How does NBA free agency work? Rights, not an open market

Every summer somebody calls NBA free agency a market, and every summer the description falls apart by the second week of July. A market has buyers who compete on price. This one has a queue, and the team the player is already on stands at the front of it holding a document that says so. Asking how does NBA free agency work is really asking who holds which right, what that right permits, and on what date it expires.

There are only a handful of rights in play. The right to negotiate. The right to refuse. The right to exceed the cap for one particular person. The right to convert an unsigned player into spending money by giving up on him. Almost every transaction you will read about in July is one of those four being exercised, and the reporting almost never says which.

The rest of this piece names them.

Why NBA free agency is a set of rights, not an open market

Start with the thing that makes basketball different from every other American league. A team is allowed to exceed the salary cap, but only by naming a written exception, and the largest exceptions exist purely so that a team can keep the players it already employs. The cap is a threshold with named doors cut into it, and the soft cap architecture underneath all of this determines who may walk through which one.

That design decision has a consequence nobody planned but everyone lives with. Since the incumbent can spend money rivals cannot, the outside market for a good player is not a bidding contest. It is a test of whether the incumbent wants him. If the answer is yes, the incumbent wins, because it can offer an extra season and larger raises. If the answer is no, the player reaches the market carrying a signal that the people who know him best have declined to pay.

So the useful mental model is not thirty buyers and a supply of players. It is a small number of teams with genuine cap room, a much larger number of teams holding exceptions of fixed size, and a set of players each attached to one team's prior claim. Money moves along the edges of that structure, never freely across it.

• 2Minimum seasons on an offer sheet
• 2Days the incumbent has to match one
• 3Two-way contracts allowed per team
• 15Standard roster spots to fill

These are structural counts written into the collective bargaining agreement rather than annual figures the league publishes each summer.

Restricted or unrestricted: the distinction that decides everything

A player whose contract has expired falls into one of two categories, and the category is decided by his service time and by whether his team filed a piece of paperwork on time.

An unrestricted free agent may sign with anybody who can pay him. His old team has no special power beyond the exceptions it holds, which is not nothing, but it cannot stop him leaving. Veterans with enough seasons behind them reach this status automatically when their contracts lapse.

A restricted free agent may also negotiate with anybody, may sign an offer sheet with anybody, and may still end up staying exactly where he was against his own preference. His old team holds a right of first refusal. Restricted status attaches to players with relatively little service time, and to players finishing a rookie scale contract, and it exists only if the team tendered a qualifying offer before the deadline.

That last clause carries more weight than its length suggests. Restricted free agency is not a status a player has. It is a status a team buys, by making an offer it may not want to honour, in order to keep a veto it may never use. A team that misses the tender deadline, or decides the qualifying offer is priced above what the player is worth, converts its own restricted free agent into an unrestricted one and loses the right of first refusal permanently for that summer.

There is a third category worth naming because it confuses people every year. A player under contract is not a free agent at all, whatever his agent is telling reporters, and a player with a team option or a player option is not a free agent until that option is resolved. Options are decided on dates set in the agreement, well before the negotiating window opens, which is why late June produces a wave of announcements that look like transactions and are really just declined choices.

What a qualifying offer really does

A qualifying offer is a standing one-year contract offer, and its value is calculated rather than negotiated.

For a player finishing a rookie scale contract, the amount is derived from his draft slot. That would be tidy except for one adjustment that front offices spend real energy on. The agreement contains a set of starter criteria, based on starting roughly half the regular season or playing a defined total of minutes, measured over the previous season or averaged across the previous two. A player drafted late who meets the criteria has his qualifying offer re-slotted upwards, to the amount attached to an earlier pick. A player drafted very high who fails to meet them has his re-slotted downwards. The result is that a team can spend the final weeks of a season quietly aware that giving a young player forty minutes on a Tuesday in April changes what it will owe him in July.

The offer then sits there until a deadline in the autumn. The player may accept at any point up to that date, and several things happen if he does. He plays one season at the calculated figure. He acquires a no-trade clause for that season, because the agreement will not let a team offer a one-year deal and then ship the player somewhere he did not choose. And at the end of it he becomes an unrestricted free agent with no strings at all.

That is a genuine option and it is priced like one. Accepting means betting on your own health and production with the downside entirely uninsured. A serious injury during a qualifying offer season removes both the contract and the leverage. Declining means holding out for a multi-year deal that the incumbent controls and that rivals are reluctant to offer.

The team has an exit too. It may withdraw the qualifying offer unilaterally up to an early-July date, and after that only with the player's agreement. Withdrawal releases the player into unrestricted free agency and cannot be undone.

The right of first refusal, and the two days that decide a summer

Now the mechanism that makes restricted free agency what it is.

A rival team that wants a restricted free agent cannot simply sign him. It signs him to an offer sheet, which is a real contract with a suspensive condition attached. The offer sheet must run at least two seasons. The rival must have the room or the exception to pay it at the moment of signing, and must keep that capacity available while the clock runs, because an offer sheet is not a promise of money, it is money set aside.

The offer sheet is then delivered to the incumbent, and a short window opens. The incumbent has two days from delivery of the notice to exercise its right of first refusal by matching. Matching does not mean bidding higher, and it does not mean negotiating. It means accepting the offer sheet exactly as written: same salary, same years, same bonuses, same options, same everything. The player is then under contract to the team he was trying to leave, on terms drafted by the team he was trying to join.

1. 1The incumbent tenders a qualifying offer.Filed by the deadline in late June. Without it the player is unrestricted and everything below this line never happens.
2. 2The player becomes a restricted free agent.He may negotiate with all thirty teams once the window opens, and may re-sign with his own team at any point instead.
3. 3A rival signs him to an offer sheet.Minimum two seasons. The rival must hold cap room or an exception large enough to cover the first year, and that capacity is frozen from this moment.
4. 4The offer sheet is delivered to the incumbent.The two-day matching window starts here. Until it closes, the rival cannot spend the money it has committed on anybody else.
5. 5The incumbent decides.Matching is all or nothing and cannot be varied. There is no counter-offer, no partial match, and no negotiation over the terms.
6. 6If matched, the player stays.He is bound to the incumbent on the rival's terms, and the agreement restricts trading him for a period afterwards, including a bar on trading him to the team that wrote the offer.
7. 7If not matched, the player leaves.The contract takes effect with the rival, and the incumbent receives nothing at all in return.

The clock runs from delivery of the offer sheet to the incumbent, not from the moment it is reported. A player may hold only one offer sheet at a time.

Two details in that sequence produce most of the interesting behaviour. The first is that a losing incumbent receives nothing. No pick, no compensation, no trade exception. It had a player and now does not. The second is that a matched player cannot be moved on freely afterwards, which stops a team matching an offer sheet purely in order to flip the contract to a third party the following week.

Why a restricted free agent's market is chilled before it opens

Put yourself on the other side. You run a team with genuine cap room, the rarest asset in the sport, and you like a restricted free agent. Consider what signing him to an offer sheet costs before you know whether you have him.

Your room is frozen for the duration of the window. Every unrestricted free agent you might otherwise have signed is being signed by somebody else while you wait. If the incumbent matches, you emerge two days later with the same money you started with, minus every player who has agreed to go elsewhere in the meantime. You paid full price for the option and received nothing.

Worse, the outcome is not random. The incumbent will match every offer sheet that represents good value and decline every one that does not, because it has better information about the player than you do and it chooses after seeing your bid. That is adverse selection in its purest form. The offers that succeed are, on average, the offers the people who know the player best considered too expensive.

Rational teams respond in one of two ways. Most simply do not bid, which is why restricted free agency in practice produces so few offer sheets and so many quiet re-signings at the incumbent's preferred price. The minority who do bid stop trying to find a fair number and start trying to construct an offer that is painful to match.

That is what a poison pill offer sheet is, and the agreement permits a specific version of it for players with only one or two seasons of service. A rival that is over the cap may offer such a player a first-year salary limited to the mid-level exception, then a modest raise in year two, then an enormous jump in year three to whatever it could have paid had it spent all of its room in year one. The rival is charged the average of the whole contract. The incumbent that matches carries the actual year-by-year amounts.

The arithmetic below is invented, with round numbers chosen so the shape is visible.

• Charge to the team writing the offer
• Charge to the team that matches

Year 124m12m

Year 224m12.6m

Year 324m35.4m

Year 424m36m

Invented figures. A four-year offer sheet totalling ninety-six million dollars, structured with a large third-year jump. The team writing the offer is charged the twenty-four million dollar average each season; the team that matches carries the salary as written.

Show the numbers

Worked example: an uneven offer sheet, charged two different ways

Item	Charge to the team writing the offer	Charge to the team that matches
Year 1	24m	12m
Year 2	24m	12.6m
Year 3	24m	35.4m
Year 4	24m	36m

Read the two bars against each other and the tactic is plain. The team writing the offer sheet has a flat, manageable commitment. The team matching it has two cheap years followed by a spike that lands in whichever season it had planned to be expensive elsewhere. If year three is the season a max extension begins, or the season the roster is meant to peak and is already brushing a spending line, the incumbent has to choose between losing a player for nothing and accepting a payroll shape it would never have designed.

Most incumbents match anyway. That is the point of the right, and it is why front offices treat their own restricted free agents as retained assets rather than as open questions.

Bird rights, and why your own team can always outbid the field

Underneath restricted free agency sits the exception family that makes the incumbent powerful even when the player is unrestricted.

Full Bird rights belong to a team that has employed a player for three consecutive seasons without him changing teams as a free agent. The clock survives a trade, because the rights move with the contract, and it survives almost everything except two events: signing elsewhere as a free agent, and clearing waivers. A player who is waived and goes unclaimed takes his own Bird rights out of the building with him, which is one reason teams think twice before waiving somebody they might want back in October.

A team holding full Bird rights may re-sign that player over the cap, at any salary up to his maximum, for up to five seasons, with the larger of the two permitted raise rates. A rival, however much room it has assembled, may offer four seasons with the smaller raise. That gap is the whole ball game. It is not a marginal edge. Across the life of a contract the incumbent can offer more total money and more guaranteed years while paying a lower average annual salary, which is the rare arrangement that is better for both sides at once.

Early Bird rights arrive after two consecutive seasons and are a narrower tool. The salary is limited to a defined multiple of the player's previous salary or to the league average, whichever is larger, and the contract carries a minimum length so it cannot be used as a one-year convenience. The Early Bird exception is what a team uses on a player it developed quickly, and it frequently produces a contract a year longer than either party wanted, because the alternative was nothing.

Non-Bird rights exist after a single season and permit only a small uplift on the player's previous salary. They are the exception a team uses to keep somebody nobody else particularly wants.

Because the tiers depend on continuous service, the rights are effectively tradeable. A team acquiring a player at the deadline acquires his accumulated Bird clock, which is why an expiring contract belonging to a good player is worth more than the same salary attached to a worse one. The acquiring team is buying two months of basketball and an exclusive right to spend over the cap in July.

The ceiling on all of this is the maximum salary, expressed as a percentage of the cap and tiered by years of service. The rules that govern a maximum contract cap how much the Bird exception can actually deliver, which means the incumbent's advantage shows up in years and raises rather than in the headline number whenever a genuinely elite player is involved.

Cap holds, and what renouncing a player actually costs

Here is where public reporting is least reliable, because the number everyone quotes in June is almost never the number a team can spend.

An unsigned free agent does not vanish from his old team's cap sheet. He is replaced by a cap hold, a placeholder charge calculated as a multiple of his previous salary. The multiple is tiered so that a player who was underpaid carries a proportionally larger hold, and a player eligible for the maximum carries a hold at his maximum whatever he was earning before. The hold sits on the books until he signs, at which point his real salary replaces it, or until the team renounces him, at which point the hold disappears and every Bird right attached to him disappears with it.

That is the trade the system is built to force. Keep the hold and you keep the exception, but the hold blocks the room. Kill the hold and you get the room, but you have destroyed your ability to exceed the cap for that player and cannot restore it that summer. You may still re-sign him, on exactly the terms available to any rival.

Two smaller holds complete the picture. A team with fewer than twelve players under contract carries an incomplete roster charge for each empty slot below that line, valued at the rookie minimum, which prevents anyone manufacturing space by simply employing nobody. And an unsigned first-round pick carries a hold from draft night at his scale amount, so his cost arrives before his signature does.

The arithmetic is worth doing once, with invented numbers.

Suppose a team has nine players under contract, and the gap between that committed salary and the cap is thirty million dollars. It has two of its own free agents unsigned: a good one carrying an eighteen million dollar hold, and a role player carrying five million. Nine players means three incomplete roster charges to reach twelve, at an invented one point two million each, so three point six million more.

Total charges against that thirty million: twenty-six point six million. Real, usable room: three point four million.

Renounce the role player and the room becomes eight point four million, at the cost of any special right to re-sign him. Renounce both and it becomes twenty-six point four million, which is a genuinely different summer, purchased by giving up the ability to exceed the cap for the better of the two.

One further wrinkle catches people out. Each signing fills a roster slot, so it also removes an incomplete roster charge. Signing somebody for eight million dollars in the situation above consumes eight million and returns one point two million, a net cost of six point eight million. Room is spent in gross and recovered in dribs, and a team that budgets by headline salary alone will find itself a million short on the last contract of the summer.

The moratorium: why deals are agreed a week before they can be signed

The league year turns over at the start of July. The cap for that league year does not yet exist.

It cannot, because the cap is a share of audited revenue, and the audit is not finished when the calendar says the new year has begun. So the agreement inserts a gap. Negotiations open at a fixed hour at the end of June, published by the league each spring. Signing does not open until several days later, once the cap, the tax line and both apron levels have been calculated and announced. The interval between those two moments is the moratorium.

During it, teams and players may talk, may agree terms, and may shake hands. They may not sign, and the agreement is explicit that what they hold is not a contract. A handshake in the moratorium is unenforceable by either side. It holds because breaking it would be remembered, not because anything obliges it to.

A short list of contracts is exempt, essentially the ones nobody negotiates: rookie scale deals for first-round picks, minimum contracts, two-way contracts, and the qualifying offer itself. Everything with a real negotiation attached waits.

One change in the current agreement altered the texture of late June more than anything else in the document. A team may now talk to its own outgoing free agents at any time, without waiting for the window. The window governs conversations with everybody else's players. The practical effect is that most re-signings are settled before the market technically opens, and the noisy opening night is mainly about the smaller group of players whose own teams have already decided against them.

Sign-and-trade, the escape valve, and the apron that narrowed it

A player wants to join a team with no cap room. That team wants him. Neither can do anything about it, because room is the only ordinary way to sign an outside free agent and the exceptions are too small.

The sign-and-trade is the answer, and it works because a team can always exceed the cap to re-sign its own player. He signs a new contract with his current team and is traded in the same motion to the destination. He gets a bigger contract than the destination could have offered by itself. His old team gets something rather than nothing.

The rules around it are deliberately tight. The contract must run at least three seasons and no more than four. Raises are limited to the smaller rate. The acquiring team is hard-capped at the first apron for the remainder of the league year the moment the deal completes, which means every subsequent signing, claim and trade that season must keep it under that line. And a team already above the first apron cannot acquire a player by sign-and-trade at all, which removed the manoeuvre from precisely the teams that used to rely on it. The restrictions stacked above the tax line turned sign-and-trade from a standard tool into a transaction available mainly to teams in the middle of the spending ladder.

There is also an obscure rule that explains why so many of these deals need a third team. If a player receives a large raise and his team is over the cap, he becomes a base year compensation player for trade purposes. His outgoing salary counts, for salary matching, as the greater of his old salary or half his new one, while the acquiring team takes on the full new figure.

An invented example makes the problem visible. A player earning five million signs a new deal at twenty million and is traded immediately. The acquiring team absorbs twenty million. The team sending him out is credited with ten million, because half of twenty exceeds his old five. The two sides of the trade are ten million apart before anyone has added a single other name, and somebody has to cover that gap. Usually it is a third team with a trade exception, which is why these deals take days to assemble and occasionally collapse over a rounding difference. The wider salary-matching rules for trades apply on top of all of it.

How does NBA free agency work for a team with no cap room?

Most of the league has no room, most summers, and still signs people. This is the toolkit.

The mid-level exception comes in three sizes, and which one a team holds is determined by where its payroll sits. The full version belongs to teams below the first apron and is the largest and longest. The taxpayer version, for teams between the aprons, is smaller and shorter. The room version belongs to teams that used cap space, and is smaller again, which is the price of having spent room in the ordinary way. Any of them may be split among several players. Using the full version, or the bi-annual, hard-caps a team at the first apron for the season.

The bi-annual exception is a smaller amount available only in alternating years and only below the first apron. Spending it commits a team to going without it next summer, a strange constraint that exists mainly to stop it becoming a second mid-level.

The disabled player exception is the one almost nobody explains. When a player suffers an injury that a league-designated physician judges will keep him out for the rest of the season, his team may apply for an exception to sign a replacement. It is worth half the injured player's salary or the non-taxpayer mid-level, whichever is smaller, it may only be used on a one-year contract, and it expires early in the new calendar year. It is a replacement mechanism, not a windfall, and the value cap ensures a team is never better off for the injury.

Trade exceptions are credits created when a team sends out more salary than it takes back. They last a year, cannot be combined with each other or with anything else to absorb one larger contract, and cannot be used at all by a team above the second apron. Their real function is to let a team do half a trade in February and the other half in August.

And the minimum salary exception, which no team can lose under any circumstances. Any team may sign any player to a minimum contract of up to two seasons at any point, regardless of payroll. For a veteran with enough service the league covers part of the cost and the team is charged at a lower defined level, which stops capped-out contenders preferring a rookie to a ten-year professional purely on price.

Two-way deals, minimums, and the last three spots on a roster

Free agency does not end when the reported deals stop. It ends in September, quietly, at the bottom of the roster.

A standard roster holds fifteen players and a team must carry at least fourteen. Alongside those, each team may hold three two-way contracts, reserved for players with limited service time. A two-way player earns a fixed proportion of the rookie minimum, does not count against the cap at all, and may be active for a capped number of regular-season games. He is not eligible for the playoffs unless his contract is converted to a standard one before the deadline for doing so, and the mechanics of how two-way deals actually function shape the last week of training camp at every club.

The minimum market is a real market, and it is the part of free agency where competitive imbalance shows up most nakedly. A contender and a rebuilding team can offer an identical minimum salary. They cannot offer an identical anything else. So the veteran minimum tier sorts itself almost entirely on non-financial grounds, which is exactly what a well-designed floor should produce, and is quietly one of the more successful pieces of the agreement.

Later in the season two further routes open. Ten-day contracts become available in the new year, and a team may sign a given player to two of them before it must either commit for the rest of the season or let him go. Hardship provisions allow a team with enough simultaneous injuries to exceed the roster limit temporarily. And the buyout market functions as a second, smaller free agency in February, constrained by the rule that a player waived after a date set each season is ineligible to appear in the playoffs for a new team, and by the first apron restriction barring the richest teams from signing bought-out players whose previous salary was above the full mid-level.

How does NBA free agency work in the first seventy-two hours?

Front offices do not improvise this. The sequence is modelled in May, argued about in June, and executed in an order where every step forecloses options in the ones after it.

1. 1Resolve every option and tender every qualifying offer.Team options, player options and the restricted free agent tenders are all settled on deadlines before the window opens. Miss a tender and a restricted free agent becomes unrestricted.
2. 2Choose to be a room team or an exception team.These are alternative modes, not a menu. A club spends cap room and gets the small room exception, or it stays over the cap and keeps the full mid-level and its Bird rights.
3. 3Decide which cap holds to renounce.Renouncing converts a hold into spending money and destroys the exception attached to that player for the summer. The decision cannot be reversed once made.
4. 4Agree terms during the moratorium.Nothing can be signed yet. This is the window in which the whole summer is verbally allocated, on agreements neither side can enforce.
5. 5Sign outside free agents first, if using room.Room is consumed before exceptions, and each signing also releases an incomplete roster charge, so the money available moves as the roster fills.
6. 6Re-sign your own free agents last.With their holds intact, the Bird exceptions permit going over the cap, for a longer term and larger raises than any rival was allowed to offer.
7. 7Check the hard cap before every remaining move.Using the full mid-level, using the bi-annual or acquiring a player by sign-and-trade fixes a ceiling for the entire league year. A transaction that would breach it is simply not permitted.
8. 8Fill the last spots on minimums and two-ways.The bottom of the roster is completed in late summer, once the market has cleared and the remaining players are choosing on grounds other than money.

Steps two and three are effectively irreversible for the summer. Everything after them is constrained by choices already made.

The compressed version is that the first day is about outside players, the second is about your own, and the third is about discovering what the first two cost you. Teams that reverse the order find their Bird rights renounced and their room already spent.

Reading a signing and naming the mechanism

The habit worth building is small. Every time a signing is reported, ask which right was exercised, and the answer is usually visible in the shape of the contract itself.

Five seasons with large raises is Bird rights. Nobody else was permitted to offer that, so the player re-signed where he was, and the size of the deal tells you nothing about whether anyone else wanted him.

Four seasons at a large number, with a team that had space, is cap room. That team renounced holds to create it, and the interesting question is which of its own players it gave up on to get there.

An identical amount that several players around the league sign for in the same week is the mid-level, and the team using it has just hard-capped itself for the season, which will explain something it fails to do in February.

Three or four seasons, arriving by trade on the day the player signed, is a sign-and-trade, and the acquiring team is now capped at the first apron whether it planned to be or not.

Two seasons, matched exactly by the incumbent, is an offer sheet that failed. The rival lost two days of a summer and the player is going back to work for the people he tried to leave.

And one season at a calculated figure, signed by a young player with no leverage, is a qualifying offer accepted, which means he has bet a year of his career on being healthier and better next summer than his own team expects him to be.

None of this requires knowing the cap number, which changes annually and is published by the league. It requires knowing which doors exist. Once the mechanisms are legible, most of July stops being surprising, and the genuinely interesting decisions turn out to have been taken weeks earlier, on renounce lists and option deadlines that nobody reports at all. The same is true across most of how the modern game is built and paid for. Free agency is only the loudest example.