NBA In-Season Tournament explained: the NBA Cup format

On the last Friday of November the league plays a full slate on courts painted end to end in a single team colour, and a side twenty games from a playoff place can still be knocked out of something. That is the in-season tournament doing its job. Three completed editions on from launch, the competition has a settled shape, a new name, a rearranged December and a prize pot that has quietly become one of the larger single-night payouts in team sport.

The mechanism is short enough to state in a paragraph. All 30 clubs are drawn into six groups of five, three per conference. Each plays four group games, one against every other side in its group, two at home and two away, on scheduled Cup Nights spread across November. Six group winners advance, plus the best second-placed side in each conference, making an eight-team single-elimination bracket in early December. Every game in the competition counts as a normal regular-season fixture except the final, which does not.

Everything else is detail, and the detail is where the interesting decisions live.

• 6Groups, three in each conference
• 4Group games per team
• 8Teams in the knockout bracket
• 67Games in the whole competition

Fixed features of the current format, not results from any edition.

What the tournament is called now, and why the name moved

It arrived for the 2023-24 season under the plain description the league had used internally for years, the NBA In-Season Tournament, with a separate name attached to the trophy itself. Sponsorship changed that within months. From the second edition the sponsor's name sits at the front of the official title, and the shorter form most people actually use, the NBA Cup, has become the working name of the whole competition rather than just the silverware.

The renaming matters more than branding usually does, because it coincided with the league treating the event as permanent furniture rather than an experiment. Year one carried the caution of a pilot: every participating court was repainted in a heavy, saturated design that drew a great deal of comment, some of it about how the surfaces played rather than how they looked. The league brought court production under a single manufacturer for the second edition and gave itself a longer build window. The designs stayed bold. The complaints about consistency mostly stopped.

Three winners have been crowned so far. The Lakers took the first edition with LeBron James named most valuable player of the tournament, the Bucks the second with Giannis Antetokounmpo taking the award, and the Knicks the third with Jalen Brunson. A media panel picks the tournament award and an all-tournament team across all rounds rather than from the final alone.

Why November was the problem the competition was built to solve

To understand the design you have to understand the hole it was dug for. The league had been circling the idea of a mid-season knockout competition for well over a decade before it appeared on a schedule, and the reason it kept coming back is a calendar problem that no amount of marketing had solved.

An 82-game season is long enough that its first two months carry very little information and almost no stakes. A defeat in the second week of November costs a team roughly one two-hundredth of its playoff positioning. Nobody is eliminated, nobody is promoted, and the standings will not resemble their final shape for another three months. Meanwhile the American sporting audience in November is committed elsewhere, to a football season that runs on a weekly rhythm where every result matters and a single defeat can end a campaign.

A league cannot manufacture jeopardy out of nothing inside a schedule where nothing is at stake. What it can do is invent a second competition that runs on top of the same fixtures and imposes a knockout structure on them, so that a group game in the third week of November becomes the last chance to reach a bracket. That is the whole trick. No extra games were created for group play, no team was asked to travel further, and the November calendar was simply relabelled.

The relabelling comes with a cost the league accepted knowingly. Because the competition sits inside the regular season, its rewards have to stay small enough not to distort the season it borrows from, which is the source of most of the criticism the format attracts. A cup that genuinely mattered would need consequences, and consequences in a closed league with guaranteed places would mean redistributing something real.

How the six groups are drawn

The draw is not free-form. Before it happens, each conference's fifteen clubs are sorted into five pots of three by the previous season's record: the top three go into pot one, the fourth to sixth into pot two, and so on down to the bottom three in pot five. One team is then drawn from each pot into each of that conference's three groups.

The consequence is a group of five containing exactly one side from each performance band, which is a deliberate flattening. A group cannot be stacked with three contenders, and no group can be a walkover for a strong side, because every group contains a team that finished in the previous season's top three and a team that finished in its bottom three. Whatever else the competition produces, the entry conditions are close to identical across all six pools.

Note what the pots use. Last season's record, not any current projection, not any market-size weighting. A club that rebuilt aggressively over the summer can find itself in pot five with a roster that belongs in pot one, and the draw will not correct for it. The reverse also happens, and a side that has lost its two best players to free agency can carry a pot-one placing it no longer deserves.

The four group games and the Cup Nights

Each side plays four group games, one against each of the other four teams in its pool, with the home and away split fixed at two each. Those fixtures are placed on designated nights that the league protects for the competition. In the current edition group play runs from the end of October to the end of November, with Fridays carrying the bulk of the schedule and a short midweek cluster around the American Thanksgiving holiday.

Concentrating the games on named nights is the whole point of the exercise. A random Tuesday in mid-November is the hardest sell in the American sports calendar, sitting directly against the professional and college football seasons. Bundling the games onto a handful of branded evenings, each with a national broadcast window attached, converts a diffuse month into a set of appointments. Whether that has repaired November viewership is arguable. That it was the intent is not.

Sixty group games get played in total, which is simply thirty teams playing four each and dividing by two.

Group play tiebreakers, in the order they are applied

Four games is a very short season, and three sides finishing on the same record is common rather than exceptional. The tiebreak order is therefore load-bearing, and it runs like this.

1. Head-to-head record in group play.
2. Point differential across the four group games.
3. Total points scored in group play.
4. The previous season's regular-season record.
5. A random drawing.

Two features of that list are worth sitting with. The first is that point differential arrives at step two rather than near the bottom, which is unusual in North American sport and imported wholesale from European football's group stages. It means a team leading by eighteen with two minutes left has a live reason to keep starters on the floor, which is precisely the behaviour the league wanted to manufacture in November. The second is that overtime scoring is removed before differential and total points are worked out. Without that carve-out a pair of teams could trade baskets through three extra periods and distort a group table by accident.

Total points scored sitting at step three creates a mild second-order incentive that rewards pace. A side that plays fast and concedes as much as it scores can hold a tiebreak edge over a slower team with the same record and the same differential.

1. 1Pot placementEach conference's fifteen clubs are sorted into five pots of three by the previous season's record, so every group ends up with one team from each performance band.
2. 2Group drawOne club is drawn from each pot into each of the three groups in its conference, producing six groups of five across the league.
3. 3Four group gamesEvery club plays each of its four group opponents once, two at home and two away, on the protected Cup Nights spread through November.
4. 4Group tableRecords are separated by head-to-head result, then point differential with overtime removed, then total points, then last season's record, then a drawing.
5. 5Eight qualifiersThe six group winners go through, joined by the second-placed side in each conference with the best group record as a wild card.
6. 6Quarter-finalsSingle elimination in December, hosted by the higher-seeded club in each tie rather than at a neutral venue.
7. 7Semi-finalsAlso single elimination and also at the higher seed's arena in the current edition, one tie per conference.
8. 8ChampionshipOne game at a designated venue between the two conference winners. It settles the trophy and the top prize but does not touch the standings.

The path a single club follows through the current format.

How the eight knockout places are decided

Six of the eight are straightforward: win your group and you are in. The other two are wild cards, one per conference, awarded to the second-placed team with the best record across its four group games. Because the two wild cards are decided conference by conference rather than league-wide, the bracket always ends up with four Eastern and four Western sides, and the two halves of the draw stay separated until the final.

Seeding then follows group performance, and the practical effect of the current arrangement is that hosting rights run almost all the way to the last game. In the present edition both the quarter-finals and the semi-finals are staged at the home arena of the higher-seeded club, with only the championship at a designated venue away from either finalist. That is a change from the first three editions, which sent the semi-finalists to a neutral city for the closing weekend.

The argument for the switch is not complicated. A single-elimination tie in front of a home crowd that has something to lose produces a better atmosphere than the same tie in a half-neutral arena, and it rewards a good group campaign with something tangible rather than symbolic. The argument against is that the league gives up a concentrated finals weekend in one city, which was among the more distinctive things the competition had.

How the bracket is seeded, and why the wild card is worth less than it looks

Once the eight qualifiers are known, they are ranked within their own conference. The group winners take the top three seeds by group record, with the same tiebreak sequence applied across groups rather than inside one, and the wild card takes the fourth seed regardless of how good its record was. That last part is fixed. A wild card that went unbeaten and lost its group only on a head-to-head result still seeds fourth, behind a group winner that went two and two.

The penalty for that is concrete. A fourth seed travels in the quarter-final, and if it wins it travels again in the semi-final, because both rounds are now staged at the higher seed's arena. A group winner that never leaves home has a materially easier path to the final than a wild card of identical strength, and in a single-elimination format staged over eight days the difference between hosting and travelling is not decorative.

That asymmetry is the strongest incentive the group stage contains. Winning a group is not merely qualification, it is two potential home games and a seeding that cannot be taken away by a better record elsewhere. It also explains why the differential tiebreak gets taken seriously in the final round of group play even by teams already certain of qualifying.

The awards, and who decides them

Two individual honours come out of the competition: a most valuable player award and an all-tournament team. Both are decided by a media panel, and both are judged across the whole competition rather than on the final alone, which matters because a bracket this short would otherwise reduce the award to whoever had the best single evening in December.

That choice keeps the group stage inside the conversation. A player who dragged an average roster through four group games and a quarter-final has a case even if his team lost the semi-final, and the voting has generally reflected that. The award has so far gone to the outstanding player on the winning side in each edition, which is what usually happens with panel votes on short competitions, but the rule as written does not require it.

Which games count towards the regular season, and which do not

Sixty-seven games are played across the competition: sixty in group play, four quarter-finals, two semi-finals and one final. Sixty-six of them are ordinary regular-season games in every respect that matters. They go into the standings, into the head-to-head records that decide playoff seeding, into every counting statistic a player accumulates, and into the injury and rest calculus that governs a modern roster.

The final is the exception, and it is a clean one. It does not count towards either team's regular-season record, and the individual numbers produced in it do not enter season statistics. They are recorded separately as championship-game figures. A player who scores forty in that game has scored forty in a competitive fixture watched by millions and zero towards his season scoring average.

That carve-out exists to keep the schedule symmetrical. If the final counted, two teams would finish the regular season having played one more game than the other twenty-eight, which would corrupt every percentage the standings are built on and every tiebreak that depends on them.

The schedule arithmetic, and why only 80 games are announced in August

The neatest piece of engineering in the whole competition is the way it fits inside a fixed 82-game calendar without anybody having to play a 90-game season.

The league publishes only 80 of each club's 82 games when the schedule is released before the season. Two slots per team are held open on the December dates the knockout rounds occupy. Once the bracket is known, those slots are filled in.

Work it through and it closes exactly.

• A team eliminated in group play has both open dates filled with ordinary regular-season fixtures. Four group games plus 78 others, and it reaches 82.
• A quarter-final loser has played one knockout game that counts, so it needs one more fixture added and reaches 82.
• A semi-final loser has played two knockout games that count, both filling open dates, and reaches 82 with nothing added.
• A finalist has played the same two counting knockout games plus a final that counts for nothing. It reaches 82 in the standings having actually played 83.

Nobody plays fewer than 82, nobody has 82 counting games plus extras, and the only asymmetry left in the system is one extra evening's basketball for the two teams that got furthest. That is a small price for a knockout competition bolted onto a league that will not shorten its season.

90%

• Group play60
• Quarter-finals4
• Semi-finals2
• Championship1

Fixed by the format. Sixty-six of the sixty-seven count towards the standings.

Show the numbers

Where the sixty-seven games sit

Item	Value
Group play	60
Quarter-finals	4
Semi-finals	2
Championship	1

What the winning players actually earn

The prize money is real, it is paid to players rather than to clubs, and it is larger than most casual viewers assume.

In the 2025 edition each member of the winning roster took a little over 530,000 dollars. The runners-up received roughly 212,000 each, the two beaten semi-finalists roughly 106,000, and the four beaten quarter-finalists roughly 53,000. The pot for the knockout stage as a whole passed eighteen million dollars.

The structure behind those numbers is more interesting than the numbers themselves, because it is written into the collective bargaining agreement rather than set by announcement. The pool is indexed to the league's basketball-related income, so it rises when league revenue rises. That is why the champion's share has climbed above the half-million-dollar figure of the first edition. Players on two-way contracts receive half the listed amount. Head coaches take the same sum as a player on their roster, and the assistants share a pool set at three quarters of the head coach's figure.

For a maximum-salary player, an extra half-million is a rounding error against a season's earnings, and this is the standing objection to the whole incentive. For a rookie on a scale deal, a two-way player, or a veteran on a minimum contract, it is a meaningful fraction of a year's pay, and those are exactly the players who fill out a December rotation. The money does not motivate the whole roster equally, but it motivates the part of the roster that most often decides whether a team competes hard in a game that is not a playoff game. Contract structure across the league is covered in the guide to how the NBA salary cap works, and the reason a maximum deal dulls the effect is set out in the piece on the maximum contract.

What the league copied, and what it deliberately did not

The design is openly borrowed from European football, where domestic cup competitions and continental group stages have coexisted with league seasons for generations. Group of five, four fixtures, differential as a tiebreak, a knockout bracket, a trophy that has nothing to do with the league title: none of that is an American invention, and the league has never pretended otherwise. Continental club basketball has run something structurally similar for years, as the guide to the EuroLeague format sets out.

What was not copied is more revealing. There is no promotion or relegation attached, no entry for teams below the top level, and no separate calendar. The competition sits entirely inside the existing 82 games rather than beside them, which is why the schedule arithmetic above had to be invented at all. A genuine cup competition in the European sense adds fixtures. This one rearranges them.

The other thing not copied is jeopardy for the club. A European cup exit costs revenue and sometimes a manager. Here it costs two December evenings that get backfilled with ordinary league games, plus a share of a prize pool that never reaches the balance sheet.

What it has changed about November

Three editions is a short evidence base and it would be easy to overclaim. Some effects are visible anyway.

The most obvious is that a handful of November games now carry a defined stake, which changes how they are coached in the closing minutes. A tied group and a live differential tiebreak is a reason to keep a starter on the floor at the end of a comfortable win, and coaches have said as much. The competing pressure, resting players through a long season, has not gone away and is unpacked in the piece on load management, but for four nights a year it now has something pushing back against it.

The second is a scheduling effect that runs beyond the tournament. Holding two fixtures per team in reserve gives the league a rare piece of flexibility in a calendar that is otherwise locked in August, and those late-filled December games can be placed with better information than any August schedule has about which matchups are worth a national window.

The third is the trophy problem. A competition three editions old has no history, and a trophy with no history is hard to make players want. The evidence so far is that the players on the floor take the knockout games seriously, which is the part the format can actually control. Whether the competition means anything in twenty years depends on things no format can legislate for.

The criticisms that have stuck

Three complaints keep returning, and all three are reasonable.

The first is that the reward is not proportionate to the effort. A team can win seven straight competitive games across November and December, take a trophy, and hold exactly the same position it would otherwise hold in the standings, because the games all counted anyway. The competition adds meaning without adding consequence.

The second is fatigue. The two finalists play an extra high-intensity game in December and travel for it, in a season already criticised for its density. It is one game, and the counter-argument is that it replaces nothing, but it lands on the two teams that have just played two knockout games inside a week.

The third is that the format flatters short samples. Four games decide a group, and a single blowout can settle a differential tiebreak that decides who plays in December. That is true of every group stage ever devised, and it is not obvious that it is a flaw rather than the point. The play-in tournament attracted much the same objection when it arrived and has largely outlived it.

What the tournament leaves untouched

A surprising amount of confusion comes from readers assuming the competition alters something in the wider season. It does not, and the list of things it leaves alone is worth stating plainly.

Playoff qualification is unaffected. Winning the trophy grants no seeding, no bye and no advantage of any kind in April. The champion enters the postseason on exactly the record its 82 games produced, and if that record is bad it misses the playoffs entirely, which is a live possibility given that the bracket takes the best group performers rather than the best teams.

The draft order is unaffected. A deep run in December does nothing to lottery odds beyond the ordinary effect of winning games, and the mechanics of that are covered in the explainer on the draft lottery.

The salary cap is unaffected. Prize money is paid outside team salary and does not count against the cap or the aprons, which is the reason a payout of this size could be agreed at all without disturbing the roster-building rules.

Trade eligibility, the deadline, roster limits and two-way conversion rules all run to their normal calendar. The only structural change any team experiences is which two December dates carry a knockout tie and which carry an ordinary fixture.

Where the format could still move

Three editions in, the league has already shown it will adjust the thing rather than defend it. The name changed, the court production changed, the group-play nights were reshuffled, and the semi-finals moved out of a neutral city and into home arenas. None of those were emergencies. They were tuning.

The obvious remaining levers are the size of the bracket, the treatment of the final, and the venue policy. An eight-team knockout out of thirty is a narrow gate, and expanding it would give more clubs a reason to care in late November at the cost of diluting what qualification means. Making the final count towards the standings would remove the awkward statistical carve-out but would break the 82-game symmetry, which is almost certainly why it has not happened. And having moved the semi-finals home, the league has left itself the question of whether a fixed championship venue is worth keeping at all, or whether the last game should also go to the higher seed.

What is unlikely to move is the core arithmetic. Four group games inside the existing schedule, two dates held in reserve, and a final that sits outside the standings is a tight solution to a hard constraint, and there is no obvious way to loosen it without either lengthening the season or shortening it.

How to read a group table in late November

Practical guidance, for the week when the tables actually matter.

Look at differential before you look at records. With four games and five teams, records bunch immediately, and the second tiebreak usually settles things. A team at two wins and one defeat with a differential of plus thirty is in a materially better position than a team with the same record at plus four, and the standings page will not make that obvious.

Check the head-to-head first, though, because it outranks differential. Two teams level on record with a completed meeting between them have already settled it, and no amount of scoring in the final game will change that.

Watch the wild-card race across the whole conference rather than inside a single group. The second wild card is decided between three second-placed teams who never play each other, which means a side can be eliminated by a result in a game it has no connection to.

And watch the closing minutes of a decided game. A coach who leaves starters in a twenty-point win in the last group round is telling you the differential tiebreak is live, which is usually more reliable information than anything said in the pre-game interview.

For the wider basketball calendar, from the draft lottery through the trade deadline to the postseason, the basketball section collects the format guides in one place, and the full article index carries the equivalent explainers for every other sport we cover.