Offensive and defensive rating: how teams are measured per 100 possessions
Offensive and defensive rating measure points scored and allowed per 100 possessions. Learn how the ratings are calculated, why they beat per-game scoring, and where their blind spots are.
Suppose someone tells you that one team scores 118 points a game and another scores 104. The first sounds like a better offense. It might be. It might also be a team that plays at a sprint, with twenty extra trips down the floor each night, and a perfectly ordinary offense that simply has more chances. The second team might be running a slower, more patient system and getting more out of every trip. Scoring average cannot tell the two apart, because it mixes the quality of an offense with the number of times it got the ball.
Offensive rating and defensive rating were designed to pull those two things apart. They answer a clean question: if a team had the ball 100 times, how many points would it score, and how many would it allow? Because the question is posed on a fixed number of trips, the answers are comparable across fast teams and slow teams, across eras with different tempos, and across the brisk pace of a rout and the grind of a close game.
This article explains what the two ratings are, how they are calculated, how net rating comes out of them, how the idea carries into college and international ball, and the places where a careful reader should be cautious. It builds on the idea of points per possession; if that is new to you, it is worth a look first.
The idea in one sentence
A team's offensive rating is the number of points it scores per 100 possessions. Its defensive rating is the number of points it allows per 100 possessions. That is the whole definition. Everything else is about how the possessions get counted and how the numbers are used.
The formula for offense is:
Offensive rating = 100 × points scored ÷ possessions
and for defense:
Defensive rating = 100 × points allowed ÷ possessions
In a game, the two teams have almost the same number of possessions, because possessions alternate, so one count serves for both. That gives a neat symmetry. If your team scores 110 points in a game that has 100 possessions each, your offensive rating is 110 and your opponent's defensive rating, from your point of view, is also 110. If you allow 104, your defensive rating is 104. A lower defensive rating is better, which is the opposite of the offensive rating, and it is the first thing that trips up new readers.
The scale was chosen so that numbers are easy to read. Dividing points by possessions gives something like 1.10, but multiplying by 100 makes it 110. Fans can read the second number as a score on a typical 100-trip night.
An example that shows why it matters
Take two games. In the first, the winning team scores 120 points and the losing team scores 110. In the second, the winning team scores 100 and the losing team scores 90. Both margins are ten points. Which is the more decisive win?
If the first game had 100 possessions and the second had 80, then the answer changes. In the first game, the winner's offensive rating is 120 and its defensive rating is 110, a gap of 10 per 100 possessions. In the second, the winner's offensive rating is 125 and its defensive rating is 112.5, a gap of 12.5 per 100 possessions. The slower game was the more lopsided one on a per-trip basis, even though both ended with the same margin. A ten-point win in a game of 80 possessions is a bigger edge per trip than a ten-point win in a game of 100.
Scale this up and you see why the ratings matter for season-long comparisons. A team that plays fast will have bigger point totals and bigger margins than an equivalent team that plays slow. The per-100 view removes that artifact, which makes it possible to compare teams that play very differently. The tempo side of the story is covered in the article on pace.
How possessions are counted
Because possessions are not printed in the standard box score, they have to be derived or tracked. The classic box-score estimate adds a team's field-goal attempts, 0.44 of its free-throw attempts and its turnovers, then subtracts its offensive rebounds. The 0.44 reflects the fact that only some free throws end a possession; the article on true shooting percentage and effective field-goal percentage explains the factor in more detail.
Some versions of the formula refine this by averaging the team's possession count with its opponent's, since both teams must have nearly the same number, and by adjusting for the details of how offensive rebounds relate to the opponent's missed shots. These refinements reduce the error but do not eliminate it.
Modern data providers do not need to estimate. Play-by-play data identifies the start and end of each possession, and tracking data confirms it. As a result, the ratings published on different sites can differ by a point or two for the same team. When the figures do not agree exactly, the cause is usually the possession count: one source is estimating from the box score and another is using the tracked event data. Neither is wrong in a deep sense, but you should compare ratings from one source at a time. The data sources page describes where the numbers on this site come from.
Net rating: the difference that matters
Subtract the defensive rating from the offensive rating and you get net rating, the number of points a team outscores its opponents per 100 possessions. A team with an offensive rating of 114 and a defensive rating of 108 has a net rating of plus 6.
Net rating is useful because it is a close cousin of point differential, but with the tempo noise removed. Teams with strong net ratings tend to win more games, and over a full season the relationship is tight. Many analysts treat net rating as the best single indicator of a team's real quality, better than win-loss record, because the record is partly the product of luck in close games while the net rating captures how the team performed across all of its games. The case for that view is made in the article on net rating, and the way points convert into expected wins is the subject of Pythagorean wins.
A rough sense of scale helps. In any recent season, a net rating of plus 5 or better is the mark of a clear contender, and the very best teams of an era can reach into the double digits. A net rating near zero is an average team. A strongly negative net rating means a team that is outscored badly. These ranges are broad rules of thumb, and the league's overall levels shift from year to year.
Why the league average moves
An offensive rating of 110 means different things in different years. In the 1990s, when scoring was depressed by slow tempos and physical defense, an offensive rating of 110 would have been outstanding. In more recent seasons, with better spacing and heavy three-point shooting, league-average offensive efficiency has risen, and 110 is closer to ordinary.
Because the league's average offensive rating and its average defensive rating are equal, as every point scored by one team is allowed by another, the average net rating is exactly zero in every season. What changes is the level at which offense and defense are both measured. So the fair way to compare teams across eras is to look at ratings relative to the league average of the same season. A team that scores 8 points per 100 possessions better than the league average has a strong offense whether the league average is 103 or 114.
This is the same principle that applies to hitting statistics in baseball: raw numbers are less informative than numbers set against the environment of their time. If you are curious why the environment shifted, the articles on the three-point revolution in the numbers and pace and space cover the main causes.
Individual ratings
The same terms appear on player pages, which can cause confusion, so it is worth sorting out what they mean.
Team ratings while a player is on the floor. This is the version most often seen in lineup and on/off data. It reports the team's offensive and defensive efficiency during the minutes a player was on the court. It is a team figure with a player's name attached, and it depends heavily on who else was playing. The article on plus-minus and on/off covers why that dependence matters.
Estimated individual ratings. Dean Oliver, whose book Basketball on Paper from 2004 laid out much of the possession-based approach, proposed estimating the points a player produces per 100 possessions he uses, and the points a player allows per 100 possessions on defense. The offensive version is built from the player's scoring, assists and offensive rebounds, adjusted for his share of team possessions. The defensive version is built from the team's defensive results, distributed among the players according to their steals, blocks and rebounds. These individual ratings are estimates, particularly on defense, where the box score offers little information about who was responsible for a stop.
Because of that last point, an individual defensive rating is the weakest of the figures. A player on a great defensive team will look good by default, and a player on a weak one will look worse, whatever his individual contribution. The reasons defense is hard to credit are discussed in the article on defensive metrics.
College and international play
The per-100 approach translates readily to other levels of the game. In men's college basketball, the best-known efficiency ratings, such as the KenPom ratings and the NET used by the selection committee, are built on points per 100 possessions, adjusted for the strength of opponents and the site of the game. The reasoning is identical to the NBA's: games between teams with different tempos need to be compared per possession, and a blowout against a weak opponent tells you less than a close game against a strong one. The article on college efficiency metrics explains how those ratings are built and how they enter the selection process.
College games are shorter, 40 minutes rather than 48, and contain fewer possessions. A college offensive rating is still expressed per 100 possessions, so it can be compared with the NBA's in principle, but the levels are not the same and the competition is not the same. A rating of 115 in one league does not translate into 115 in another. Comparisons across leagues need the context of each league's average.
International competitions follow the same logic. FIBA play uses 40-minute games and different rules about the three-point line and the shot clock, so possession counts differ again. The principle holds: whenever you want to say which team is better at scoring or stopping scoring, count per trip.
Where the ratings fall short
Offensive and defensive rating are among the most reliable tools in the analyst's kit, but they are not complete. Several limits are worth keeping in mind.
Opponent strength. A raw rating takes no account of whom the team played. A team with a soft schedule will have inflated ratings. Adjusted versions correct for it, as explained in the article on strength of schedule.
Garbage time. When a game is decided, teams change how they play. Starters sit, tactics loosen, and the final minutes produce possessions that are not representative. Some data providers filter out those minutes. Others do not. The difference affects teams that blow out opponents often.
Lineups and injuries. A season-long rating averages over all of a team's lineups, including ones that rarely play together in the future. A team whose best player missed a third of the season has a rating that blends the versions. For understanding what the team is when everyone plays, lineup-level data is more informative, though also noisier; see lineup data and five-man units.
Sample size. Ratings over a few games swing widely. Shooting luck can move an offensive rating by several points over a short stretch. Over a full season the noise shrinks, but even then a team's opponent three-point shooting can vary in ways that a defense does not control. The point is covered in the article on small samples.
What the ratings leave out. They report outcomes, not causes. A team with a poor offensive rating could be missing shooters, turning the ball over, failing to rebound, or sending opponents to the line too rarely. To see which, you need to break the rating apart. The four factors do exactly that, splitting each end of the floor into shooting, turnovers, rebounding and free throws.
How to read a rating line
When you open a team page and see an offensive rating, a defensive rating and a net rating, a quick routine will get you most of the value.
- Compare the offensive and defensive ratings with the league average. Which end carries the team?
- Look at net rating for the overall picture, and compare it with the team's record. A strong net rating with a mediocre record often reflects bad luck in close games. A good record with a modest net rating often reflects good luck.
- Check the sample. Early in a season, treat the numbers as provisional.
- If the team's profile looks surprising, break it into the four factors to see where the strength or weakness lies.
- Remember that the numbers describe what happened, not what will happen. The standings page shows records; the ratings help explain why the records look the way they do.
A note on defenses and what they control
Fans sometimes assume that a defensive rating is wholly a measure of defensive skill. It is mostly that, but not entirely. A defense controls many things: the shots it allows, the turnovers it forces, the free throws it gives up, the rebounds it secures. It controls less of what happens once a shot goes up. Opponents' three-point percentage on open shots, for example, varies from team to team and season to season in ways that appear to be partly luck. Analysts who study this find that teams have much more control over the quality of shots they allow than over whether those shots fall.
That does not make defensive rating meaningless. Over a season the luck evens out and the rating reflects the sum of what a defense does. It just means that a surprising drop or jump in defensive rating over a short span should be read with skepticism, and that explanations built on a handful of games are usually less certain than they sound.
What to take away
Points per game confuse how good an offense is with how often it gets the ball. Offensive and defensive rating separate the two by counting per 100 possessions, which lets you compare a sprinting team with a deliberate one, one era with another, and one level of the sport with another so long as you compare to the league average. Subtracting the two gives net rating, one of the most trusted summaries of a team's real strength.
Use the ratings as a starting point. They tell you how good a team has been, and the four factors, the lineup data and the schedule context tell you why. When a team's scoring average and its rating tell different stories, trust the rating, and ask what made the average mislead.