Basketball stats glossary: from PER to RPM in plain English
A plain-English glossary of modern basketball statistics, from true shooting and usage to PER, win shares, BPM, VORP and RPM, with what each one measures, how to read it and where it misleads.
Basketball statistics used to fit on a single line: points, rebounds, assists. Today a broadcast might mention a player's true shooting percentage, his usage rate, his net rating and some kind of all-in-one number with an acronym that sounds like a government agency. The vocabulary can feel like a barrier. It does not have to be. Almost every advanced statistic is an attempt to answer a plain question, such as how efficiently a player scores, how much of an offense he uses, or how much a team improves with him on the floor.
This glossary explains the most common modern basketball statistics in everyday language. For each term it gives the name, the question it answers, how it is built in outline, how to read a typical value and what the number tends to miss. It is organized into groups: shooting, possession and role, team measures, all-in-one player ratings, and plus-minus metrics. You do not need to memorize formulas. The goal is to know what a number is trying to say, so you can tell when it is helping an argument and when it is just decoration.
A word of caution first. Every statistic here is an estimate built from the box score or from tracking data, and none of them captures everything a player does. Defense, in particular, is hard to count. The best use of these numbers is as a starting point for a conversation, and the best conversations combine more than one of them with what you see on the court.
Shooting efficiency
Field-goal percentage (FG%) is the share of shots that go in. It is the oldest efficiency measure, and its flaw is that it treats every shot as equal. A three-pointer that goes in is worth more than a two-pointer, and a free throw is worth less, but FG% does not know.
Effective field-goal percentage (eFG%) fixes the first problem. It gives extra credit for three-pointers, because each is worth 50 percent more than a two. The formula is field goals made plus half of the threes made, divided by field goals attempted. A player who shoots 45 percent on two-point shots and makes a lot of threes will usually have a much higher eFG% than FG%. Values around 50 percent are typical, and well above that signals an efficient shooter.
True shooting percentage (TS%) goes one step further by adding free throws. It measures points per shooting possession, using a standard estimate for how many possessions free throws use. In outline, it is points divided by twice the sum of field-goal attempts and a fraction of free-throw attempts. TS% is the best single number for scoring efficiency, because it counts everything. If you read nothing else, read this one. We cover it in depth in true shooting percentage and effective field-goal percentage, explained.
Free-throw rate (FTr) is free-throw attempts divided by field-goal attempts. It shows how often a player gets to the line. A high rate means that a player draws fouls, which is valuable because free throws are an efficient way to score.
What they miss. Efficiency numbers say nothing about difficulty. A player who takes only open shots will often look more efficient than a player who creates his own shots against pressure. That is why efficiency should be read next to usage, which brings us to the next group.
Possessions, role and pace
Usage rate (USG%) estimates the share of a team's possessions that a player ends while he is on the floor, through a shot, a trip to the free-throw line or a turnover. A high usage rate means the offense runs through him. Typical starters are around the low 20s in percentage terms, and stars are often higher. Usage is not a measure of quality. A high-usage player can be great or wasteful, which is why it should be paired with efficiency. Our fuller treatment is in usage rate: how much of the offense a player uses.
Assist percentage (AST%) is the estimated share of teammates' made field goals that a player assisted while on the floor. It adjusts for playing time and pace, so it is fairer than assists per game. Turnover percentage (TOV%) estimates turnovers per 100 plays. Together they describe a playmaker's value and cost.
Rebound percentage (REB%) is the share of available rebounds a player grabs while on the floor. Like AST%, it removes the effect of pace and minutes. There are offensive and defensive versions. Steal percentage and block percentage work the same way, expressed per opponent possession or opponent shot attempt.
Pace is the number of possessions a team uses per 48 minutes. A fast team has more possessions, which inflates raw totals such as points and rebounds. Pace is why per-game numbers from different eras or different teams are not directly comparable.
Per-36 and per-100-possession numbers rescale a player's production to a standard amount of playing time, or to a standard number of possessions. They are useful for comparing players who play different minutes. They are less useful for predicting what would happen if a bench player played starter's minutes, because the role and the opponents would change.
Team ratings
Offensive rating (ORtg) and defensive rating (DRtg) are points scored and allowed per 100 possessions. Using possessions rather than games removes the distortion of pace: a slow team that scores 100 points may be more efficient than a fast team that scores 105. Both ratings apply to teams, and there are estimated individual versions as well. The team versions are the foundation of modern analysis.
Net rating is offensive rating minus defensive rating. It is the point margin per 100 possessions, and it is one of the best predictors of how good a team really is, better than its win-loss record in a small sample. A team with a strong net rating but a mediocre record has usually been unlucky in close games. A team with the reverse has probably been fortunate. The idea is explained in net rating.
The four factors are shooting (eFG%), turnovers, rebounding and free throws. They were identified by the analyst Dean Oliver as the elements that most strongly explain winning, and they apply on both offense and defense. A team's profile across the four factors tells you how it wins: by shooting well, by protecting the ball, by controlling the glass or by getting to the line.
The all-in-one ratings
An all-in-one statistic tries to boil a player's contribution into a single number. They are tempting, and each has weaknesses.
Player efficiency rating (PER) was created by John Hollinger. It adds up positive contributions such as points, rebounds, assists, steals and blocks, subtracts negatives such as misses and turnovers, and scales the result to a league average of 15. A PER above 20 is excellent, and the top seasons in history go well beyond that. PER is easy to calculate from the box score, but it leans toward volume scoring and tends to underrate players whose value is in defense or in work that does not appear in a box score. The details are in player efficiency rating: what it measures and what it misses.
Win shares (WS) attempts to divide a team's wins among its players. It is built from offensive and defensive components, and it is expressed as a total or as win shares per 48 minutes, where the league average is about 0.100. The idea has roots in baseball analysis. Win shares favor players on good teams and with big minutes, which is both a feature and a bug.
Box plus-minus (BPM) uses a regression to estimate a player's contribution from box-score numbers, in points per 100 possessions relative to an average player. Zero is average, and positive numbers are above average. It is better at including defensive and rebounding signals than PER, but it is still based on the box score.
Value over replacement player (VORP) turns BPM into a cumulative total by comparing a player with a hypothetical replacement-level player and multiplying by playing time. It is a useful way to compare players across seasons, since it rewards both quality and quantity. All three of these are covered in win shares, BPM and VORP.
Reading an all-in-one number
A few habits help. Treat a difference of a few tenths as noise. Compare players in similar roles. Look at several metrics, and if they disagree, ask why. And remember that all of them are built from the same underlying events, so agreement among them is less independent than it looks. When two all-in-one numbers diverge, the explanation usually lies in how they treat defense, usage or team context.
Plus-minus and its descendants
Plus-minus is the simplest measure: the team's point margin while a player is on the floor. If a team outscores its opponent by 8 while a player is in the game, his plus-minus is plus 8. The problem is that it depends enormously on teammates and opponents. A bench player who plays with the starters in a blowout can post a spectacular number. For that reason, the raw figure is unreliable in small samples. Our explanation is in plus-minus, on/off and why the raw number lies.
On/off differential compares the team's net rating with a player on the court and with him on the bench. It is an improvement, because it accounts for the team's baseline, but it still mixes in the effect of the teammates who share the floor.
Adjusted plus-minus tries to solve the problem with statistics. It uses a regression across thousands of lineups to estimate each player's effect while adjusting for the teammates and opponents he played with. The raw version is noisy, and later versions use a technique called regularization to stabilize the results.
Real Plus-Minus (RPM) is a metric introduced by ESPN in the mid-2010s that builds on adjusted plus-minus and adds a box-score element to improve stability. It is expressed as points per 100 possessions above average, and it is often split into offensive and defensive components. Other public models with similar goals exist under names such as EPM, LEBRON and DARKO. They share an approach: use play-by-play lineup data to estimate each player's impact, with various adjustments.
What they miss. Plus-minus metrics are the closest we have to measuring defense and off-ball contributions, but they need many possessions to be reliable, and they can be unstable from year to year. Their results depend on modeling choices, and two models can disagree about the same player.
A quick reference
The table below gathers the main measures and the question each one answers.
| Measure | The question it answers |
|---|---|
| eFG% | How efficient is the shooting, counting threes? |
| TS% | How efficient is the scoring, counting free throws? |
| USG% | How much of the offense does this player use? |
| AST%, TOV% | How much does he create, and how often does he waste possessions? |
| REB% | What share of the available rebounds does he secure? |
| Pace | How many possessions does the team play? |
| Net rating | How much does the team outscore opponents per 100 possessions? |
| PER | How productive is he per minute, in one number? |
| WS | How many team wins can be assigned to him? |
| BPM, VORP | How good is he, and how much value has he provided? |
| RPM and similar | How much does the team improve with him on the floor? |
Terms from the tracking era
Since cameras began following every player in NBA arenas, a second layer of vocabulary has appeared. You will meet these terms in broadcasts and analysis, and they are easier to follow once you know what they are reaching for.
Shot quality is an estimate of how likely a shot is to go in, given where it was taken, how closely it was defended and sometimes the type of shot. A shooter's actual results can then be compared with the expectation. A player who regularly beats the expectation is either an excellent shooter or fortunate, and a longer sample usually tells which. Expected points applies the same idea to possessions.
Contested and uncontested shots describe how close the nearest defender was when the shot was released. They help separate the shooter who creates his own looks from the one who is fed by teammates. Touches, passes and time of possession describe how the ball moves and how long a player holds it. Hustle statistics, such as deflections, loose balls recovered and screen assists, capture small actions that the box score ignores.
Defensive field-goal percentage allowed tracks how well opponents shoot when a given player is the nearest defender. It sounds like the answer to the defensive-measurement problem, but it has its own flaws. A defender who guards weak shooters, or who is assigned to the player who rarely shoots, will look excellent. Because of that, many analysts still prefer team-level defensive ratings and on/off numbers, and they check the numbers against video.
Replacement level is a concept borrowed from baseball. It describes the level of play that a team could get from a freely available player, such as one signed from the G League or off the waiver wire. Statistics such as VORP count value above that level, which explains why they give credit to players who are merely good but play a lot of minutes, and not only to stars.
How to use the numbers
Having the vocabulary is only the first step. Here are a few principles that experienced analysts follow.
- Match the metric to the question. Use TS% for scoring efficiency, rebound percentage for rebounding, and net rating for team quality. Using the wrong tool leads to wrong conclusions.
- Respect the sample. A hot month is not a talent. Even large numbers of games can be misleading for statistics that depend on small events such as blocks or steals.
- Consider context. A statistic reflects a role, a team and a style of play. A player on a bad team with the ball in his hands will post big usage and low efficiency, and a role player on a good team will look efficient because the team creates open shots for him.
- Use more than one number. Efficiency, usage and impact each tell part of the story. Together they are much more useful than any single metric.
- Remember defense. Most of these measures are strongest on offense. If a metric makes a defensive specialist look ordinary, the metric is probably the problem.
These principles also explain why analytics and film are partners. A number can point to a pattern, such as a defender who allows few points at the rim, but the video explains how it happens. The best evaluations use both.
If you want to see these measures in action, the leaders page ranks players across the standard categories, and the about our data page explains how often the underlying numbers are refreshed.
What to take away
Modern basketball statistics are tools for answering specific questions. Shooting measures describe efficiency, usage and percentage stats describe role, team ratings describe the quality of a lineup, all-in-one numbers summarize production, and plus-minus models try to capture impact on winning. Each has limits, and no single number ends an argument.
Learn the question behind each number, and you can read almost any stat line. When you meet a new acronym, ask what question it was built to answer and what it leaves out. That habit will serve you better than any formula.