Net rating: the best single number for a team
Net rating is the gap between the points a team scores and allows per 100 possessions. Here is how it works, why it predicts so well, and where it can mislead you.
If you could look at only one number to judge how good a basketball team is, which would you pick? Wins and losses are the obvious answer, but a record is a blunt instrument. A team can win ten close games by a point each and look dominant on paper, or lose ten close games and look mediocre while playing like a contender. Points per game is better but still flawed, because it ignores how fast a team plays and what it gives up at the other end. The number that analysts return to again and again is net rating.
Net rating is simple to state: it is a team's offensive rating minus its defensive rating, where each of those is measured in points per 100 possessions. A team that scores 115 points per 100 possessions and allows 110 has a net rating of plus 5. That one figure folds together how well a team shoots, how well it defends, how often it turns the ball over, how it rebounds and how often it gets to the free-throw line, and it does so on a scale that does not care whether the team plays fast or slow.
This article explains where net rating comes from, why it tracks winning so closely, how to read it for a team, a lineup or a single player, and, just as important, where it lets you down. It builds on the ideas in offensive and defensive rating: how teams are measured per 100 possessions, so if you want the full mechanics of those two inputs, start there. You can see the underlying team numbers on the standings and leaders pages.
What net rating actually measures
Start with a possession. A possession ends when a team scores, turns the ball over, or gives it up after a defensive rebound. Over a game, the two teams have almost the same number of possessions, because each possession for one side ends in a way that starts a possession for the other. Counting them precisely requires play-by-play data, but a standard estimate works from the box score: field-goal attempts, minus offensive rebounds, plus turnovers, plus a fraction of free-throw attempts, usually about 44 percent, to account for trips to the line that end possessions only on the last shot.
Once you have possessions, you can ask how many points a team produces per 100 of them. That is offensive rating. Flip the question to the opponent and you get defensive rating. Subtract one from the other and you have net rating. A positive net rating means a team outscores opponents on a per-possession basis; a negative one means the opposite.
The use of 100 is a convention, not a law of nature. It gives numbers on a friendly scale: average NBA teams hover near the league average each season, which in recent decades has moved from the low hundreds up toward the 110s and beyond, and the best teams have net ratings that are comfortably positive, usually in the high single digits or beyond. The exact league average changes from year to year, so it is always more useful to compare a team with its own season's league than with a team from a different era.
One subtlety is worth knowing. Net rating is not the same thing as average point margin. A team that plays at a fast pace has more possessions per game, so its point margin is stretched compared with its net rating; a slow team's margin is compressed. Net rating removes that stretching. We cover the margin version of the story in margin of victory and what it says about a team's true level, and the effect of tempo in pace: possessions per game and why it changes everything.
Why per 100 possessions beats per game
Imagine two teams that each average 105 points per game. One plays at a crawl, with about 90 possessions a game. The other races, with about 105. The slow team is scoring 1.17 points per possession and the fast team just 1.00. Per game, they look identical; per possession, one of them has an offense that is far more efficient. If you evaluated them by points per game alone, you would be confusing speed with skill.
The same trap exists on defense. A team that allows 100 points a game might look like a defensive stalwart, but if it plays so slowly that its opponents get only 85 possessions, those 100 points are a poor defensive result. By contrast, a fast team allowing 108 points might be giving up fewer points per possession than the slow team. Without a per-possession view, the slow team gets credit it has not earned and the fast team gets blamed for something it did not do.
This is why the shift toward possession-based thinking was such an important step in basketball analysis. When you hold the number of opportunities constant, you can finally compare teams that play different styles. Net rating inherits that benefit. A team built around a deliberate half-court offense and a team built around running at every opportunity can be placed on the same ruler.
The reason one number works
Why would a single subtraction predict so much? Because the thing that wins a game is, quite literally, outscoring the opponent. Net rating measures exactly that, on a per-opportunity basis, over many games. Every other statistic, whether it is three-point percentage, rebounding or turnovers, matters to the extent that it feeds into scoring and preventing scoring. Net rating sits at the end of that chain, so it captures everything upstream.
The practical consequence is a very strong link between net rating and winning. Across a full 82-game NBA season, each point of net rating is worth, as a rough rule, about two and a half wins. A team with a net rating of plus 5 should therefore win somewhere around 50 or 51 games, a bit above the mid-40s that a plus 2 team would collect, and a team at plus 10 should land in the high 60s. This is the same logic that underlies Pythagorean wins, which turns points scored and allowed into an expected record. The two ideas are close cousins: Pythagorean wins uses totals, net rating uses per-possession rates, and they usually agree.
The relationship is not exact, and no one should expect it to be. A team's actual record will wander from its expected record because of luck in close games, injuries, and the order in which things happen. But over a season the wandering is limited. Teams that win far more games than their net rating suggests have often been fortunate in tight finishes, and in the following season they commonly slide back. Teams that win far fewer than their net rating suggests tend to bounce up. That pattern is an example of the broader principle in regression to the mean.
A worked example
Consider a hypothetical team, which we will call Team A, over a season. It scores 8,900 points in roughly 7,800 possessions and allows 8,500 points in roughly 7,800 possessions. Its offensive rating is about 114 and its defensive rating about 109, giving a net rating of plus 5. Now consider Team B, which scores 8,000 points and allows 7,600 points, but in only 7,000 possessions because it plays slower. Its offensive rating is about 114 and its defensive rating about 109 as well, so its net rating is also plus 5.
The two teams differ by 900 total points scored, which looks like a gigantic gap, yet their per-possession efficiency is identical. On a points-per-game basis, Team A might look like an offensive powerhouse and Team B like a plodding squad. On a per-100 basis, they are equals, and both should win roughly the same number of games. The example is invented to make the arithmetic visible, but the principle is real and shows up every season when fast and slow teams are compared.
Reading a net rating
A rough sense of scale helps. Think of net ratings in bands.
| Net rating | What it usually means |
|---|---|
| Plus 8 or higher | A team in the title conversation |
| Plus 4 to plus 8 | A comfortable playoff team, possibly a contender |
| Plus 1 to plus 4 | A solid team on the playoff bubble or just above |
| Minus 1 to plus 1 | An average team |
| Below minus 4 | A team in a rebuilding or struggling year |
These bands are a guide, not a rule, and the cutoffs shift with context. The more important habit is to ask what the number is built from. A plus 6 produced by a great offense and an average defense is a different team than a plus 6 produced by an elite defense and a middling offense. Pairs of numbers tell you more than one number alone, which is why most sites display offensive rating, defensive rating and net rating together.
The second habit is to ask how much of the season the number covers. Early in the year, a net rating is built on a few dozen games, and a handful of blowouts can drag it far from where it will settle. Our piece on small samples in basketball explains why a number drawn from twenty games deserves more suspicion than one drawn from eighty.
Where net rating can mislead you
Net rating is the best single number for a team, but it is not a perfect one, and it helps to know its failure modes.
Schedule
A team that has played a soft schedule will have an inflated net rating, and a team that has faced a tough one will have a deflated rating. Raw net rating treats every opponent the same. The fix is to adjust for the strength of the teams faced, which is the subject of strength of schedule and how to adjust for it. In the NBA, where everyone plays everyone, the effect is smaller than it is in college basketball, where conferences vary wildly, but it still exists, especially when one conference is stronger than the other or when a team's schedule is front-loaded with home games.
Garbage time
When a game is decided, coaches empty the bench. The final minutes then feature reserves against reserves, and the score drifts in ways that do not reflect the quality of the starters. A team that wins by 25 and sits its stars in the fourth quarter can see its net rating shrink a little; a team that is blown out and then gives up easy points late can see its rating worsen. Some analysts filter out garbage-time possessions to get a cleaner reading, but the filters are judgment calls, and published net ratings usually include everything.
Health and roster changes
A season-long net rating averages across every version of a team. If a star missed thirty games, the net rating blends the team with him and the team without him. That blend describes the past accurately and tells you almost nothing about how the team will perform when everyone is healthy. This is the most common mistake in using net rating for playoff predictions: the number may belong to a roster that no longer exists in the same form, or to a version of the team that will not be the one on the floor in a seventh game.
Style and the playoffs
In the playoffs, rotations tighten. Starters play more minutes, teams run more of their best actions, and opponents scout deliberately for weaknesses. A team that built a fine regular-season net rating with a deep bench can look different when only eight players matter. Conversely, a team with great top-end talent can be better in the playoffs than its regular-season number implies. Net rating remains a strong guide, but it does not carry over perfectly.
Luck on the margins
Opponent three-point shooting is the classic source of noise. Teams have much more control over the shots they take and the shots they allow than over whether the opponent's shots go in. A defense that holds opponents to unusually poor three-point shooting for a stretch may be seeing luck rather than skill, and its defensive rating and net rating will drift back toward normal. Looking at the components, such as the four factors of shooting, turnovers, rebounding and free throws, can show which parts of a rating are durable.
Net rating for lineups and players
The same arithmetic can be applied to any group of players: take the possessions during which they were on the floor, and compute points scored and allowed per 100. A five-man lineup's net rating answers the question, how did these five do together? A player's on-court net rating answers, how did the team do while he played? And the difference between the on-court and off-court figures is the basis of the on/off number discussed in plus-minus, on/off and why the raw number lies.
Here the cautions pile up. Lineups often play only a few hundred possessions in a season, which is a small sample, and the results can swing wildly from the quality of the opposition and from three-point luck. A player's on-court rating is contaminated by his teammates and by who he plays against: a reserve who plays alongside three starters will look better than one who plays with other reserves, and a star who rests while the other team's bench is on the floor will look worse than he deserves. The usual remedy is to use these figures as hints rather than as verdicts, and to combine them with other evidence, which is why lineup data and five-man units are best read with the possession counts in plain view.
At the team level, the sample is large enough that these distortions largely wash out. That is exactly why net rating is a better tool for teams than for individuals.
How to use it well
A short checklist helps turn the idea into practice.
- Look at the pair. Check offensive and defensive ratings separately before you trust the difference.
- Check the sample. A net rating after fifteen games is a sketch; after sixty it is a portrait.
- Adjust for schedule when it matters. If a team has played far easier or tougher opponents than the average, mentally shade the number.
- Compare with the record. A big gap between wins and net rating usually signals luck in close games, and gaps tend to close.
- Ask about the roster. Was the full lineup together? Who is missing from the number?
- Treat it as a guide to quality, not a guarantee. The best team by net rating does not always win the title, but it wins more often than any other single-number approach would predict.
If you want to experiment yourself, how to build a simple team rating at home walks through a simple version that uses the same inputs, and the rankings page is a handy place to compare what different approaches say about the same teams.
What to take away
Net rating works because it measures the thing that decides every game, scoring more than your opponent, on a scale that removes the distortion of tempo. For a full season, it is the strongest single summary of a team, and it forecasts future results better than the win-loss record alone. Its weaknesses are real but manageable: schedule, garbage time, roster changes and luck. Keep those in mind, read the offensive and defensive halves separately, and net rating will serve you well whenever you need to size up a team quickly.