Assist percentage, turnover percentage and the passing numbers
Assists per game and turnovers per game are shaped by minutes and teammates. Learn how assist percentage and turnover percentage work, what the assist-to-turnover ratio hides, and how to read passing numbers.
A great pass is one of the most satisfying things in basketball and one of the hardest to count. The bounce pass threaded between two defenders, the skip across the court to a shooter in the corner, the look-away feed to a cutter: all of them are worth the price of admission, and the box score records each, at best, as a single tick in a column. Even that tick depends on the shooter converting. A perfect pass that the receiver bricks is, statistically, nothing.
Turnovers are the mirror image. They are the cost of handling the ball, and the same player who creates dazzling plays also tends to commit more of them, simply because he handles the ball more. A fair assessment of a passer has to weigh what he creates against what he gives away, and it has to do so in a way that does not simply reward the player who has the ball the most.
This article looks at the numbers used to do that: assist percentage, which measures how much of a team's scoring a player sets up, and turnover percentage, which measures how often a player's possessions end in a giveaway. It explains how each is calculated, why they are fairer than their per-game cousins, how they relate to the old assist-to-turnover ratio, and what the passing numbers cannot capture.
What counts as an assist
Start with the definition, because the whole structure rests on it. An assist is credited to a player whose pass leads directly to a made basket. In general, the receiver must score without a significant dribble or move that effectively creates the shot himself, though the rules give the official scorer discretion about what counts. A pass to a player who takes a dribble and scores may or may not be credited, depending on the judgment of the scorer.
Two features of that definition shape everything that follows.
First, it depends on someone else. A passer cannot get an assist unless the teammate makes the shot. A player who throws the same quality of pass to poor finishers will record fewer assists than one who throws to excellent ones. An assist number partly measures the shooters.
Second, it involves judgment. Scorers in different arenas or leagues have been more or less generous, and the standard has shifted over time. Analysts have long noted that assist totals can differ for the same play from one scorer to another, and that home teams have sometimes seemed to record more assists than visitors. The effect is small compared with the signal, but it is another reason assist counts are best read as approximate. The site's data sources page describes where the figures here come from.
The tracking era has added more detailed counts, including passes that lead to a shot attempt regardless of whether it goes in, often called potential assists, and the pass before the assist, sometimes called a secondary or hockey assist. These help show a passer's true contribution and reduce the dependence on shooters. They are available only for recent seasons and are not part of the standard record.
Why assists per game mislead
The same problem that afflicts all per-game totals applies to assists: the count depends on opportunity. Several factors inflate or deflate it.
- Minutes. More time on the floor means more chances to pass.
- Pace. A fast team has more possessions, more shots and more made baskets to be assisted.
- Team shooting. A team that makes a high percentage of its shots has more baskets to assist. A passer on a team of excellent shooters will rack up more assists than the same passer on a team of poor ones.
- Role. A player who runs the offense handles the ball on nearly every possession and has the most opportunities. A forward who passes occasionally has fewer.
- Teammates who create their own shots. If the team's scorers mostly generate their own baskets in isolation, there are fewer assistable baskets, and every passer's total drops.
This is why comparing assists per game across eras and teams can mislead. John Stockton's season of more than 14 assists per game in 1990-91 is a spectacular total, but it came with heavy minutes, a rhythm of pick-and-roll basketball with Karl Malone and a tempo that gave him plenty of possessions. Stockton remains the all-time leader in assists by a wide margin, with a career total that no one is close to approaching; the article on John Stockton describes his craft. But to compare him with a passer from a different era, you need a rate.
Assist percentage
Assist percentage estimates the share of a team's made field goals that a player assisted while he was on the floor. The formula is:
AST% = 100 × assists ÷ (((player minutes ÷ (team minutes ÷ 5)) × team field goals made) − player field goals made)
In plain terms, the denominator estimates the number of baskets scored by teammates while the player was on the court, since a player cannot assist his own shot. The numerator is the number of assists he recorded. The ratio is the percentage of teammates' baskets that he set up.
A worked example helps. Suppose a guard plays the full length of a game and his team makes 40 field goals, of which he scores 8 himself. His teammates scored 32. If he recorded 12 assists, his assist percentage is 12 divided by 32, or 37.5 percent. He set up more than a third of his teammates' baskets while he was in the game.
What does the scale look like? An average player will be in the range of 10 to 15 percent. Wings who share playmaking duties often run at 15 to 25. Primary ball handlers frequently run at 30 percent or higher, and the best playmakers of an era may reach the 40s. Centers may be in the single digits. These figures vary with the team and the league environment, so compare within the same season.
The advantages are the usual ones for a rate. It corrects for minutes, because the denominator counts only the baskets scored while the player was on the floor. It partially corrects for team shooting and pace, because the denominator moves with the number of baskets. And it makes comparison across positions and eras more meaningful.
What it does not correct for is the role. A point guard who runs every pick-and-roll will have a higher assist percentage than a guard who plays off the ball, whatever their passing ability. And the rate still depends on the shooters: a passer who feeds weak finishers will have a lower figure.
Turnover percentage
A turnover is the loss of the ball without a shot. Turnovers include bad passes, dribbles lost, traveling and other violations, offensive fouls, and so on. A raw turnover count has the usual problem: a player who handles the ball more commits more of them. A point guard who runs the offense might commit three or four a night without being careless, while a center who handles the ball rarely and commits two is arguably more careless.
Turnover percentage addresses this by asking what share of a player's possessions end in a turnover. The formula is:
TOV% = 100 × turnovers ÷ (field-goal attempts + 0.44 × free-throw attempts + turnovers)
The denominator is the number of possessions the player used. It is the same set of events used in the usage rate formula: shots, trips to the line and turnovers. The figure is the number of turnovers per 100 of those plays. A turnover percentage of 15 means that out of every 100 possessions the player used, 15 ended in a giveaway.
Typical values for individual players are in the low-to-mid teens, with careful ball handlers in the single digits to low teens and high-volume creators somewhat higher. The very best ball handlers combine high assist percentages with low turnover percentages, which is rare, because creating plays involves risk.
The same measure is computed for teams, with the same formula applied to the team's totals, giving the share of its possessions that end in turnovers. It is one of the four numbers in the four factors, the framework that identifies shooting, turnovers, rebounding and free throws as the main drivers of winning. A team that cuts its turnover percentage by a couple of points gains a noticeable number of extra shots per game, since each turnover prevented is a possession converted into a chance to score. The value of a possession, as described in the article on points per possession, explains why: a turnover is a trip with zero points, so each one costs a team roughly what an average possession is worth, a little over a point.
The assist-to-turnover ratio and its flaws
For many years, the standard way to evaluate a passer was the assist-to-turnover ratio: assists divided by turnovers. A player with 8 assists and 2 turnovers has a ratio of 4. It is quick, intuitive, and found on many leaderboards.
It has some real virtues. It rewards the player who creates without giving the ball away. It punishes carelessness. It is easy to compute.
Its flaws, however, are significant.
- It ignores volume. A reserve who plays 12 minutes, makes 2 assists and commits 0 turnovers has an undefined or infinite ratio. A reserve who makes 1 assist and commits 1 has a ratio of 1. These small samples are meaningless.
- It rewards conservatism. A player who never takes risks can have a superb ratio and add little to the offense. The most creative passers take risks, and risk produces turnovers. The ratio penalizes the very play that creates the most value.
- It mixes different things. The numerator depends on teammates' shooting, and the denominator depends on the player's own carelessness. Dividing one by the other produces a number that is hard to interpret.
- It ignores usage. A player who uses a large share of his team's possessions will have more turnovers just from handling the ball. The ratio makes no allowance.
The percentage-based approach is cleaner. Assist percentage measures creation. Turnover percentage measures carelessness. The two can be read side by side, which allows a more nuanced picture: a player with a high assist percentage and a moderate turnover percentage may be an excellent creator who accepts some risk; a player with a low assist percentage and a high turnover percentage is likely a liability when handling the ball.
Great passers in the numbers
Looking at passing through rates allows some enlightening observations about the great passers.
Magic Johnson, described in the article on Magic Johnson, was a big point guard whose passing in the open court defined the Showtime Lakers. His assist rate was very high, and the team's fast pace multiplied the chances. Stockton's numbers reflected a different style, a methodical pick-and-roll partnership, but the rates are comparable in their dominance. Steve Nash, discussed in the article on Steve Nash, combined high assist rates with very efficient scoring in his Phoenix years, a rare combination for a primary creator. Jason Kidd, covered in the article on Jason Kidd, was a triple-double threat whose passing was complemented by rebounding and defense.
The rates also reveal a different type of passer: the big man or forward who handles playmaking duties. A center with a high assist rate, such as those who run offense from the high post, can open an offense in ways that the position traditionally did not. The article on playing through a point guard versus a point forward discusses how different teams have organized their playmaking.
What the passing numbers cannot see
Even the better rates leave much out.
The pass before the pass. An offense is full of actions where the player who makes the key decision is not the one who gets the assist. A skip pass that draws a defender, followed by a swing and a shot, credits the last passer. The first made the play. Analysts have used the idea of the secondary assist to capture this, but standard records ignore it.
Passes that create fouls. A pass that leads to a drawn foul and free throws creates points but not an assist.
Passes that create open shots that are missed. The pass was as good as any, and the passer gets nothing.
Passes that avoid turnovers. A safe pass that keeps the possession alive is the foundation of ball movement, and it is not recorded. The article on ball movement and assist numbers argues that a team's passing quality is not fully captured by its assist total, and that some of the best offenses make many passes per possession without a high assist count.
Reading the defense. The most valuable skill of a point guard is the ability to read what a defense is doing and respond before the opening closes. This is the subject of the article on how to read a defense. The decisions are visible on film and invisible in a statistic.
The role of the offense. Systems differ. Some offenses, such as the triangle, spread playmaking among several players, so no one has an extreme assist rate even though the team passes extremely well. Others funnel playmaking through a single player. Rates reflect the system as well as the passer.
Additionally, there is some subjectivity in the very definition of an assist, as noted. Over a full season the effect averages out, but in a single game the difference between an assist and no assist on a given play can be arbitrary.
Turnovers have causes
A turnover percentage, like an assist percentage, is a description rather than an explanation. Turnovers come in several kinds, and a high figure can reflect very different problems.
- Live-ball turnovers, such as steals and bad passes that the defense picks off, are especially costly because they often lead to fast breaks. A team that commits many of these concedes easy points.
- Dead-ball turnovers, such as traveling or stepping out of bounds, are costly in that they end the possession, but they do not give the opponent a running start.
- Offensive fouls end a possession and, in some cases, add to the team's foul count.
- Pressure turnovers reflect the defense's skill as much as the offense's carelessness. A team facing a full-court press will commit more, as explored in the article on breaking the press.
A rate that mixes these cannot say which is which. The standard statistics separate steals from the other turnovers only on the defensive side, not on the offensive.
Reading the numbers in practice
A few habits will make passing numbers more useful.
- For a passer, look at assist percentage and turnover percentage together. High assist percentage with moderate turnovers is a sign of a creator.
- Compare players within the same role and era. A big man's assist rate should be measured against other big men.
- Consider the supporting cast. A passer with excellent finishers will have a higher rate than the same passer with weak ones.
- Consider usage. A high-usage player will commit more turnovers per game; use the percentage to see whether his rate is reasonable.
- Remember what is not recorded: the pass before the pass, the safe pass, the read.
- Treat short stretches with caution, since these rates, like all others, are noisy over a few games.
The leaders page lists assists and turnovers by the per-game measure, and the players section shows more detail. Use them as the starting point, then apply the rates to get a fairer view.
What to take away
Assists per game reward minutes, pace and good shooters. Assist percentage corrects for those by measuring the share of teammates' baskets a player created while on the floor. Turnovers per game reward the player who rarely has the ball; turnover percentage measures how often a player's possessions end in a giveaway. The old assist-to-turnover ratio is quick but rewards caution and ignores volume, and the two percentages are a cleaner pair.
Even so, the numbers see only part of passing. They miss the pass before the pass, the safe pass that keeps an offense moving and the read of the defense that comes before the pass is thrown. Use them to ask better questions: who creates, at what risk, and with whom? Then let the game itself supply the rest.