Learn the metrics

Every metric, explained

Plain-English definitions for every stat across the app, from net rating to Synergy play types, with a note on which direction is good.

How to read the numbers

A few conventions used across every page.

League percentile

Most colored cells are shaded by league percentile, a value's rank against every qualifying player, lineup, or team in the same season type (above a minutes or possessions minimum). Green is the top of the league, red is the bottom, so very different stats compare on one 0–100 scale.

Per 100 possessions

Rate stats are normalized to 100 possessions so fast- and slow-paced teams compare fairly. 'Totals' instead show raw cumulative counts.

Season type

Stats split into Regular Season and Playoffs. Playoff samples are much smaller, so treat playoff splits as directional.

Efficiency & ratings

Per-possession measures of how good a team or unit is.

OFF_RATINGHigher

Offensive Rating - Points scored per 100 possessions

DEF_RATINGLower

Defensive Rating - Points allowed per 100 possessions (lower is better)

NET_RATINGHigher

Net Rating - Point differential per 100 possessions (Offensive Rating - Defensive Rating)

CHEMISTRYHigher

Chemistry - How far this five-man unit outperforms the sum of its parts: actual net rating minus what our RAPM model predicts from the ten players on the floor (per 100, opponent-adjusted). Our estimate; Regular Season lineups with enough possessions only

WPAHigher

Scoring WPA - Win probability added by made baskets, weighted by leverage: a buzzer-beating game-winner is worth ~+0.9 wins, a garbage-time bucket ~0. Misses aren't debited, so it reads as leverage-weighted scoring. Our estimate from play-by-play

PACE

Pace - Number of possessions per 48 minutes

PIEHigher

Player Impact Estimate - Overall statistical contribution measure

PLUS_MINUSHigher

Plus/Minus - Point differential when lineup is on the court

Shooting

How efficiently shots go in.

EFG_PCTHigher

Effective Field Goal Percentage - Adjusted field goal percentage that accounts for 3-pointers being worth more

TS_PCTHigher

True Shooting Percentage - Shooting efficiency that accounts for 2PT, 3PT, and free throws

FG_PCTHigher

Field Goal Percentage

FG3_PCTHigher

Three-Point Field Goal Percentage

FT_PCTHigher

Free Throw Percentage

Playmaking & ball security

Creating shots for others and protecting the ball.

AST_PCTHigher

Assist Percentage - Percentage of field goals that were assisted

AST_RATIOHigher

Assist Ratio - Percentage of possessions that end in an assist

AST_TOHigher

Assist to Turnover Ratio - Number of assists per turnover

TOV_PCTLower

Turnover Percentage - Percentage of possessions that end in a turnover

Rebounding

Share of available rebounds secured.

REB_PCTHigher

Rebound Percentage - Percentage of available rebounds grabbed by the team

OREB_PCTHigher

Offensive Rebound Percentage - Percentage of available offensive rebounds grabbed

DREB_PCTHigher

Defensive Rebound Percentage - Percentage of available defensive rebounds grabbed

Box score

The classic counting stats.

PTSHigher

Points

REBHigher

Total Rebounds

ASTHigher

Assists

STLHigher

Steals

BLKHigher

Blocks

TOVLower

Turnovers

PFLower

Personal Fouls

FGM

Field Goals Made

FGA

Field Goals Attempted

FG3M

Three-Point Field Goals Made

FG3A

Three-Point Field Goals Attempted

FTM

Free Throws Made

FTA

Free Throws Attempted

OREB

Offensive Rebounds

DREB

Defensive Rebounds

GP

Games Played - Number of games this lineup has played together

MIN

Minutes played by the lineup

W_PCTHigher

Win Percentage - The percentage of games won when this lineup plays together

WOWY / on-off

How a team changes with a player on versus off the floor.

On / Off rating

The team's net rating per 100 possessions while the player is on the court versus on the bench.

ΔNET (net swing)Higher

On-court net rating minus off-court net rating. Positive means the team plays better with the player on the floor.

Padding (regression to mean)

Small on/off samples are noisy. A player with 80 minutes can post a wild swing. We 'pad' ΔNET by shrinking it toward zero based on on-court minutes, so a big number only survives if enough minutes back it. The adjusted figure is more trustworthy than the raw one; leaderboards sort by it.

Reading WOWY safely

On/off is shaped by which teammates and opponents share the floor and by sample size. It's a correlation, not a clean causal effect. Pair it with lineup and play-type data before drawing conclusions.

Clutch

Performance when the game is on the line.

Clutch time

The last 5 minutes of a game with the score within 5 points.

Clutch net ratingHigher

Net rating per 100 possessions during clutch time.

Clutch eFG% / TS%

Shooting efficiency in clutch time. Useful, but clutch possessions accumulate slowly. Single-season clutch splits are noisy.

Clutch lift (leverage)Higher

A player's on-court net rating in clutch time minus their overall on-court net rating. Does the team get better or worse with them on the floor late? Padded by clutch minutes (which are few), so treat it as directional.

Play types (Synergy)

How possessions are used, and how efficient each type is.

Play type

A Synergy classification of how a possession is used: isolation, pick-and-roll ball handler or roll man, post-up, spot-up, hand-off, cut, off-screen, transition, putbacks, and miscellaneous.

PPP (points per possession)Higher

Points produced per possession finishing in this play type. The headline efficiency number. Roughly 1.00 is league-average.

Frequency (POSS%)

Share of a player's or team's possessions that come from this play type, how often they run it.

Synergy percentileHigher

League rank (0–100) for PPP within a play type. 90 means more efficient than 90% of the league.

Offensive vs defensive

Offensive measures the player or team running the action; defensive measures how they defend it. On defense, a lower PPP allowed is better.

Player tracking

Movement- and touch-based stats from optical tracking.

Drives

Touches that start at least 20 feet out and are dribbled within 10 feet of the rim, excluding fast breaks.

Catch & shoot

Jump shots taken right after a catch with no dribble and a defender no closer than ~4 feet, a proxy for spot-up shooting.

Pull-up

Jump shots taken off the dribble, a proxy for self-created shot-making.

Touches & time of possession

How often a player touches the ball and how long they hold it, a read on usage and on-ball role.

Rim protectionLower

Opponent field-goal percentage at the rim when this player is the closest defender, lower is better.

Hustle

Effort plays the box score misses.

DeflectionsHigher

Times a defender gets a hand on the ball outside of steals, a disruption signal.

Loose balls recoveredHigher

Loose balls a player secures for their team.

Charges drawnHigher

Offensive fouls drawn by taking a charge.

Screen assistsHigher

Screens that directly lead to a teammate's made basket.

Box-outsHigher

Times a player boxes out an opponent to help secure a rebound.

Contested shotsHigher

Shots a defender contests by closing out with a hand up.

Advanced impact (our estimates)

Self-computed impact metrics. See the Methodology page for exactly how each is built and what it can't tell you.

IPMHigher

Informed Plus-Minus - our flagship impact estimate: RAPM re-fit to shrink each player toward a box-score prior trained on our own nine seasons of RAPM, instead of toward zero. Stable and current-season; beat pure RAPM in 17 of 17 held-out tests. Our estimate; see Methodology

O_IPMHigher

Offensive IPM - the offensive half of Informed Plus-Minus. Our estimate

D_IPMHigher

Defensive IPM - the defensive half of Informed Plus-Minus. Positive is good defense. Our estimate

RAPMHigher

Regularized Adjusted Plus-Minus - our ridge-regression estimate of a player's per-100 impact, controlling for teammates and opponents, from reconstructed play-by-play. Single-season is noisy. Our estimate; see Methodology

RAPM_3YRHigher

3-year RAPM - the RAPM fit pooled over the last three seasons with recency weighting. The stable read on a player's impact; prefer it over single-season RAPM. Our estimate

O_RAPMHigher

Offensive RAPM - the offensive half of RAPM (per-100 impact on team offense, opponent- and teammate-adjusted). Our estimate

D_RAPMHigher

Defensive RAPM - the defensive half of RAPM. Positive is good defense. Our estimate

WPAHigher

Scoring WPA - Win probability added by made baskets, weighted by leverage: a buzzer-beating game-winner is worth ~+0.9 wins, a garbage-time bucket ~0. Misses aren't debited, so it reads as leverage-weighted scoring. Our estimate from play-by-play

CHEMISTRYHigher

Chemistry - How far this five-man unit outperforms the sum of its parts: actual net rating minus what our RAPM model predicts from the ten players on the floor (per 100, opponent-adjusted). Our estimate; Regular Season lineups with enough possessions only

BPMHigher

Box Plus/Minus (BPM 2.0) - per-100 contribution estimated from the box score + team context. Our estimate on NBA.com data; tracks Basketball-Reference within ~0.2-0.3

VORPHigher

Value Over Replacement Player - BPM scaled by minutes against a replacement-level baseline. Our estimate

SHOTMAKING_OVER_XEFGHigher

Shot-making over expected - actual eFG% minus the league-average eFG% on the player's exact shot-location mix (in eFG points). Isolates finishing from shot selection. Our estimate

BOX_CREATIONHigher

Box Creation (Ben Taylor) - open shots created for others per 100 possessions, from the box score. Our estimate; see Methodology

OFFENSIVE_LOADHigher

Offensive Load (Ben Taylor) - the share of a team's offense a player shoulders (creation + scoring load + turnovers), per 100. Our estimate

TEAM_IPMHigher

Team IPM - minutes-weighted sum of player IPM (five on court): what the roster is worth. Predicts future game margins better than results-based ratings in our held-out tests. Our estimate; see Methodology

SRSHigher

SRS - point margin adjusted for opponent and venue, solved by least squares over every game. The results-based team rating. Our estimate

O_SRSHigher

Offensive SRS - the offensive half of SRS (points scored, opponent-adjusted, per game). Our estimate

D_SRSHigher

Defensive SRS - the defensive half of SRS. Positive is good defense. Our estimate

SOS

Strength of schedule - average opponent SRS a team actually faced. Higher = harder schedule. Our estimate

OSOS

Offensive SOS - average opposing defense faced (higher = tougher defenses for your offense). Our estimate

DSOS

Defensive SOS - average opposing offense faced (higher = tougher offenses for your defense). Our estimate

Estimated metrics

NBA's modeled, single-number player impact estimates.

Estimated (E_) metrics

Regression-based estimates of player impact (E_OFF_RATING, E_DEF_RATING, E_NET_RATING, E_USG_PCT, E_PACE and more), modeled from box-score and contextual inputs rather than measured directly. Handy as quick single-number ratings.

E_NET_RATINGHigher

Estimated point differential per 100 possessions attributable to the player.

E_USG_PCT

Estimated share of team plays a player uses while on the floor.