it takes in the team ahead and the team chasing alike.
Related: Minutes (MP) / “—” (blank) / Shrinkage / PTS (points)
A team’s offense and defense put on a common footing of points per 100 possessions. Offensive efficiency (ORtg), defensive efficiency (DRtg) and the gap between them (NetRtg) — the three that appear on team pages.
Points and points allowed per game leave the difference in tempo in place. A team that runs both scores and concedes more, so as they stand the numbers do not say whether the offense or the defense was any good. Dividing by turns with the ball takes that difference out.
Players carry values under the same names, and theirs are the team’s result while that player was on the floor. The scale is the same, so setting them beside the team’s own figures shows whether the team is better with them out there.
Related: ORtg (offensive rating) / DRtg (defensive rating) / NetRtg (net rating) / PACE / Points scored and allowed per game
Wins ÷ (wins + losses). This is what the standings are ordered by. There are no ties in the sport, so ordering by wins alone gives the same result.
It is conventionally written to three decimal places with the leading zero dropped, as in .780.
Related: Games back (GB) / Conference and division / Points scored and allowed per game
The distance to the leader, converted into games. It is (the gap in wins + the gap in losses) ÷ 2.
One win adds one to your own column and closes the gap to theirs by one, so each win moves it by 0.5. That is why it comes in half-game steps.
Its advantage is that it compares teams that have played different numbers of games, and unlike win percentage alone it tells you whether catching up is still on.
Related: Winning percentage / Conference and division
Geographic divisions of the league. The larger split is the conference, and each conference is split again into divisions.
The standings are ordered within each conference. Two teams on the same win percentage can sit in different places East and West. Teams play more games against others in their own grouping to keep travel down, so the strength of the schedule varies by grouping too.
Related: Winning percentage / Playoffs and play-in
After the full schedule, the leading teams play a knockout tournament — the playoffs. The higher the seed, the kinder the draw, and the more games are played at home.
Just below those places sit short games for the remaining spots — the play-in. Win and the playoffs follow; lose and the season is over.
Team pages on this site show how far that team went in a given season.
Related: Conference and division / Winning percentage
The matchups and results of the knockout rounds (the playoffs), laid out as teams advance. The figures under each team are their regular-season record, and the small number on the left is the seed — their place within the conference.
Seeds start in 1970-71 — before that no order was assigned, so they show “—” (the groupings of the day were divisions, not conferences).
The rounds are named counting back from the Finals — some seasons gave the top seeds a bye in the first round, which throws the count off from the front in those years.
The play-in is not included (those games decide who gets a place; they are not part of the knockout itself).
Related: Playoffs and play-in / Conference and division
The standings and the team numbers for that season. The figures are per game, and the order is within the grouping (the league-wide view lists them by win percentage).
Anything not recorded at the time is left out as a whole column — steals, blocks and offensive rebounds begin in 1973-74, turnovers in 1977-78, and three-pointers in 1979-80.
Conferences exist from 1970-71; before that there were only divisions. Team abbreviations are the ones in use that season.
None of this is connected to the grades on this site — grades are assigned only from 2017-18 onward, and what appears here is the NBA’s official record as it stands.
Related: Conference and division / Winning percentage
How strong a team was (the four factors on each side of the ball) and how it played (tempo, passing, where it shot from), read as its place within that season’s league. When the league as a whole shifts, the same figure means something different, so everything is converted to a rank before it is lined up.
It is available from 1996-97 — that is as far back as opponent numbers and shot locations reach. Earlier seasons have only the standings and the team stats per game.
Related: Season tables (standings and team stats) / Conference and division
Points scored and points conceded per game. The gap between them is the margin, a number that often says more about a team’s real strength than its record does.
The effect of tempo remains in it, though. A team that runs both scores and concedes more, so for comparisons the per-100-possession version (team ratings) is the more accurate one.
Points allowed are not published, so they are computed here from points scored and +/−.
Related: Team ratings / +/− (plus-minus) / PACE
The number of fouls drawn from opponents — the mirror image of the fouls a player commits (PF).
A high number means a player is going at the rim or has a way of forcing contact. It moves closely with free-throw attempts but is not the same thing — fouls drawn away from a shot are in there too.
It is recorded for teams as well, and the gap between fouls drawn and fouls committed shows which way the whistle was leaning.
Related: PF (personal fouls) / FT / FTA / FT% / Blocked attempts (BLKA)
A fast break finished before the defense can set. It is the most efficient play type there is (unsurprisingly, since the shot goes up before the other side is in place).
A player with a high share of it on the mix bar is the type who can run, or who looks up the floor off a rebound.
A teammate comes to the ball-handler and sets a wall — a screen — to shake the defender loose, and the gap is attacked. It is the most frequently used play in the modern game.
What it asks of the player with the ball (the handler) and the player setting the wall (the screener) is entirely different, so the mix bar keeps the two apart.
Related: Screen assists
Not carrying the ball but waiting in a set position, catching a pass and shooting on the spot — or attacking off one dribble.
It exists because a teammate drew the defense in, so it depends as much on how the team is built as on the individual. Its share of the league as a whole has grown over the past decade.
Running into a gap in the defense without the ball and catching it near the basket to finish. The distance is short, so the efficiency is high.
There is nothing flashy about it, but it shows whether a player keeps moving and can read the timing. This site grades it as the Cutting metric.
Teammates deliberately clear out to open up space, and the player attacks one-on-one.
It is the lead option’s play, but the defense can concentrate on one man too, so the efficiency itself runs low. Its value is as a last resort — when the clock is gone, or when the defense has set.
The mix bar shows it added together with post-ups (both are ways of creating one-on-one).
Setting up near the rim with your back to the basket, backing in and then attacking.
Across the league it has halved in a decade, from 6.8% to 3.4%, displaced by threes and the more efficient play types. Grading is by rank, so the number of players it applies to does not change, but the same “post-heavy player” carries a different weight from year to year.
The two corners, left and right, where the free-throw line meets the painted area. They bend like an elbow, which is where the name comes from. It is also known as the high post (the end nearer the basket being the low post).
It is a relay point within reach of both the rim and the perimeter, so a big man catching and distributing from here creates a starting point for the offense somewhere other than a guard’s hands. The High-Post Hub tag comes from how often a player catches here.
Related: Post up
Taking an offensive rebound and putting it straight back up.
The mix bar keeps it separate. Unlike the other types it is the result of having stayed in a good place, not a play that was created.
Related: REB% / OREB% / DREB% / Box-outs
A shot from inside the semicircle of four feet (about 1.2m) measured from the center of the rim — the arc marked on the floor. It has the highest percentage of any zone, and the most attempts.
Getting a shot here is itself the result of having beaten the defense. The high percentage reflects not just finishing but having won good position.
This site’s 2P finishing looks at this zone together with the paint and mid-range (after correcting for the different going rate in each).
A shot from inside the painted area — a rectangle 16 feet wide (about 4.9m) running 19 feet (about 5.8m) from the baseline to the free-throw line — excluding the semicircle right under the rim.
Even within “inside”, directly under the rim and a step or two back from it convert at very different rates. That is why they are counted separately.
Shots here are often released short of finishing through, so the percentage falls below the rim.
Related: Restricted area / Points in the paint / Post touches
A shot outside the paint but inside the three-point line. By distance it is roughly the band from 8 feet to 23 feet (about 2.4 to 7m).
It is worth two points and goes in less often than a three, so the league has been shrinking it for a long time. It survives all the same for when the defense has set, or the clock is gone.
The Mid-Range Shooter tag goes to players with a high share of their points from this zone. It does not mean they are good at it — it means that is where they make their living.
Related: Paint / Three-point line / Post up
A shot made from outside this line counts three. It is an arc 23 feet 9 inches (7.24m) from the center of the rim.
Along the sides the arc is cut off and runs straight. The court is 50 feet wide and the arc does not fit inside it, so those straight sections sit 3 feet (0.91m) from the sideline and only 22 feet (6.71m) from the rim.
That is why the corners alone are 1.75 feet (about 53cm) closer, and why the percentage there is higher.
Related: Corner 3 / Above the break 3 / 3P / 3PA / 3P%
A three taken from beside the sideline, where the arc runs straight. It is 22 feet (6.71m) to the rim, 1.75 feet shorter than from the arc.
Being shorter, it goes in more often. So within the same “3P%”, a player who shoots mostly corners and one who does not are shooting different things.
This site’s 3P finishing corrects corners and everything else separately before putting them together. Without the distinction, players whose role is to wait in the corner would gain an advantage.
Related: Three-point line / Above the break 3 / Spot up
A three from above the point where the arc turns straight — the break — in other words, anything that is not a corner. Straight on and at 45 degrees either side all belong here, at a distance of 23 feet 9 inches (7.24m).
Being farther out than the corners, the percentage is lower. Attempts are far more numerous here; this is where threes are taken.
Related: Corner 3 / Three-point line / Pull-up
A shot taken several steps farther back from the three-point line. The distance costs percentage, but the shot is still worth three. This site cuts it at 25 feet (7.6m) from the rim (the arc being 23 feet 9 inches, that is about a step outside it).
If the defense is coming out to the line, dropping farther back pulls them out. A player who can do this routinely is also opening up space for teammates.
Shots are recorded by distance band, so this range can be cut out and counted on its own.
Related: Shot distance / Above the break 3 / Three-point line
Shots sorted into five-foot bands (about 1.5m) by distance from the rim. It is a different cut from the zones on the floor — restricted area, paint, mid-range, three — the same shots lined up again by distance alone.
Closer is not always better. The lowest point comes just outside the restricted area, and the figure recovers a little farther out. That band is where shots released short of finishing through collect.
Related: Deep three / Restricted area / Paint
A shot taken off a pass without a single dribble.
Not having to create it leaves the body in better shape, so the efficiency per turn with the ball is high. But it leans heavily on whether teammates draw the defense in.
FG% makes it look the opposite way round. Nearly all of these are threes, and FG% counts one shot as one shot, which penalizes them. Compare on eFG% or TS%.
The Catch & Shoot Shooter tag goes to players with a high share of their threes in this form.
Related: Pull-up / Spot up / eFG% (effective field goal)
A shot taken by stopping mid-dribble and making the space yourself.
It is taken with a defender attached, so it is less efficient than catch and shoot. Even so, when the clock is gone or the defense has set, an offense with nobody who can do it has no way to finish.
The Pull-Up Shooter tag goes to players with a high share of their threes in this form.
Related: Catch and shoot / Pick and roll / Isolation
Points scored in the semicircle right under the rim and in the paint outside it, added together. It gathers up the points taken from inside, and broadcasts show it often.
But it contains two different things. Right under the rim the shots go in; outside it they do not. From the total alone there is no telling whether a player is finishing through or stopping short.
Related: Restricted area / Paint / Post up
The shot goes in and the foul comes with it, so one free throw is added. English calls it an and-one; Japanese broadcasts call it a basket count.
One attack becomes three or four points, which makes it the most valuable way a possession can end.
It is because these are mixed in that the TS% calculation puts a coefficient of 0.44 on free throws: one free throw does not correspond to one turn with the ball.
Related: FT / FTA / FT% / TS% (true shooting) / PF (personal fouls)
The number of times a player received the ball. It is used as the denominator for Ball security — how rarely a player turns it over.
The denominator is touches because that is the number of chances to make a mistake. USG% could serve instead, but it is pulled around by the size of a player’s role.
Related: TOV (turnovers) / TOV% (turnover ratio) / Possession (POSS)
Times the ball was received on the far side of half-court — the attacking half. The rest were received in a team’s own half, which shows involvement from the point of bringing it up.
Its share of all touches varies a great deal by role. A low share belongs to the player who carries the ball; a high one to the player waiting up front to receive.
The Ball Handler tag uses being on the low side of this as one of its conditions.
Related: Touches / Time of possession
The time from receiving the ball to letting it go, accumulated over a season.
Holding it longer is neither a virtue nor a fault. It simply reflects the difference between carrying the ball and catching and moving it on. At the same number of assists, one player holds it and then distributes, another taps it on the moment it arrives.
Divided by touches it becomes how many seconds a player holds it each time.
Related: Touches / Dribbles per touch
The average number of dribbles between receiving the ball and letting it go. Together with seconds per touch (time of possession ÷ touches), it describes ball handling.
The high side carries and attacks; the low side catches and moves it on.
But it is strongly tied to height. Measure a role by this alone and what you are measuring turns into whether the player is big or small.
Related: Touches / Time of possession
Times the ball was received inside the painted area — the colored rectangle under the basket.
It counts receiving, not scoring. Catching here draws the defense, which opens both the route to attack and the route back out to the space that leaves.
Whether a player shot or passed from there is recorded separately.
Related: Post touches / Post up / Touches
Times the ball was received near the rim with the back to the basket, set up in that position.
This too counts receiving, so whether a player shot or passed from there is recorded separately. Catching with the back turned and then distributing outside is one of the ways a big man becomes a starting point for the offense.
Related: Post up / Paint touches / Elbow (high post)
Times a player held the ball away from the rim and drove it in close with the dribble.
What happened at the end of it — scored, drew a foul, passed, lost it — is recorded separately, so the type that drives and finishes and the type that drives and kicks out can be told apart.
This site’s Drives metric is built from the number of them and the share that drew a foul. The Slasher and Drive & Kick tags are separated by the share that ends in a pass.
Related: Isolation / FT / FTA / FT% / Touches
The number of passes made and the number received. Whether anything came of them does not matter.
Setting the share of passes that turned into assists against the number made separates how much a player distributes from how much of it went in.
Related: AST (assists) / Potential assists / Touches
The number of passes that would have been assists had the receiver made the shot. Misses still count, so it comes out clearly higher than assists.
An assist depends on how the receiver plays; this comes closer to the work of the passer. The weaker the finishers around a player, the wider the gap between the two.
Related: AST (assists) / Passes made and received / Secondary assist (hockey assist)
Credited to the player who made the pass immediately before the assist. Borrowed from the sport on ice, it is also called a hockey assist.
When the ball swings, draws help and then finds the open man, the pass before the last one is sometimes the one that did the work. It is one of the few records that catches the build-up that never reaches the stat sheet.
Related: AST (assists) / Potential assists / Passes made and received
The total points a player’s teammates scored off their assists. Each assist is worth two or three, so it carries a little more weight than counting them.
It grows with the number of teammates who make threes, so it is pulled around by the cast. In mixing a player’s own skill with how the team is built, it is no different in kind from the assist count.
Related: AST (assists) / Potential assists / 3P / 3PA / 3P%
Rebounds split by whether an opponent was nearby when they came down: the ones won in a contest, and the ones picked up with nobody around.
The same count can hold different things. A player may simply be collecting balls nobody contests because teammates sealed them off.
Related: Rebound chances / Box-outs / REB / OREB / DREB
Times a rebound came down near a player and they left it for a teammate.
Because this is counted, the ones deferred can be taken out of the denominator when working out how many of the available chances a player secured. “Could not get it” and “chose not to” are different things, and mixing them makes players with a strong rebounder alongside read unfairly low.
This site’s Rebound conversion metric is built on that thinking.
Related: Rebound chances / REB% / OREB% / DREB%
The average distance from the rim at which a player’s rebounds came down. It is recorded in feet.
The near side seals and takes them under the basket; the far side picks up the ones that spin out. It is a separate axis from the number of them — at the same count, two players are collecting them in different places.
Related: Rebound chances / REB / OREB / DREB
The distance covered on the floor, derived from the camera tracking.
Player pages give it per game (kilometers or miles (mi) can be chosen under Settings on My page; the NBA publishes miles). Left as a total it would move almost in step with playing time, which would make it another way of saying “played a lot”.
Covering more ground does not make a player better. How much movement a role requires varies, so it is shown as one of the things that make a player who they are.
Related: Average speed / Minutes (MP)
Average speed while on the floor (km/h or miles per hour (mph) can be chosen under Settings on My page; the NBA publishes mph).
The levels differ between offense and defense, so they are given separately. Players move faster attacking and slower defending, and a combined figure alone does not say which side a player is moving on.
It averages in the time spent standing still, so it is not a top speed.
Related: Distance traveled / Minutes (MP)
Times a player got a hand on an opponent’s pass or dribble and disturbed its path. The ones that never became steals are recorded too.
A steal is a result — did the ball come loose — whereas a deflection measures the act of keeping the hands active. That makes it less exposed to luck and steadier from year to year.
Related: Contested shots
Times a player reached out to disturb an opponent’s shot. The result does not matter.
Whether a shot goes in depends heavily on the shooter’s skill, so as a measure of the defender’s work, “did they get a hand up every time” comes closer to their own ability. It is recorded separately for threes and twos.
Related: Deflections / Matchups
After a shot goes up, putting your body between the opponent and the basket to seal them off. It is work done not to get the rebound but to stop the other side getting it, so the player who does it is often not the one who ends up with the ball.
The league publishes a column for the “quality” of them too, but it moves almost identically to the count, so this site measures the count only.
Related: REB% / OREB% / DREB% / Rebound chances
Times a ball came down near a player. It is recorded by the tracking.
A rebound count on its own mixes together “was in the right place” and “came away with it”. Putting chances in the denominator isolates the second. That is this site’s Rebound conversion metric.
Related: REB% / OREB% / DREB% / Box-outs
Credited when a teammate scores as a direct result of a screen — a wall — that this player set.
It is one of the few columns that puts a number on work that never reaches the stat sheet at all, and this site grades it as the Screening metric.
Related: Pick and roll
Defense played by stepping into the path, standing your ground and drawing an offensive foul.
It is a choice to take the pain, and it owes more to willingness than to skill. Close to half of all players go a whole season without one, so while it can find the players who do it a lot, it cannot line up the ones who do not.
Related: Loose balls recovered
Times a player dived on a loose ball and came away with it for their team. It is recorded on offense and defense separately.
Together with charges drawn it makes up the Hustle metric.
Related: Charges drawn
A paired record of what happened while one player was guarding another — time spent on them, the points they scored, their shooting percentage and so on.
It is estimated from the tracking, not an assignment made by a coach. Switches and help defense move the attribution around, so a single game is noisy. Added up over a season it holds as a tendency.
This site uses it for Ace suppression, part of on-ball defense: how far a player held an opponent’s lead option (USG% of 25% or more) below that opponent’s usual scoring.
Related: Contested shots / USG% (usage rate) / Field goals defended
The shots recorded against a player as the nearest defender at the moment they went up. It gives how many were taken (the count) and how many went in (the percentage). The Editor uses the same names, with defensive metrics prefixed Opp.
It is not a record of assignments. It is attributed automatically by proximity, so it falls on whoever came over to help as well.
The count describes where a player defends. Players who guard near the rim have many; players who guard outside have few.
There are records split by location too. Alongside the overall figure, near the rim (within six feet) and threes can be taken separately. This site’s Rim protection is built from the near-the-rim part.
Related: Matchups / Versus their normal FG% / Contested shots / Restricted area
The difference between the shooting percentage opponents managed against a player and their own usual percentage. A negative figure means they were held below their norm.
Read the raw percentage and defenders who guard good shooters look worse. Subtracting each opponent’s usual level cancels that out.
The difference only shows up near the rim. It shrinks as the shot moves out, and taken as a whole almost nothing remains — the farther away the shot, the narrower the band a defender can move.
That is why this site’s Rim protection looks only near the rim.
Related: Field goals defended / Matchups / Restricted area
Awarded to the most valuable player of the regular season. It has run since 1955-56.
It is decided by a vote of the media and broadcasters, each ranking up to five players. Places carry points — 10 for first, 7 for second, 5 for third, 3 for fourth, 1 for fifth — and the totals set the order.
The playoffs do not enter it. Work in the Finals is the business of a separate award, the Finals MVP.
Related: Finals MVP / DPOY (Defensive Player of the Year) / Grade
Awarded to the best player across the NBA Finals series. Its full name is the Bill Russell NBA Finals Most Valuable Player Award.
The winner does not have to come from the winning team — in 1968-69 it went to Jerry West of the defeated Los Angeles Lakers, the only time in its history.
It is a separate vote from the regular-season MVP.
Related: MVP (Most Valuable Player)
Awarded to the best player on defense. It was created in 1982-83.
Like the MVP it is a media vote, with 5 points for first, 3 for second and 1 for third — fewer places than the MVP, so the weight sits higher up.
It is not an award decided by numbers. Blocks and steals do not line up neatly with who wins; rim protection, how a player handles switches and other work that never reaches the box score all count.
Related: MVP (Most Valuable Player) / All-Defensive Team
Awarded to the best player among those in their first year.
There has been a tie — in 1970-71 Dave Cowens and Geoff Petrie finished level and shared it.
The order of the voting can only be retrieved by machine from 2017-18 onward. Second and third place before that are not on this site.
Related: MVP (Most Valuable Player)
Awarded to the best player among those who came off the bench more often than they started.
To be eligible, a player must have appeared as a substitute in more games than they started. It does not mean playing few minutes — most winners are out there about as much as the starters.
Related: MIP (Most Improved Player)
Awarded to the player whose improvement on the previous season was greatest.
It is not an award for young players, though in practice it tends to go to players in their second to fourth year whose minutes have grown.
How improvement is measured is not fixed; it is settled by the vote. It has nothing to do with the grades on this site either.
Related: 6MOY (Sixth Man of the Year) / Grade
Awarded to the best player in clutch situations — the last five minutes of the fourth quarter or overtime, within five points. Created in 2022-23, it is a new award, formally the Jerry West Trophy.
It has nothing to do with the grades on this site — figures taken from clutch minutes alone never reached a size that could be read as a difference in ability, however many seasons were added together.
Related: Clutch / MVP (Most Valuable Player)
The season’s finest players, named in three teams of five (1st / 2nd / 3rd). It has run since 1946-47.
It became three teams in 1988-89; before that there were two (1st / 2nd). From 2023-24 the position groupings were dropped, and the top fifteen are now selected regardless of position.
Related: MVP (Most Valuable Player) / All-Defensive Team
The season’s finest defenders, named in two teams of five (1st / 2nd). It has run since 1968-69.
It is nearly twenty years older than the DPOY — the slot for honoring individual defense came first in the form of a selected team.
Related: DPOY (Defensive Player of the Year) / All-NBA Team