Play contexts

Not all offensive plays are equal. On average, a layup off a steal before the defense can get back has a much higher likelihood of producing points than a long two with two seconds left on the shot clock. Both plays are recorded as a single field goal attempt in the box score but over the course of a game (and indeed a full season) the types of plays which a team prioritises can provide valuable insights into their offensive strategy. This is where play contexts fit in.

Play contexts split a team's offense by the situation each shot or turnover happens in and we can learn two very useful metrics for every context: how often a team finds itself there (frequency), and how efficiently it scores when it does (points per 100 plays).

We sort offense into four mutually exclusive primary contexts, which together account for all of a team's plays:

• Transition
• Early offense
• Putbacks
• Half court

We then split half court plays one level deeper according to how much shot clock was left, into mid clock and late clock (4 seconds or less remaining).

One possession can have multiple plays

Play contexts are measured at the level of a "play" rather than a possession. The distinction matters when a team rebounds its own miss. For example, a single possession that contains a missed shot, an offensive rebound, and a putback is regarded as one possession but counts as two discrete plays. Each play is characterised by either a shot attempt, a trip to the free throw line, or a turnover.

Transition

A play is transition when the defense has not had time to get set after the offense gains the ball. For shots and free throws we take the league at its word: the NBA and WNBA tag fast break plays in their own play-by-play data. For turnovers, which the league never tags, we use a time-based rule instead: a turnover is transition when it is the first play of a possession, the team won the ball "live" (a defensive rebound or a steal), and it happens within six seconds of gaining possession. Either way the defining feature is the same. The offense is attacking a less organised defense, which is why transition plays are consistently the most efficient context in the game.

One important nuance applies to the time-based component. Those six seconds have to be an unbroken live sequence. If play stops first (a timeout, or a defensive foul that sends the ball to a sideline inbound) the defense gets a chance to get set during the stoppage, so we do not treat what follows as transition or early offense, even when it still falls inside the six second window.

Early offense

Early offense differs from transition in one crucial aspect: it occurs when a team gains possession from a dead ball rather than a live one. The most common case is pushing the ball up the floor off a made basket. We classify it using the same six second window we use for transition turnovers, but the situation is different. After a made basket the defense has a moment to get back and organise, so even a quick attack is working against a defense that is largely set. Tellingly, the league itself doesn't tag these quick pushes as fast breaks.

For this reason we bucket early offense plays separately from transition plays rather than lumping all fast-starting possessions together. Both are valuable (catching a defense before it is fully organised is an edge worth having), but the live-ball break and the push off a make are different challenges. Statistically, transition plays are more efficient than early offense plays so collapsing them into a single "fast" bucket would hide that difference.

Putbacks

A putback is a second-chance play (a play that follows an offensive rebound) where a shot or turnover comes within three seconds of the offensive rebound. It is the immediate tip, gather-and-finish, or quick kick out for a three pointer. If the play exceeds three seconds, this usually means a team has pulled the ball back out, and is running a fresh action.

The three second window follows the standard set by Ben Falk of Cleaning the Glass, whose putback definition is the common reference point for this category. Keeping the window tight is deliberate: it isolates the genuine immediate second-chance attempt, which is typically a high-efficiency play, from a reset possession that happens to begin with an offensive rebound but otherwise looks like ordinary half court offense. Putbacks are carved out of the half court bucket so the two are not double counted.

Half court

Half court plays represent all other play types: any play that is not transition, early offense, or a putback. In practice this is the offense walking the ball up and running its offense against a set defense. For most teams transition makes up around two thirds of all plays and is the single largest context by a wide margin. So a team's half court efficiency is usually the single best indicator of how good its offense really is.

Mid clock and late clock

Within half court plays, we further split plays by how much shot clock was left when the play ended. Late clock is a half court play with four seconds or less remaining; mid clock is everything else in the half court (any half court play with more than four seconds left on the shot clock). The two together make up the half court bucket exactly, so they are best read as a sub-split of half court rather than as peers of transition or early offense.

We use four seconds because offensive efficiency declines steadily as the shot clock runs down and then falls off a cliff at the very end. The final few seconds are typically where an offense is simply trying to get something up before incurring a shot clock violation. Four seconds captures that "have to shoot now" zone. A wider cutoff would dilute it with normal late-clock execution and a tighter one would miss too much of the drop-off. Four seconds sits at the point where efficiency noticeably declines.

NBA play-by-play data does not include shot clock data, so we estimate it from the game clock and the type of play: a fresh twenty four seconds for a first attempt, or fourteen seconds after an offensive rebound. We then subtract the time elapsed in the play, capped by the time left in the period for end-of-quarter situations.

Why the league's fast break tag (and six seconds for turnovers)

The NBA and WNBA tag fast break plays in their own play-by-play data but they only do so on field goals and free throws, never on turnovers. For shots and free throws the fast break tag is the cleanest signal there is: it is the league's own judgement of what was a fast break so we use it directly rather than second-guessing it with a stopwatch. Our transition shot and free throw numbers therefore match the league's own numbers.

A turnover on the break is every bit a transition play though and the tag never covers turnovers. So for turnovers we fall back to a time-based rule: a live-ball turnover (off a steal or a defensive rebound) within six seconds of gaining possession. This is the one place we still lean on a clock window, and it lets transition plays include the turnovers that happen while a team is pushing the ball, not just the shots. We acknowledge that this is imperfect but it is a necessary compromise in the absence of a more direct signal.

We settled on six seconds by calibrating against the league's own fast break shots since this is the one set of plays where we have a definitive label to check ourselves against. For each candidate window we measured two things: how many of the league's tagged fast breaks the window catches, and how many shots inside the window the league did not tag. Six seconds is the balance between catching genuine fast breaks and not over-counting and it lands in the same place for both leagues.

Frequency and efficiency

Each context is measured two ways. Frequency is the share of a team's plays that fall into that context, which describes a team's style: how often it runs, how reliant it is on second chances, and how often it gets stuck in late clock situations.

The raw play counts from the four primary play contexts (transition, early offense, putbacks, and half court) add up to a team's total plays, which is roughly its field goal attempts, plus turnovers, plus trips to the free throw line (a little under half of all free throw attempts, since most trips are two shots).

Efficiency is points per 100 plays in that context. Crucially the denominator is plays, not shots, so turnovers (which score zero) are included. That makes it a true measure of how productive a team is in a context, not just how well it shoots there. A team that scores efficiently in transition but coughs the ball up on the break will see that reflected in its transition number.

Offense and defense

Every context can be viewed from both sides of the ball. The offense view is a team's own play mix and efficiency. The defense view flips it to a team's opponents: how often opponents reach each context against this team, and how efficiently they score there. It is the same six contexts aggregated for the other team in each game.

A team that allows a high opponent transition frequency and efficiency has poor transition defense, and opponents tend to lean into it. A lower efficiency allowed is usually indicative of a good defense but frequency is more nuanced. Allowing a high frequency of late clock plays is usually a good thing since the defense is forcing opponents to take tough shots at the end of the clock. However, if this is paired with a high putback frequency this would tend to indicate a poor defensive rebounding team. Generally speaking a good defensive team will try to increase the frequency of their opponents half court plays since their defense is set in these contexts.

Garbage time

By default, play context numbers exclude garbage time, so blowout minutes against deep benches do not distort a team's profile. This matches how we treat efficiency ratings. A toggle in the site header switches to raw numbers which include every play.

See it on the leaderboards

• Team play contexts (frequency + efficiency by context)

---

Related reading: Efficiency ratings and Possessions.