How the NHL salary cap works, and why LTIR is not free money

Somewhere in late February, a broadcast panel will say that a club has three million dollars of cap space, and then say the same club is about to acquire a player carrying a six million dollar cap hit. Both statements will be correct. Nobody will explain why, because explaining why takes four hundred words and there is a game starting.

That gap between what the numbers say and what the numbers mean is most of the confusion around the NHL salary cap. The cap is not a bank balance. It is a daily accounting test, run against a ceiling that is itself a share of revenue the league has not finished earning, enforced by a withholding mechanism most fans have never had explained to them, and bent in half a dozen entirely legal ways that all carry a price.

What follows is the whole system, mechanism by mechanism, with the arithmetic done rather than asserted.

What the NHL salary cap actually limits

The cap constrains a defined quantity called Averaged Club Salary: the sum of the cap hits of every contract a club is responsible for on a given day, plus a list of charges for players who are not on the roster at all. It does not constrain what an owner spends in cash. It does not constrain what any individual player is paid. Those are separate numbers, and the whole system depends on their being allowed to differ.

Hockey's version is the strictest ceiling in North American sport, and the comparison is worth making properly because the four big leagues have picked four genuinely different instruments. Baseball has no cap, only a tax on payroll above a threshold, which is a price rather than a prohibition and which a determined owner simply pays. Basketball has a soft cap so thoroughly qualified by exceptions that most contending teams spend the whole season above it. Football has a hard cap sitting on top of contracts that are mostly not guaranteed, which makes the club's escape route cheap and the player's position weak. Hockey has a hard cap sitting on top of contracts that are fully guaranteed, which is the combination that produces all the interesting behaviour, and it is why the way different leagues cap spending is a better question than which cap is highest.

Fully guaranteed matters. A signed NHL contract is money the player will receive. There is no equivalent of the March release that costs the club only what it has already paid, and there is no non-guaranteed year sitting at the end of the deal as a free option. If a club no longer wants a player, it has to trade him, loan him to the minors, or buy him out, and each of those has a cost that stays on the cap sheet. The guarantee structure a league chooses determines almost everything else about how its front offices behave, and hockey sits at the strict end.

Three structural limits frame everything else. A club may hold no more than fifty standard player contracts at once, which is a hard constraint on how many prospects it can sign and a genuine planning problem for clubs with deep systems. The active roster is capped at twenty-three players for most of the season. And on any given night a club dresses eighteen skaters and two goaltenders, a number that mattered very little for cap purposes until recently and now matters a great deal.

• 50%The players' share of hockey related revenue
• 15%Upper and lower limits, as a margin either side of the midpoint
• 50%Most of a cap hit one club may retain in a trade
• 7.5%The performance bonus cushion, as a share of the upper limit

Rule-defined shares from the collective agreement, not figures from any particular season.

Where the NHL salary cap number comes from

No one sets the cap by judgement. It is derived, and the derivation starts with a defined term: hockey related revenue.

Hockey related revenue is, broadly, everything the league and its clubs earn from the business of playing NHL hockey. Gate receipts. Broadcast and streaming rights, national and local. Sponsorship, licensing and merchandise. Concessions and parking attributable to game nights. Playoff income. A defined share of arena revenue. What sits outside it is revenue from businesses that would exist whether or not the hockey team did, and certain streams enter the pool only after their direct costs are deducted. The definition runs to many pages in the agreement because every line item in it is worth real money to both sides, and both sides argued over every line.

Once the pool is defined, the chain is short. Project hockey related revenue for the coming season. Take the players' share of it, which under the current agreement is one half. Adjust for benefits and for the accounting carried over from previous seasons. Divide by the number of clubs. What comes out is the midpoint of the payroll range.

The two limits are then fixed proportions of that midpoint. The upper limit sits fifteen per cent above it. The lower limit sits fifteen per cent below it. Every club in the league gets the identical pair of numbers, and the band between them is thirty percentage points wide. That band is the entire spread of payroll the system permits, which is a remarkable thing to say out loud: the richest club in the league and the poorest are legally confined to the same narrow strip.

Two consequences follow, and both are larger than they sound.

Because the cap tracks revenue, the denominator grows. A contract that looks enormous against this season's ceiling looks ordinary against the ceiling four seasons from now, without a dollar of it changing. Front offices price that in, and it is one of the reasons long deals for genuinely elite players age better than the reaction to them suggests.

Because every club shares the same ceiling and the same floor, competitive advantage cannot come from spending more. It has to come from spending better, or from spending at a moment when nobody else can. That is a fundamentally different game from the one played in European football, where a club's ceiling is a function of its own revenue and the constraint is proportional rather than absolute. Hockey's version compresses the league deliberately, and the compression is the product.

Growth is not automatic, though, and the reason is the mechanism that actually delivers the split.

Escrow is what makes the fifty-fifty split real

A fifty-fifty division of revenue sounds like a statement of fact. It is closer to an engineering problem, because the two halves of the equation happen at different times.

The cap has to be published before the season, because clubs cannot sign players against a number that does not exist yet. Revenue is not known until the season has finished and the books are closed. Contracts are written in dollars, not in percentages, and they are guaranteed. So the league sets the ceiling on a projection, thirty-something clubs then commit somewhere inside the payroll range, and the total they commit will not equal half of the revenue that eventually arrives. It cannot, except by coincidence.

Escrow is the correction. A percentage is withheld from every player's pay throughout the season and held in an escrow account. When the season ends and hockey related revenue is finally calculated, the accountants work out what the players' half actually amounts to and compare it with what was actually paid. If the players were collectively underpaid, the withheld money comes back to them, with a further payment from the clubs if the shortfall is larger than the escrow held. If they were collectively overpaid, the withheld money goes to the clubs, up to the amount needed to restore the split.

That is the whole mechanism, and it is the part of the system most likely to be described in a single clause and then dropped. It deserves better, because without it the cap does not do what people think it does.

The cap is an ex ante control on each individual club. Escrow is an ex post correction on the aggregate. Neither one alone produces a fifty-fifty split. Thirty-something clubs each obeying a ceiling tells you nothing about whether the total lands on half of a revenue figure that was still a guess when the ceiling was published. Escrow is the thing that turns a stated share into an enforced one.

Three consequences fall out of it, and each explains something that otherwise looks strange.

A player's stated salary is a ceiling in real terms rather than a promise. The contract is guaranteed as a number of dollars. What reaches the bank is that number less the escrow withheld, plus whatever comes back after reconciliation. Two players with identical contracts in different seasons can take home meaningfully different amounts, and neither club has done anything wrong.

Escrow rates move with the gap between projection and reality, which means a bad revenue year lands on the players immediately and lands on the clubs not at all. That asymmetry is why escrow is the single most resented word in the sport, and why the union's negotiating energy goes into capping the rate rather than into the headline share.

And when revenue falls far enough that one season of escrow cannot close the gap, the shortfall becomes a debt. The two sides have handled that by deferring a portion of salary, capping escrow at agreed percentages so the withholding never becomes punitive, and holding the upper limit flat until the accumulated balance is repaid. That is why the ceiling can sit almost still for several seasons while revenue is visibly recovering, and then move sharply once the balance clears. Clubs plan around that schedule, because a flat cap is a very different planning environment from a rising one and everyone knows in advance which they are in.

The short version: raise the cap and cut escrow are the same demand, made from opposite ends of the same equation.

Why the NHL salary cap has a floor as well as a ceiling

The lower limit is not decoration, and it is not a gesture towards competitive balance. It does three jobs.

The first is arithmetic. If every club spent near the floor, the players would receive far less than their half, escrow would return everything and the clubs would still owe a shortfall payment. The split would then be delivered by a lump sum at the end of the year rather than by wages during it, which is a worse outcome for everybody: players want the money in their contracts, and clubs want the money to buy them players. The floor keeps aggregate payroll close enough to the target that escrow stays a correction rather than a redistribution scheme.

The second is competitive. A hard ceiling without a floor is a licence to field a cheap team, collect the national broadcast money and the revenue share, and finish last profitably.

The third is that eighty-two games are worth less if a meaningful number of them are played by clubs that have stopped trying. The floor protects the value of the schedule, which is the asset the broadcast deals are actually buying.

Here is the part that gets skipped. The floor is measured in cap hits, not in cash. A club satisfies the lower limit with Averaged Club Salary, and Averaged Club Salary includes a lot of money that is not being paid to anyone currently on the ice.

Retained salary counts towards the retaining club's total. Buyout charges count. Cap recapture penalties count. Contracts signed at thirty-five or older count whether the player plays or not. Salary loaned to the minors counts for whatever portion sits above the burial threshold. A club can therefore approach the floor while carrying a roster substantially cheaper than the floor implies.

That gap turns the league's least wealthy clubs into its most useful ones. A club sitting under the lower limit that agrees to retain half of a departing player's cap hit is doing two things at once: it is being paid a draft pick by the contender that needs the space, and it is moving closer to a compliance requirement it had to satisfy anyway. Taking an unwanted contract off a cap-squeezed rival works the same way, and is also paid for in picks. The clubs that get called cheap in February are frequently the most active brokers in the league, precisely because their constraint is somebody else's scarce asset. The agreement also moves money directly from the largest clubs to the smallest through a revenue sharing pool, with conditions attached, which is what makes the floor survivable for a club that could not otherwise reach it.

Average annual value is the unit of account, not salary

Every contract has exactly one cap hit and it never changes. Total compensation over the term, meaning base salary plus signing bonuses, divided by the number of years. That figure is charged in year one and in the final year and in every year between, regardless of what the club is actually paying in any of them.

This is the deepest structural difference between hockey's cap and football's. In the NFL, a player's cap charge moves every year as base salary and prorated bonus combine, which is what makes restructures possible and what produces the dead money accelerations that dominate the coverage. The way proration and acceleration drive the NFL's cap has no real hockey equivalent. In the NHL the charge is flat and the cash is lumpy, which moves all the cleverness to a different place.

Take an invented five-year contract worth thirty million dollars, with salaries of eight, seven, six, five and four million.

• Cash paid
• Cap charge

Year 18m6m

Year 27m6m

Year 36m6m

Year 45m6m

Year 54m6m

An invented $30m contract paid on a declining schedule. The cap hit is the average and never moves. The salary shape is what the club and the player actually negotiate over.

Show the numbers

Worked example: cash paid versus cap charged, five-year deal

Item	Cash paid	Cap charge
Year 1	8m	6m
Year 2	7m	6m
Year 3	6m	6m
Year 4	5m	6m
Year 5	4m	6m

The cap hit is six million in all five years. Nothing about the payment schedule alters it. So why does anyone negotiate the shape at all?

Three reasons, all of them about what happens if the relationship ends early.

Signing bonus money is paid regardless of nearly everything. It is owed in full whether the player is injured, whether the club buys him out, and, historically, whether the season is being played at all. Money written as a signing bonus is money the player is going to receive. Agents push hard for it, and a contract paid mostly in signing bonuses is effectively buyout-proof, because a buyout is calculated on remaining salary and there is very little salary to reduce.

Cash matters to the owner even though only the cap hit matters to the cap. A club acquiring a player with two years left, a two million dollar cap hit and nine million of remaining cash is buying a cheap cap charge and an expensive payroll. Clubs with tight cash budgets turn those deals down, and clubs with deep pockets buy them. That is one of the very few places inside a hard cap where an owner's actual wealth still moves the market, and it is why the phrase "cap hit and real dollars are different problems" is a permanent fixture of general manager interviews.

And the shape is regulated, because it used to be abused. Under the current agreement no year may differ from the next by more than thirty-five per cent of the highest year's salary, and no year may be less than half the highest year. The agreement taking effect in September 2026 narrows the year-to-year band further and raises the floor on the lowest year, tightening the same screw again. Both rules exist for one reason.

Front-loading was the loophole, and cap recapture was the punishment

The attack on average annual value is obvious once stated. The cap hit is total money divided by total years, so adding years reduces it. Add years the player will never play, price those years at close to the league minimum, and the cap hit falls a long way below what the player is actually being paid during the seasons he intends to play.

A twelve-year contract with the last four years at minimum salary is not a twelve-year contract. It is an eight-year contract with a discount stapled to it, and every person in the room knows the player will retire rather than skate those tail years for a fraction of his market rate. The cap hit was real on paper and fictional in substance, and clubs with the cash to pay heavily in the early years bought a cheaper cap charge than clubs without it. A hard cap designed to equalise had produced an advantage available only to the wealthy.

Two rules closed it, and one of them is the harshest instrument in the agreement.

The first is term and shape. Maximum contract length is currently eight years when a club re-signs its own player and seven when a player signs elsewhere, and from September 2026 those fall to seven and six. Combined with the variance rules, a modern contract simply cannot carry the long, cheap, unplayable tail that made the tactic work.

The second is cap recapture, and it was applied retrospectively to contracts already signed under the old rules. If a player retires before such a contract expires, the club is charged the difference between the cash it actually paid him and the cap hits it actually charged, spread across the years that remain on the deal. The charge follows the benefit rather than the player: if he was traded partway through, the penalty is divided between the clubs in proportion to what each of them gained from the discount.

Recapture is punitive by design and it is unusual in sport for being retrospective. It converted a closed piece of accounting into a dated future liability that no buyout could clear and no trade could shed. The league's justification was straightforward and difficult to argue with: the discount was only ever legitimate if the player actually skated the tail years, so if he does not, the discount is repaid. Those contracts belong to a closing era, but recapture charges outlive them, and they are among the items that carry over into the new playoff cap calculation described below.

The same reasoning produced the thirty-five and over rule at the other end of a career. A contract signed by a player who is thirty-five or older on the thirtieth of June preceding it counts against the cap whether or not he plays, and buying it out does not clear the charge in the way an ordinary buyout does. A club cannot sign an ageing player to a long deal at a suppressed cap hit and then be relieved of it when he stops playing.

Long term injured reserve is permission to exceed the cap, not money

This is the single most misreported mechanism in the sport, so it is worth being precise about what it is and what it is not.

A club may place a player on long term injured reserve when he is expected to miss at least twenty-four days and ten games. That is the threshold. It is a genuine injury designation, and the club has to be able to justify it.

Long term injured reserve does not remove the player's cap hit. The contract stays on the books at full value for the whole time he is there. Every sentence that begins "the club cleared eight million by putting him on LTIR" has described an operation that does not exist.

What the designation grants is permission to exceed the upper limit by a calculated amount, and the calculation happens at the moment of placement. The club's existing cap space is consumed first, and the relief covers the remainder, up to the injured player's cap hit. A club sitting hard against the ceiling when it makes the placement receives close to the full hit as relief. A club with room receives correspondingly less, on the reasonable principle that it should use the room it already has before the league grants it more.

Then comes the cost, which almost never makes the broadcast. A club operating above the upper limit under long term injured reserve does not accrue daily cap space. Accrual is explained in full in the next section, and it is the engine of in-season roster building. Forfeiting it for a whole season is routinely worth more than the relief the club received. This is why competent front offices delay placement until the day they actually need the room, and why a club with real space will often carry an injured player against the cap rather than take relief it does not yet require.

Two further properties are worth holding onto. The relief is a fixed pool set at placement, not a balance that grows; sign a replacement whose cap hit exceeds the pool and the club is simply non-compliant, with no mechanism to fix it other than sending someone down. And the injured player cannot be activated until the club is compliant without the relief, which is why a club that spent aggressively in January sometimes cannot bring back a healthy player in March.

For years the designation had one more property, and it was the one that made people angry. Compliance was a regular season obligation, so a club could hold a genuinely injured star out for the entire regular season, spend the relief on reinforcements, and then dress him for game one of the first round at no cost whatsoever. The rules permitted it. The optics were dreadful.

The current agreement attacks that from both ends. On the regular season side, replacement players are now constrained: the total salary of the players brought in may not exceed that of the player they are replacing, and the average cap hit of those replacements generally may not exceed the previous season's league average salary, unless the injured player is genuinely finished for the season and the playoffs or the league and the union agree otherwise. A club can still cover an injury. It can no longer convert one into an upgrade.

The playoffs used to be a different game, and from 2026 they are not

The postseason was historically outside the cap entirely. Once the last regular season game was played the compliance test stopped, and a club could ice whatever lineup it liked. This was not an oversight. Daily accrual, prorated charges and the whole apparatus of Averaged Club Salary are built around a fixed-length regular season, and nobody had designed an equivalent for a variable-length tournament.

From the 2026 postseason, that changed, and the design chosen is neat enough to be worth walking through.

Compliance is now tested game by game, and it is measured against the lineup rather than against the roster. The combined cap hits of the eighteen skaters and two goaltenders a club dresses must fit inside a playoff figure, with carryover charges added for the things that are not attached to a player in the lineup at all: buyouts, contracts signed at thirty-five or older, retained salary, salary loaned to the minors above the burial threshold, grievance awards and cap recapture penalties. Lineups are submitted to the league before a deadline on the day of the game.

Counting the dressed lineup rather than the whole roster is the clever part, and it does two things at once. It means an injured player's cap hit is simply not counted in the postseason, which removes the need for long term injured reserve in the playoffs altogether and therefore removes the thing there was to abuse. And it means a club cannot bank a large cap hit through the winter and collect the player in April, because in April the hit only counts on the nights he plays, and on those nights somebody else has to come out of the lineup to make room for him.

The knock-on effect is that a deep run is now a roster-shape problem as well as a health problem. A club carrying several large contracts it cannot dress simultaneously has a constraint in April it never used to have, and the way a playoff bracket is built and survived now has a line of accounting running underneath it.

Cap space accrues by the day, which is why deadline players look cheap

Compliance is daily, and that word is doing an enormous amount of work.

On each day of the regular season a club is charged one day's share of every cap hit it carries: the annual figure divided by the number of days in the regular season, a divisor the league fixes each year. A club that sits below the ceiling on a given day banks the unused portion. The next day it banks again. By February a club that has carried an open roster spot since October and kept a couple of million spare is holding a pool of accrued space substantially larger than the gap between its payroll and the ceiling.

The charge for an arriving player prorates the same way. A player acquired with a quarter of the season remaining is charged a quarter of his annual cap hit, because there are only that many days left to charge him for. Nobody pays a full year's cap hit for two months of hockey.

Put those together and the contradiction from the opening paragraph dissolves. A club described as having three million of space in late February can comfortably absorb a player with a six million dollar annual cap hit, because the six million becomes a fraction of itself over the days remaining, and because the accrued pool is a different and larger quantity from the headline figure being quoted. The panel was not confused. The two sentences were measuring different things.

Three practical consequences follow.

Cap space is worth more in October than in March, and the market prices it accordingly. A club that intends to be a buyer at the deadline starts manufacturing space in the autumn by carrying a thin roster, not on deadline morning by making phone calls.

The cost of adding a player rises the earlier in the season you add him, which is why so much business waits, and why a good player on a bad team spends four months in trade rumours before anything happens.

And a club operating in long term injured reserve overage accrues nothing at all. It arrives at the deadline holding exactly the relief it was granted in November and not a dollar more, while a rival that stayed compliant has been quietly compounding since opening night. That is the real reason the designation is expensive, and it is why the sequencing of a placement is one of the more consequential decisions a general manager makes all year.

Buyouts, and the two-thirds arithmetic that surprises people

A buyout is how a club ends a guaranteed contract early. It is not free, it is not always available, and the cap consequences are not what most people assume.

The main window opens shortly after the Stanley Cup final and closes at the end of June. Clubs that go to salary arbitration get a short second window afterwards, and there is a separate right to walk away from an arbitration award above a threshold set in the agreement. Outside those windows the option does not exist. A club that decides in January that it has made a mistake waits until summer.

The contract goes on unconditional waivers first. If another club claims it, there is no buyout: the player changes teams at full price and the original club is relieved of the whole thing, which occasionally happens and is a much better outcome for everyone except the claiming club's cap sheet.

Then the arithmetic, which is worth doing slowly because the shape of it surprises people.

1. 1The window opensA club may only buy a contract out in a defined period, the main one running from shortly after the Stanley Cup final to the end of June. Outside it the option does not exist.
2. 2The contract clears waiversIt goes on unconditional waivers first. A claim by another club cancels the buyout and the player simply changes teams at full price.
3. 3Count the salary still owed, not the cap hitEight plus six plus four is eighteen million dollars. Signing bonus money is excluded from this figure entirely, because it is owed in full whatever happens next.
4. 4Apply the rate for his ageHe is 28, so the club pays two thirds. Two thirds of eighteen million is twelve million. Had he been under 26 on the date of the buyout, the rate would be one third and the cost six million.
5. 5Stretch the payments over twice the termThree years remaining becomes six years of payments. Twelve million over six years is two million of cash a year.
6. 6Work out each year's cap chargeCap hit less the cash the club no longer pays that year. Year one is six million minus eight million of salary less the two million still going out, which is nothing at all. Year two is two million. Year three is four million.
7. 7Carry the tailYears four, five and six carry two million each, for a player who left three seasons earlier and is very likely playing somewhere else.
8. 8Add up what changedEighteen million of charges across three years became twelve million across six. The club bought space now and paid for it in duration.

Worked with an invented contract. A player aged 28, three years left, salaries of $8m, $6m and $4m against a $6m cap hit.

Two things in that sequence deserve emphasis.

The rate is applied to remaining salary, not to the cap hit and not to total contract value. This is why the shape of a contract matters so much, and why a deal paid heavily in signing bonuses is close to buyout-proof. If most of what remains is bonus money, there is almost nothing for the two-thirds rate to bite on, and the club would be paying nearly the full remaining value to be rid of him.

And the annual cap charge is not the annual cash payment. It is the cap hit minus the saving, and the saving is that year's salary less that year's buyout payment. In a front-loaded year, where salary is well above the cap hit, the saving is large and the charge collapses.

• Cap charge if kept
• Cap charge after the buyout

Year 16m0m

Year 26m2m

Year 36m4m

Year 40m2m

Year 50m2m

Year 60m2m

Years four to six carry nothing in the first series because the contract would have expired. The buyout replaces three years of charges with six.

Show the numbers

The same invented contract, kept against bought out

Item	Cap charge if kept	Cap charge after the buyout
Year 1	6m	0m
Year 2	6m	2m
Year 3	6m	4m
Year 4	0m	2m
Year 5	0m	2m
Year 6	0m	2m

Eighteen million of cap charge over three years becomes twelve million over six, and the six million difference is precisely the third of the salary the club is no longer paying. The relief is concentrated in year one, which is exactly where a club that has miscalculated needs it, and the price is three extra years of a small charge for a player nobody will remember is on the books.

Push the front-loading harder and the arithmetic runs past zero. Under the older rules, where the salary shape could be far more extreme, the same calculation on the same eighteen million of remaining salary paid as ten, six and two would have produced a negative charge in year one. A club could buy a player out and have more cap space that season than if it had kept him, which is the sort of result that makes a rule look ridiculous and is one of the reasons the variance limits exist.

The strategic reading is straightforward. A buyout converts a large, immediate, unavoidable problem into a small, distant, unavoidable one. It is the right move when the club needs the space inside a contention window and is prepared to be worse in three years' time, and it is the wrong move roughly whenever a club does it twice in the same summer.

Retained salary, and the three limits that stop it becoming a market

Because the cap is hard and cap hits are flat, clubs need a way to move a player whose charge no acquiring club can absorb. Retained salary is that mechanism, and it is one of the few genuinely elegant things in the agreement.

In a trade, the selling club may retain up to half of a player's cap hit and half of his salary, in the same proportion, for the entire remaining term of the contract. Not for the season. For the term. A club that retains half of a four-year contract has committed itself to that charge for four years, long after everyone has forgotten why.

Three limits keep it from becoming a free market in cap space. A club may carry no more than three retained-salary contracts at any one time. The aggregate of everything it has retained is capped as a share of the upper limit. And a single contract may only be retained on twice in its life.

That last limit is the one that produced the three-team trade as a standard piece of technology. The selling club retains half. The player moves to a middle club, which immediately flips him to the buyer while retaining half of what it just took on. A quarter of the original cap hit arrives at the club that actually wanted the player, and the middle club is paid a draft pick for the use of its cap sheet and one of its three retention slots.

50%25%25%

• The original club keeps4m
• The middle club keeps2m
• The buying club carries2m

Invented figures. Each retention is capped at half of what the retaining club is carrying, and no contract may be retained on more than twice.

Show the numbers

What double retention does to an $8m cap hit

Item	Value
The original club keeps	4m
The middle club keeps	2m
The buying club carries	2m

Every part of that is priced. The selling club is accepting a charge for the full remaining term and will normally receive a lower return in the trade because of it. The middle club is renting out a slot it may need in July and will charge accordingly. The buying club is paying picks to two other organisations for the privilege of a cap hit it could have afforded outright in a world without a ceiling.

Two smaller rules complete the picture. A player whose salary has been retained cannot be traded back to the retaining club for a period after the trade, which stops a club parking a contract elsewhere and collecting it later. And retained salary counts towards the retaining club's own Averaged Club Salary, which is what makes floor clubs the natural sellers of the service and closes the loop with the lower limit.

Entry level contracts, the slide rule and the bonus you pay for next year

Young players sign under a separate regime, and it is the most valuable regime in the sport because it is the only place a club can acquire genuine production at a price the market does not set.

Length is fixed by age at signing. A player who signs at eighteen through twenty-one gets three years. At twenty-two or twenty-three it is two years. At twenty-four it is one. From twenty-five he signs a standard contract like anyone else. Base salary is capped at a maximum the agreement sets and raises over its term, so the difference between the best young player in the league and the fiftieth best is close to nothing in cap terms. That gap between production and price is the strongest force in roster construction, which is why the draft and how clubs actually value picks matters far more in a hard cap league than the drama of the night itself suggests.

Performance bonuses are the pressure valve. Entry level players can earn bonuses in two schedules: a set of individual achievement thresholds covering games played, goals, assists, points and ice time, capped per player per season; and a much larger schedule for league-leading finishes, major trophies and postseason awards, available to a narrower class of players. A great rookie season can therefore be paid something closer to its worth without breaking the salary maximum.

The cap treatment of those bonuses is where clubs get caught. During the season a club may exceed the upper limit by up to seven and a half per cent of it to accommodate bonuses that might be earned. That is the performance bonus cushion, and it is not a grant. If the bonuses are actually earned and the club finishes above the ceiling as a result, the overage is charged to the following season's cap.

That charge is a bonus overage carryover, and it is the least glamorous line on any cap sheet. A club with three excellent rookies can find itself starting the next season a few million dollars poorer because those rookies were excellent, which is a strange incentive to have written into an agreement and is nonetheless exactly how it works. Front offices with a young roster budget for it in advance. Front offices without one discover it in July.

The slide rule is the other wrinkle, and it exists so clubs are not punished for a short look at a teenager. If an eighteen or nineteen year old signs an entry level contract and plays fewer than ten NHL games that season, the contract slides: the year does not count, and the deal begins the following season instead. A player can slide twice, and where a player's twentieth birthday falls in the closing months of the calendar year an extra year is added to the term. The practical effect is that a club can give a young player nine games, decide he is not ready, send him back, and lose nothing at all from the three cheap years it will eventually want. Ten games and the clock has started.

One related rule sits alongside all of this. A contract loaned to the minors only comes off the NHL cap up to a threshold, set as the league minimum salary plus a fixed sum written into the agreement. Anything above that threshold stays charged. This is what stops a club burying an expensive mistake in the American league and pretending it never happened, and it is why the phrase "we can just send him down" is usually wrong for exactly the players a club most wants to send down.

Expansion draft protection is a cap problem written years in advance

Expansion is not a permanent feature of the calendar, but the rules governing it are permanent enough to shape contracts signed years before anyone announces a new club, which makes them a cap-adjacent constraint worth understanding.

When the league expands, every existing club submits a protection list. The two permitted shapes are seven forwards, three defencemen and one goaltender, or eight skaters in any combination and one goaltender. Everyone not on the list is available to the expansion club, which selects one player from each existing team.

Protection is only half of it. There are exposure requirements, and they are the half that hurts. A club must expose at least one defenceman and two forwards who are under contract for the following season and who meet a games-played standard, defined as forty games in the previous season or seventy across the previous two. It must also expose a goaltender who is under contract or a restricted free agent with a qualifying offer. First and second-year professionals and unsigned draft choices are exempt from selection altogether and do not need protecting, which is why a club with a strong young core loses less than the headlines suggest.

The cap connection runs through the no-movement clause. A player who holds one must be protected unless he agrees to waive it, and that agreement is a negotiation with a price. A club that handed out a no-movement clause four years ago as a way of saving a few hundred thousand dollars on a cap hit can find that the clause now forces it to protect a declining player and expose a useful one. Clubs have paid draft picks to persuade an expansion team to take a contract they wanted gone, and they have paid players to waive clauses so the list could be built properly.

That is the real lesson of expansion for cap purposes, and it generalises well beyond expansion years. Every concession in a contract that is not a dollar is still a cost. Trade protection, movement clauses and signing bonus structures are all paid for somewhere, usually in a year when the general manager who granted them has moved on.

How a general manager actually plans a cap sheet

Everything above is mechanism. Here is how it assembles into a job.

A cap sheet is not a number, it is a ladder of expiries. What a general manager is actually managing is the alignment between when his best players are cheap and when his roster is good, and those two things are set by rules rather than by ambition.

Player control runs on a fixed schedule. An entry level contract gives three cheap years. After it the player is a restricted free agent, which means his club holds his rights, must make a qualifying offer to keep them, and may face a salary arbitration hearing where a third party sets the number. He becomes an unrestricted free agent at twenty-seven or after seven accrued seasons, whichever comes first. That means every young player arrives with a known, dated schedule attached: three cheap years, then a negotiation, then a second negotiation, then the open market.

The central decision on the second contract is bridge or term. A short bridge deal keeps the cap hit low while the player is still proving himself and pushes the real negotiation two or three years down the road, by which point he will be more expensive and closer to free agency. A long deal signed immediately costs more now and locks the price in before it rises. Neither is correct in general. What makes one correct is where the club is in its window, because a bridge deal that expires in the same summer as three other contracts has created a problem the club will not be able to solve.

That is the actual planning discipline: reading the cap sheet three and four years out and looking for the years where too much comes due at once. A club with four important players all reaching restricted free agency in the same offseason is going to lose one of them, and it will know that two years in advance if anyone bothers to look.

The window itself is defined by the cheap contracts. A club with several good young players on entry level deals can afford to pay market rate everywhere else at the same time, and that combination is what a contending roster is made of. It lasts exactly as long as those contracts do. Everything a general manager does during it, the deadline additions, the retained salary trades, the buyout that clears a mistake, is an attempt to convert a dated advantage into something permanent before the arithmetic catches up. This is the same logic that runs through most of how the modern NHL is built and played, and it is why two clubs with identical payrolls can be at completely opposite points in their life cycle.

The mistakes are correspondingly predictable. Buying at the deadline in a year the club was not actually close, and paying in the picks that would have started the next window. Extending a good player at thirty on the term he earned at twenty-seven, and carrying the last three years of it into the next rebuild. Handing out movement protection to save cap dollars, then discovering the protection is the binding constraint. Placing a player on long term injured reserve in November because the relief looked useful, and forfeiting five months of accrual for a room the club did not need until March.

None of those is a failure to understand the rules. They are all failures of sequencing, which is what cap management actually is.

What to check on any club's cap sheet

If you want to judge a club's position rather than react to a February headline, five things tell you nearly everything, and all five are public.

Committed cap hits for the season after next, not this one. The current season is a solved problem by the time it starts; every club is compliant or it does not play. The season after next is where the choices are still visible.

How many roster spots that money covers. Thirty million dollars committed to six players and thirty million committed to fourteen are completely different situations, and only one of them can be fixed by signing somebody.

How many entry level contracts expire in the same summer. This is the single clearest signal of when a window closes. Four at once is a cliff and it is visible three years out.

Dated charges that are attached to nobody. Buyout tails, retained salary running to the end of a term, cap recapture penalties, contracts signed at thirty-five and over. These cannot be traded, bought out or waived away, and a club carrying a lot of them has already spent a season it has not played yet.

Whether the club is accruing space or sitting in long term injured reserve overage. A club that has been compliant since October is quietly getting richer every day. A club in overage is not, and will arrive at the deadline holding exactly what it had in November.

Five figures, none of them secret, all of them stable. They will tell you more about the next three seasons of any club in the league than the entire month of trade deadline coverage that is about to be produced about it.