The NBA salary cap explained: a soft cap with hard edges

Every summer a team with no cap space signs somebody for a fortune, and every summer somebody concludes the cap must be decorative. It is a reasonable thing to think and it is wrong, but it is wrong in an interesting way. The NBA salary cap explained honestly is not a story about a ceiling at all. It is a story about a ceiling with a list of named doors cut into it, and about who is allowed through which door.

The doors are not loopholes. They were negotiated, they are written down, and most of them exist to solve one problem the league decided it cared about more than it cared about a clean number: a team that drafts and develops a great player should not lose him purely because its accountants ran out of room.

Everything else follows from that decision. The exceptions, the tax, the aprons, the cap holds, the dead money. All of it is scaffolding built around the choice to let teams keep their own.

Soft cap, hard cap, and why the difference is not a detail

A hard cap is a wall. The NFL runs one, and its clubs must be under the number when the league year opens and stay under it, with no fee available to anyone who would rather not be. There is no exception, no tax, no escape hatch for a rich owner.

The NBA runs the opposite arrangement. Its cap is a threshold, and a team that is over it loses the ability to sign players in the ordinary way. It does not lose the ability to sign players. What it loses is the freedom to sign anyone, replaced by permission to sign particular people under particular conditions.

There is a second difference that shapes every decision a general manager makes. NBA contracts are, with narrow exceptions, fully guaranteed. In football a club can end almost any deal in March and pay only what it has already advanced. In basketball, a signature is money the player will receive whether he plays, sulks, gets hurt or retires. A team cannot cut its way out of a bad contract; it can only move it, stretch it, or wait it out. That single fact makes NBA cap management a different discipline, and it is why the sport's front offices treat a four-year deal for the wrong player as a strategic wound rather than an accounting nuisance.

Put those two together and the shape of the league becomes legible. Teams are allowed to overspend, and everything they sign is real money. The system then spends several hundred pages trying to make overspending expensive enough to matter without making it impossible.

• 3Seasons with one team to earn full Bird rights
• 5Maximum contract length when re-signing your own free agent
• 4Maximum contract length signing with any other team
• 3Maximum salary tiers, set by years of service

Each is fixed by the collective bargaining agreement rather than set annually.

Where the NBA salary cap number comes from

Nobody chooses the cap. It falls out of an equation, and the equation starts with a defined revenue pool called basketball related income.

Basketball related income, always shortened to BRI, is not the same thing as everything the league earns. It is a negotiated category, and the argument over what belongs inside it is at least as consequential as the argument over the percentage split. Nationally sold broadcast money is in. Gate receipts are in, net of defined deductions. So are the sponsorship deals, the naming rights, the concessions, the parking, the playoff gate, the merchandise, and the share of arena revenue that the agreement attributes to basketball rather than to the building's other business. Certain streams are excluded outright. Others are counted at a defined proportion, because the arena also hosts concerts and hockey and conferences and nobody wants to relitigate that every April.

The players receive a share of that pool. The exact percentage is set in the agreement, sits a shade either side of half, and can move within a narrow band depending on how revenue actually performs against forecast. The precise figure is published in the agreement itself and is not worth guessing at, because it moves the entire system when it changes by a point.

From there the arithmetic is mechanical. The league projects BRI for the coming season. It multiplies by the players' percentage. It subtracts projected benefits, which are real costs the players receive that are not salary: pensions, insurance, the various welfare funds. What remains is the total the league's teams are collectively expected to pay in salary. Divide by the number of teams, thirty at the time of writing, and you have the per-team number, and the cap is set at a defined percentage of it. The luxury tax line comes out of the same base figure with a larger percentage applied. The two aprons are then set above the tax line at levels the league calculates and publishes alongside it.

Two features of this deserve more attention than they normally get.

The first is that the cap follows revenue with a lag and a governor. Because it is calculated off projections rather than results, and because the agreement limits how far the number may move from one season to the next, a colossal new media contract does not arrive as a single summer of free-spending chaos. The current agreement caps the year-on-year rise at ten per cent. That sounds like a technicality and is actually one of the most powerful rules in the document, because it smooths the curve. It also means the players' actual share and the cap can drift apart temporarily, which is what the escrow mechanism exists to correct.

The second is that every team gets the same number. The most valuable franchise in the league and the least valuable have an identical cap, an identical tax line, an identical pair of aprons. Financial advantage in basketball is therefore not a matter of having more room. It is a matter of being willing to pay the tax, and of having accumulated the right exceptions at the right time. The wider comparison across leagues makes the point sharply: the NBA is the only major North American competition where the gap between a rich owner and a frugal one shows up in the tools available rather than in the ceiling.

The escrow account, and how the split is trued up

The cap is built on a projection. Projections are wrong. The escrow system is the correction.

Throughout the season, a fixed proportion of every player's salary is withheld and held in escrow. The withholding rate is set in the agreement and currently runs at a tenth of each pay cheque. That money is not the team's and it is not yet the player's. It sits in an account waiting for the audit.

When the season ends, the league measures what BRI actually was. It calculates the players' guaranteed share of the real figure. Then it compares that share against what the players were actually paid, salary plus benefits, across all thirty teams.

If the players were collectively paid more than their share, the difference is taken out of escrow and returned to the teams. Every player contributes proportionally, whether his own team overspent or not. If the players were collectively paid less than their share, the escrow is returned to them in full, and if there is still a gap the league pays the difference directly, distributed among the players.

This is the part that makes the whole structure coherent. The cap is not really a limit on what players receive in aggregate; the split is. The cap is a limit on how the money is distributed between teams, and escrow guarantees that whatever teams do individually, the collective total lands where the agreement says it should. A summer of wild spending across the league does not enrich the players as a group. It moves money between them and triggers a larger escrow clawback.

It follows that when players complain about escrow, they are not complaining about the mechanism. They are complaining about the forecast that produced the gap.

Why the NBA salary cap is soft: the re-signing exceptions do the real work

Now the doors.

A team over the cap cannot simply offer a contract. It has to name the exception it is using, and the exception determines how much it can pay, for how many years, and with what raises. The largest and most important family of these is the set that lets a team keep its own free agents.

Full Bird rights, formally the qualifying veteran free agent exception, are earned by a player who has spent three consecutive seasons with the same team without changing teams as a free agent. Get traded and the clock keeps running, because the rights travel with the contract. Sign elsewhere as a free agent and the clock resets. A team holding full Bird rights may re-sign that player for any amount up to his maximum, entirely over the cap, for up to five years, with the larger of the two permitted annual raise rates.

Early Bird rights come at two consecutive seasons. The exception is narrower: the salary is limited to a defined multiple of the player's previous salary or to a defined multiple of the league average salary, whichever is larger, and the contract has to run at least two seasons so that it cannot be used as a one-summer dodge.

Non-Bird rights apply after a single season and are worth very little. The team may offer a modest uplift on the player's previous salary, or a modest uplift on the minimum, and that is all. A team relying on Non-Bird rights to keep a player it actually wants has already lost the argument.

The genuine advantage here is not only the money. It is the years and the raises. A rival team, however much cap room it has assembled, can offer four years. The incumbent can offer five. The incumbent's permitted annual raise is larger. Over a full contract that difference compounds into a materially bigger total, and it is the most reliable retention tool in American professional sport.

0 to 6 seasons25%

7 to 9 seasons30%

10 or more seasons35%

Set by the agreement as a percentage of the salary cap rather than as a dollar figure, so the tiers move with the cap every season.

Show the numbers

Maximum salary as a share of the cap, by years of service

Item	Value
0 to 6 seasons	25%
7 to 9 seasons	30%
10 or more seasons	35%

Re-signing your own free agent, in the order it actually happens

The most misunderstood thing in NBA free agency is that a team cannot use its cap room and then its Bird rights on the same summer's business without paying for the privilege. The two modes are alternatives, and the choice is forced by the cap hold, which gets its own section further down.

The sequence matters, and it is genuinely a sequence. Each step forecloses options in the ones after it.

1. 1Check the service clock.Three consecutive seasons with the club without changing teams as a free agent gives full Bird rights. Two gives Early Bird. One gives Non-Bird, which is barely worth having.
2. 2Decide whether you are a room team or an exception team.A club operates in one mode or the other. It cannot spend cap room in the ordinary way and then reach for a Bird exception on the same set of signings.
3. 3Leave the cap hold on the books, or renounce it.The hold is a placeholder charge for the unsigned free agent. Keeping it preserves the Bird exception and blocks the room. Renouncing it converts the hold into usable room and destroys the exception permanently for that summer.
4. 4Spend the room first, if you have chosen to use it.Cap room is consumed before exceptions. Whatever exception survives afterwards is the smaller room version rather than the full mid-level.
5. 5Re-sign your own player last.With the hold intact and the other business done, the Bird exception allows the club to go over the cap to finish the deal, for a longer term and larger raises than any rival is permitted to offer.
6. 6Check the apron before the ink dries.If a signing would take the team across a line it has hard-capped itself at, the transaction is simply not permitted, whatever the exception would otherwise allow.

Steps three and four are where most summers are won or lost. Renouncing a hold is irreversible for that offseason.

Front offices run this sequence in a spreadsheet in May and argue about it until July. The interesting decisions are almost never about the headline contract. They are about which holds to keep, which to kill, and whether the room a team could create is worth more than the exception it would have to destroy to create it.

The mid-level, the bi-annual, and the rest of the toolkit

Below the Bird exceptions sit the tools a capped-out team uses to add players it did not previously employ.

The mid-level exception exists in three sizes. The full version, available to teams below the first apron, is the largest and permits the longest contract. The taxpayer version, for teams between the aprons, is smaller and shorter. The room version is for teams that used cap space, and is smaller again, which is the price of having operated as a room team. The exception can be split across several players, and it is the main reason a contender that appears to have no money at all can still add a rotation piece every summer.

The bi-annual exception is a smaller amount usable only in alternating years, and only by teams below the first apron. It is deliberately awkward, and a team that spends it has committed to going without it the following summer.

The minimum salary exception is the one nobody can lose. Any team, at any point, may sign any player to a minimum contract of up to two years, regardless of cap position. The minimum itself is a scale that rises with years of service, and there is a quiet subsidy attached: for a veteran with enough service, the team is charged against the cap only at a lower defined level, with the league paying the difference. The design intent is transparent, and it works. Without it, a capped-out contender would always prefer a rookie to a ten-year veteran for the fifteenth roster spot purely on price, and the league would have quietly abolished the veteran bench player.

A trade exception is created when a team sends out more salary than it takes back. The difference becomes a credit, valid for a year, that can absorb an incoming salary in a later deal without any matching salary going the other way. Trade exceptions cannot be combined with each other or with other exceptions to swallow one larger contract, which limits their usefulness to precisely what they are: a way of doing a trade in two halves separated by months.

Sign-and-trade is the mechanism that lets a player leave for a team without cap room while still collecting something close to what only his own team could pay. He signs a new contract with his current team and is immediately traded. The deal must run at least three years and no more than four, and the acquiring team is hard-capped at the first apron for the rest of the season as a consequence. It is a compromise transaction, and both the length rules and the hard cap exist to stop it becoming the default route around the entire system.

Salary matching governs ordinary trades. A team over the cap must send out salary roughly comparable to what it takes back, with the permitted ratio and the small fixed cushion both set in the agreement, and with the allowance narrowing as a team climbs the spending ladder. This is why apparently absurd contracts get attached to good players in trades. Somebody has to make the numbers meet.

The luxury tax, and why the brackets escalate

Above the cap sits the tax line, produced by the same BRI arithmetic with a higher percentage. Crossing it is permitted and expensive.

The tax is not a flat charge on the excess. It is assessed in bands. The first slice of money above the line is taxed at one rate, the next slice at a higher rate, the slice after that higher again, and the escalation continues for every further band. The structure is deliberately punitive at the top, so that the marginal cost of the last few million dollars of a payroll is a large multiple of the cost of the first few.

The current agreement did something clever with those rates, and it took effect part-way through the deal rather than on day one. From the 2025-26 season the two lowest bands were made cheaper than they had been, while every band above them was made dearer. A team that dips a few million over the line therefore pays less than it once would have, and a team that lives fifty million over pays considerably more. The tax was rebalanced to stop punishing the club that misjudged its February injury cover, and to squeeze harder on the club that never intended to be under the line at all.

The arithmetic is easier to see than to describe. The figures below are invented, with round numbers chosen so the mechanism is visible. The real bands and rates are set in the agreement and republished each season.

Suppose a team finishes twelve million dollars above the tax line, and suppose the bands are five million wide with rates of one and a half, two, and three dollars of tax per dollar of salary. The first five million costs seven and a half million in tax. The second five million costs ten million. The last two million, sitting in the third band, costs six million. Total bill: twenty-three and a half million dollars, on twelve million of salary.

Now add three million more in salary. All of it lands in the third band or above, so that three million costs at least nine million in tax. The team has spent twelve million dollars to add a player earning three, which is the entire point of the design.

Then there is the repeater rate. A team that has paid the tax in a defined majority of recent seasons, set in the agreement as three of the four preceding years, pays an additional increment in every single band. The repeater rate is not a penalty for spending. It is a penalty for spending habitually, and it is aimed squarely at the small group of franchises that would otherwise treat the tax as a permanent line item.

The bill is assessed on the final day of the regular season, based on the payroll on that date. That single sentence explains a large fraction of the trade deadline. A team that has fallen out of contention and sits four million above the line has an enormous incentive to get under it, and will happily attach a draft pick to somebody else's problem in order to do so.

Tax receipts do not vanish into the league office. A defined portion is distributed among the teams that did not pay, and the remainder is retained by the league for its own purposes, including revenue sharing. Every dollar a taxpaying team spends therefore partly funds its competitors, which is the sort of arrangement that survives only because both sides signed it. The same logic drives baseball's competitive balance tax, though that sport has no cap underneath to give the tax a floor to work from.

There is a floor as well as a ceiling. Teams must spend at least ninety per cent of the cap in salary, and a team finishing below that pays the shortfall directly to the players who were on its roster. A permanently thrifty franchise cannot pocket the difference. It can only choose which players to hand it to.

The two aprons, and what each one takes away

The tax charges money. The aprons remove tools, and that turns out to hurt far more.

Both aprons sit above the tax line, at levels set in the agreement and published each season. Crossing them does not trigger a payment. It triggers a set of prohibitions, and the prohibitions are cumulative: everything the first apron takes, the second apron also takes, plus its own list.

1. 1Below the cap.The team may sign anyone with the room it has, in the ordinary way, without naming an exception at all.
2. 2Above the cap, below the tax.Every signing must now come through a named exception. The full mid-level, the bi-annual and all the Bird exceptions remain available.
3. 3Above the tax line.Nothing is forbidden, but every dollar is now taxed in escalating bands, with an extra increment for repeat payers.
4. 4Above the first apron.The full mid-level shrinks to the taxpayer version, which is capped at two seasons and cannot be used to absorb a player by trade or waiver claim. The bi-annual exception disappears. Sign-and-trade acquisitions are barred. A trade can no longer bring back more salary than it sends out. A trade exception generated in a previous season cannot be used. A waived player whose pre-waiver salary was above the full mid-level cannot be signed.
5. 5Above the second apron.Salaries can no longer be combined to match one larger incoming contract. Every mid-level exception is gone, including the taxpayer version. Cash cannot be sent in trades. A player sent out in a sign-and-trade cannot be used to match incoming salary. The team's first-round pick seven drafts into the future is frozen and cannot be traded, and staying above the line in two of the following four seasons pushes that pick to the end of its round.

The restrictions are cumulative. A team above the second apron is also subject to everything the first apron removes.

Read that ladder as a design document and the intent is obvious. The tax was a tax, and a sufficiently rich owner treated it as a cost of doing business. The aprons cannot be paid off. A team above the second apron loses the ability to build in the ordinary way, and the loss compounds, because the tools it forfeits are exactly the ones a team needs in order to fix a roster mistake.

The hard-cap mechanism is what gives the aprons their teeth during the season. Certain transactions bind a team to a ceiling for the remainder of the league year, and the ceiling depends on which transaction it was. Using more than the taxpayer slice of the mid-level, using the bi-annual, acquiring a player by sign-and-trade, signing a bought-out player who earned more than the full mid-level, taking back more salary than it sends out, or using a trade exception left over from a previous season: each of those hard-caps a team at the first apron. A different and shorter list hard-caps at the second apron: using the taxpayer mid-level, combining two or more salaries to match an incoming contract, sending cash in a trade, or using a signed-and-traded player to absorb salary. Once hard-capped, a team may not exceed that figure for any reason at any moment, including for a single day in February when an injury replacement would have been convenient. General managers now run their in-season roster with that ceiling permanently in view, and the second apron in particular has become the number every front office plans backwards from.

The frozen draft pick is the quietest and probably the most severe of the restrictions. Finish a season above the second apron and the club's first-round pick seven drafts away is frozen: it cannot be traded, and it cannot be used to sweeten anything. Stay above the line in at least two of the next four seasons and that pick is moved to the end of the first round regardless of where the team finishes. Duck under in three of those four and the freeze lifts. The cost is not felt for years, which is precisely why it works on decision-makers who are otherwise willing to write any cheque today. A pick that would have landed high in a good draft instead lands at the bottom, and the team that most needed cheap young talent is the one that ends up without it.

Nobody involved pretends this arrangement is settled. The union has said in public that it did not propose the second apron, does not like what the apron has done to player movement, and intends to attack it. What it has not done is change it. The agreement signed in 2023 runs through the 2029-30 season with a mutual opt-out available after 2028-29, and until either side exercises that right or the two of them agree a mid-term amendment, every rule described here is the rule. Treat proposals to soften the apron as what they are: negotiating positions, not regulations. The distinction matters, because a great deal of comment written about the NBA cap describes a system somebody would like rather than the one the clubs are actually operating under.

Rookie scale contracts, the max tiers, and the supermax

The entry point to all of this is the draft, and drafted players do not negotiate their first contract in any meaningful sense.

Each first-round slot carries a scale amount. The team may pay between eighty and one hundred and twenty per cent of it, and in practice pays a hundred and twenty per cent every single time, because refusing achieves nothing except annoying the player. The contract runs two guaranteed seasons with two further team options, each exercised a year ahead of time. Second-round picks sit outside the scale, with a separate exception that gives them a defined route to a standard deal.

At the end of a rookie scale contract, the player becomes a restricted free agent provided his team has extended a qualifying offer. Restricted free agency gives the incumbent a right of first refusal: another team may sign the player to an offer sheet, and the original team has a short window set in the agreement to match it exactly and keep him. Offer sheets are consequently rare and strange, because the only ones worth making are the ones structured to be painful to match. How that interacts with the rest of the summer is a subject in itself, and it drives a great deal of what happens once the free agency window opens.

Maximum salaries are expressed as a percentage of the cap rather than as dollars, in three tiers by years of service, as charted above. A player with fewer than seven seasons is limited to the lowest tier, seven to nine seasons reaches the middle, and ten or more the highest. There is also a floor: a max contract cannot pay a player less than a defined uplift on his previous salary, which prevents the tier system from producing a pay cut for someone who has just earned a raise.

Two mechanisms let a player reach a tier early, and both are decided by award voting rather than by negotiation.

A player finishing a rookie scale contract can reach the middle tier by meeting objective criteria written into the agreement: selection to an All-NBA team, the Most Valuable Player award, or Defensive Player of the Year, each within a defined recent window. His team designates him, the criteria are checked against the actual voting, and the salary adjusts on that basis alone.

The designated veteran extension, universally called the supermax, is the same idea at the top tier. A player with seven or eight years of service can be paid at the highest percentage before he is otherwise eligible, but only by the team that drafted him or that acquired him while he was still on his rookie contract, and only if he meets the same category of award criteria: an All-NBA selection or Defensive Player of the Year in the season just finished or in two of the last three, or the Most Valuable Player award in any of the last three. A team may carry only a limited number of these at once.

Since the current agreement, there is a gate in front of all of it, and it has become one of the most consequential rules in the document. A player is not eligible for MVP, Defensive Player of the Year, an All-NBA team or an All-Defensive team at all unless he has appeared in at least sixty-five regular season games. An appearance only counts if he played at least twenty minutes, with an allowance for two games of at least fifteen minutes. There is a narrow exception for a player who reaches sixty-two qualifying games, has featured in the large majority of his team's games to that point, and then suffers a season-ending injury, and a separate challenge process for genuinely extraordinary circumstances.

Read that alongside the max tiers and the effect is obvious. Availability is now worth a percentage of the salary cap. A player who misses eighteen games in the wrong season is not merely absent from an awards ballot; he is ineligible for the criteria that would have unlocked a higher tier, which can cost him tens of millions of dollars across a five-year contract that he cannot get back the following year. The rule was written to stop load management. What it has actually produced is a category of injury argument that is really a contract argument, conducted by medical staff who know exactly what the sixty-fifth game is worth.

The supermax is the most revealing rule in the whole agreement, because it is the one that most obviously fails on its own terms. It was designed to help small-market teams keep superstars. What it actually does is present a player with a choice between the largest contract available anywhere and the freedom to go somewhere better, while presenting his team with a contract it cannot trade for a period afterwards and cannot get out of at all. A supermax signed by a player who then declines is close to the worst asset in professional sport: enormous, guaranteed, immovable, and occupying the third of the cap that would otherwise buy a supporting cast. The rule has produced more trade requests than it has prevented.

Cap holds: the money a team owes to nobody

A cap hold is a charge against the cap for a player who is not under contract. It is the accounting device that makes the whole free agency sequence work, and it is where most public analysis goes wrong.

When a player's contract expires, his salary comes off the books and a hold goes on. The hold is a multiple of his previous salary, tiered so that a player who was on a small salary carries a proportionally larger hold, and a player eligible for the maximum carries a hold at his maximum. It stays there until one of two things happens. He signs a new contract, at which point the hold is replaced by the actual salary. Or the team renounces him, at which point the hold vanishes and so does every exception attached to him.

The purpose is to stop a team having it both ways. Without holds, a team could spend all of its cap room on outside free agents in the morning and then use Bird rights to re-sign its own players in the afternoon, which would make the cap meaningless inside a single day. The hold forces the choice: keep the hold and keep the exception, or kill the hold and get the room.

Because holds are frequently larger than the contract the player will actually sign, a team's reported cap space in June is often nothing like its real capacity. A club showing no room at all may be carrying thirty million dollars of holds it fully intends to renounce. A club reported to have vast space may be counting on renouncing three players it has every intention of re-signing, and will discover in July that it cannot do both.

Two smaller holds complete the picture. A team with fewer than twelve players under contract carries an incomplete roster charge for each empty spot, valued at the rookie minimum, which stops anyone creating artificial room by simply having nobody on the roster. And an unsigned first-round pick carries a hold at the top of his permitted scale range from the moment he is drafted, so his cap cost arrives on draft night rather than whenever the paperwork is filed. A team holding a high pick therefore starts its summer with less room than the raw contract numbers suggest.

Renunciation is the irreversible move in the sequence. Once a team renounces a free agent, it cannot restore his Bird rights that summer. It can re-sign him, but only with room or with an ordinary exception, on the same terms as any other bidder in the market.

Waivers, dead money, and the stretch provision

Guaranteed money is the reason all of this is difficult.

When a team waives a player, his contract goes on to the waiver wire for a claim window of forty-eight hours. Any team may claim him, and a successful claim means the claiming team assumes the contract in full, salary and cap charge alike. Claims are rare for exactly that reason. Teams claim contracts worth having, and contracts worth having are rarely waived.

If nobody claims him, he clears waivers and becomes a free agent. His original team still owes every protected dollar and still carries it against the cap. What the team gets back is set-off: if the player signs elsewhere, the original obligation is reduced by a defined portion of his new salary, calculated as roughly half of the amount by which the new deal exceeds a defined minimum. Set-off is real money but it is rarely transformative, and it does nothing at all if the player does not sign anywhere.

The stretch provision is the only tool that changes the shape of the charge. A team waiving a player may elect to spread the remaining protected money across twice the number of remaining seasons plus one. The total does not change. The annual charge falls, and the tail lengthens.

• Charged normally
• Charged if stretched

Year 17m3m

Year 27m3m

Year 37m3m

Year 40m3m

Year 50m3m

Year 60m3m

Year 70m3m

Invented figures. Twenty-one million dollars of protected money remaining across three seasons, stretched over twice the remaining years plus one, which is seven. The total paid is identical; only the annual cap charge changes.

Show the numbers

Worked example: stretching a waived contract with three years left

Item	Charged normally	Charged if stretched
Year 1	7m	3m
Year 2	7m	3m
Year 3	7m	3m
Year 4	0m	3m
Year 5	0m	3m
Year 6	0m	3m
Year 7	0m	3m

Look at the two bars and the trade is plain. Stretching buys four million dollars of annual relief in each of the next three seasons, in exchange for three million a year of dead money in four seasons where the team currently has none. If those later years are the ones the team expects to be good, it has just mortgaged the wrong seasons.

The agreement limits the damage in two ways. Stretched dead money is capped as a share of the cap in any single season, so a team cannot stretch its way into a roster made entirely of ghosts. And the decision has to be made at the point of waiving. A team cannot waive a player, watch the cap sheet for a year, and then decide it would rather have stretched him.

There is a date attached, and the current agreement tightened it. To stretch the salary owed for the season about to be played, the player has to have cleared waivers by the end of August, not merely been placed on them. Requesting waivers on the last day no longer works, because the clearance falls the wrong side of the line. That is why a run of otherwise inexplicable late-August releases happens every year, and why a front office that has decided a contract is dead moves in the third week of the month rather than the fourth.

A buyout is the negotiated version. The player agrees to give back a portion of his guaranteed money in exchange for being released to sign elsewhere, usually with a contender. The team's dead money falls by whatever he surrenders. The first apron restriction bites here: a team above that line cannot sign a bought-out player whose pre-waiver salary was above the full mid-level, a rule written specifically to stop the richest rosters hoovering up every good veteran who talks his way out of a bad team in February.

Dead money in basketball is therefore a different animal from dead money in football. In the NFL it is an accounting tail on cash already paid out. In the NBA it is frequently cash still being paid, week by week, to a player currently employed by somebody else. That difference explains why guaranteed contracts change the whole calculation of what a signing actually risks, and why an NBA front office will trade a first-round pick simply to make a contract somebody else's problem.

Reading an NBA salary cap sheet the way a front office does

None of the above is planned one rule at a time. It is planned as a single multi-year sheet, and the sheet has a particular shape that is worth learning to read.

The first column is not this season. This season is largely settled by October. The roster exists, the tax bill is arithmetic, and the only live questions are injury replacements and the deadline. The interesting columns are two and three years out, because that is where the options, the extensions and the frozen picks live.

What a capable front office actually tracks, in rough order of how much it worries about each:

The distance to each line, this season and next. Not the cap, which most contenders abandoned years ago, but the tax line and both aprons. The question is never "are we over" but "how far, and which tools does that cost us in February".

Which summer the sheet resets. Every roster has a season in which a cluster of contracts expires at once. That summer is the team's real chance to change direction, and everything before it either builds towards it or wastes it. Extensions signed today move that date, which is why a routine-looking extension is often a strategic decision disguised as housekeeping.

Cap holds and the renounce list. Long before free agency opens, the team has decided which holds it will keep and which it will kill, and it has modelled the summer both ways. The version with room and the version with exceptions are two entirely different plans, and committing to one closes the other for good.

Which contracts are movable. A movable contract is one whose salary roughly matches its value: large enough to be useful in matching, small enough that somebody wants the player attached to it. A roster full of unmovable deals, whether too big or too small, cannot make a trade even when everyone in the building agrees it should.

The frozen pick. For any team near the second apron, a first-round selection several years out is either safe or exposed, and the decision that exposes it is usually taken in a summer when the consequence feels abstract and the alternative feels like giving up.

The repeater clock. A team that has paid the tax in two of the last three seasons is one season away from a materially higher rate on every dollar. That fact alone has closed more competitive windows than any single trade.

Those six items are all public and all stable, and they will tell a reader more about a team's next three seasons than any amount of July excitement. They read the same way whether the subject is a genuine contender or a team pretending to be one, and the same discipline transfers with only modest changes to the way the women's game handles its own cap and to most of the rest of how modern basketball is built and paid for.

The mistake worth avoiding is treating the cap as a scoreboard. It is not a measure of ambition and it is not a measure of health. It is a description of which doors a team may still walk through. The teams that get into trouble are almost never the ones that spent too much in a single summer. They are the ones that spent enough, for long enough, that every door had quietly been shut behind them.