Penalty kill systems in hockey: box, diamond and wedge

A penalty kill does not defend. It concedes, deliberately, in a place it has decided it can afford to.

That is the sentence the whole subject hangs on, and penalty kill systems in hockey only make sense once you accept it. Four skaters cannot cover the ice five attackers occupy. The arithmetic of that is settled and nobody is trying to beat it. What a kill chooses is which of the five attackers gets to be free, where he stands, and how long he gets to stand there before somebody arrives.

Every structure described below is an answer to that question, and every one of them is wrong in a different place. The box is wrong at the perimeter. The diamond is wrong on the flanks. The wedge is wrong at the point. The aggressive kill is wrong everywhere at once for short bursts and hopes the puck is gone before anybody notices.

Here is what each one buys, what each one pays, and why the best kills spend most of their time nowhere near their own net.

What a penalty kill is actually trying to do

The naive objective is to prevent a goal for two minutes. That is the result a kill wants and it is not the thing a kill is organised around, because organising a defence around surviving two minutes of sustained pressure is organising to lose slowly.

The real objective is narrower and much more achievable: touch the puck four times and send it away four times.

Work through why that is the same thing. A puck cleared the length of the ice has to be retrieved by the attacking team, organised into a five-man unit, brought back up the sheet and taken over the attacking blue line against a defence that is set and waiting. Call that fifteen seconds as a constructed example with round numbers. Four clears at fifteen seconds each is a minute. Add the face-offs, the puck frozen against the boards and the two or three seconds lost each time a unit resets, and the two minutes have gone without the kill ever having to defend a settled possession for long.

So the shape in front of the net is not the primary instrument. It is the thing that keeps the kill alive long enough to get a touch on the puck, and the touch is what it is actually playing for.

That reframing changes what counts as a good kill. A unit that blocks eleven shots has had a terrible two minutes. A unit that blocked two and cleared four times had an easy one, and the difference was made at the walls and at the blue line rather than in the slot.

• 4Skaters a team defends with while serving a single minor
• 5Minutes to be killed for a major, served in full regardless of goals
• 3Minimum skaters a team may ever have on the ice
• 2Minors a double minor is actually made of

Values set by the rulebook rather than by any season's results. No kill percentage, save rate or scoring figure is asserted here or anywhere in this article.

The clock rules that define the job

A kill's tactics change with the kind of penalty being served, and the differences are rule-driven rather than stylistic.

A minor is two minutes and it dies on a goal. The moment the power play scores, the penalised player is released and the teams are even. That exit is what makes a minor survivable in a particular way: a kill under heavy pressure knows the worst case is one goal rather than three.

A double minor is four minutes and it is two penalties, not one. This matters to the kill more than to the power play. If the attacking team scores during the first minor, only that minor ends, and a fresh two minutes begins from the instant of the goal. So a kill that concedes thirty seconds into a double minor has not bought itself much relief. It has two full minutes still to serve and has already given up a goal, which is the worst version of the situation.

A major is five minutes and goals do not shorten it. Nothing about the structure changes, but everything about the management does. Five minutes is long enough that personnel become the constraint rather than the shape. A kill running four forwards and two defencemen through five minutes is asking for shifts of thirty to forty seconds, and a tired killer is a killer who stops moving his feet, which is how a structure that was holding suddenly is not.

Overlapping penalties produce a two-man disadvantage and a queue. A team already two men down that takes a third penalty does not go to three skaters, because the floor is three. The third player goes to the box and his clock does not start until one of the earlier penalties expires, so the queue can stretch well past the arithmetic of the individual calls. The full ordering of how overlapping penalties are assessed and released is set out in the account of how penalties are classified and served.

One further detail catches people out, and it is a rule rather than a convention. If a goal is scored during a two-man advantage created by two minors, the penalty with the least time remaining is the one that ends. The kill goes from three skaters to four and still has a penalty running.

The one advantage the rulebook hands the short-handed team

A team below the numerical strength of its opponent is permitted to shoot the puck the length of the ice without being called for icing.

That single exemption is worth more to a kill than any formation, and it is the reason the structures below can afford to be as conservative as they are. A kill that wins a puck anywhere in its own zone has a legal, zero-skill, guaranteed way of making it somebody else's problem. Fire it down the ice and the clock keeps running.

Without the exemption, every clear would have to be a controlled one. A puck put out of the zone and over the centre line without a teammate touching it would be whistled back for a face-off in the defending zone, which is precisely the situation a kill is trying to escape. The exemption converts a skill problem into a decision.

Two limits on it are worth knowing, because both get forgotten.

The exemption covers icing and nothing else. A killer who shoots the puck over the glass and out of play from his own defensive zone is still assessed a minor for delay of game, and his team is still short-handed, so the result is a second penalty on top of the first. The length of the ice is available. The seats are not.

And the exemption is about numerical strength rather than about serving a penalty. A team that has killed its penalty and is back to even strength loses the privilege the instant the player steps out of the box, which occasionally produces an icing call against a team that was legally clearing the puck two seconds earlier.

The full text of the icing rule and the stoppage procedures around it are covered in the guide to icing and offside.

The kill before the kill: the neutral zone layer

The best penalty killing in hockey happens before the puck is anywhere near the net being defended, and television almost never shows it.

A power play that is denied entry to the offensive zone has to retrieve the puck, reorganise five players and travel the length of the sheet again. That is time spent on transport rather than on offence, and the penalty clock does not care which it was. A kill that denies two entries has effectively shortened a two-minute penalty to something closer to ninety seconds of actual exposure, and it has done so without making a save or blocking a shot.

So the first structural decision a kill makes is not which in-zone shape to use. It is how far up the ice to contest the entry.

A kill that pressures early meets the power play in the neutral zone, sometimes before the attacking blue line, and tries to force the puck in without possession. A kill that retreats concedes the entry and sets up in front of its own net, accepting that it will have to defend a settled possession.

Neither is free. Pressing up the ice with four skaters means four skaters are above the puck, and a power play that gets through that pressure is through it with nobody home. Retreating means a controlled entry every time, which is handing the attacking team the better category of entry on every possession. The full argument for why controlled entries are worth so much more than uncontrolled ones, and what a defence is really trying to force at the line, is worked through in the piece on zone entries.

Most kills contest the line for the first clear and then adjust based on what happened.

The 1-3 and the 1-1-2, and what a pressure kill provokes

Two neutral zone shapes cover most of what you will see.

The 1-3. One killer pressures the puck carrier high, often as the power play is still regrouping in its own end, and the other three form a line across the ice behind him. The three-man wall is the structure, the single forechecker is the irritant. What it takes away is the easy controlled entry, because the carrier has a player in his face and a wall of three behind him with no gap through the middle. What it concedes is the drop pass, because the lone pressurer can be beaten by the oldest entry play there is: draw him in, leave the puck behind, and he is now behind the play.

The 1-1-2. One killer pressures, a second supports him a few feet back, and the two defencemen hold the blue line. The second man is specifically there to eat the drop pass. What it takes away is the choreographed entry. What it concedes is width, because two players stacked through the middle cannot also cover both walls, and a power play with a confident passer can get the puck to a winger on the boards and enter wide.

What a pressure kill is actually provoking is a rushed decision rather than a turnover. Turnovers are a bonus. The objective is to make the power play arrive at the line a half-second before it wanted to, with a player receiving the puck earlier than he expected.

That half-second is the whole bet. A power play is a sequence of pre-agreed reads, and a player under unexpected pressure makes the safe read instead of the intended one. Safe reads do not score. Enough of them and the penalty expires with nothing in it.

The counter-bet, which coaches of pressure kills accept, is that a power play that solves the pressure once is in on a broken defence, and a broken defence four against five is a goal more often than not.

The box: conceding the perimeter on purpose

The box is the default in-zone structure and it is honest about what it is doing.

Four killers at the corners of a rectangle in front of the net. Two forwards high, near the tops of the circles. Two defencemen low, flanking the net front. The shape collapses towards the middle as the puck goes low and expands upward as the puck goes high, and it blocks what it can reach.

What the box takes away is the slot and the net front, which is where almost all goals are scored from. It keeps four bodies and four sticks between the puck and the most dangerous ice on the sheet, permanently, and it never voluntarily leaves.

What the box concedes is everything else. Shots from the point. Shots from the half-wall. Unhurried passing around the perimeter. The box does not contest any of it, because contesting it means leaving the rectangle, and leaving the rectangle is how the middle opens.

The honest case for the box is that it is the structure least likely to produce a catastrophe. Nothing in it depends on winning a one-on-one battle or on a read being correct. Four players hold a shape, shuffle in the direction of the puck, and get sticks into lanes. A tired unit can run a box. A unit with no particular personnel advantage can run a box. It is the structure that degrades most gracefully.

The honest case against is equally plain. Against a patient power play with good passers, the box loses eventually, and everyone involved knows it. Enough perimeter passes will move four players once too often and a seam will open. The box is a bet that two minutes is shorter than the time it takes for that to happen, and against a very good unit that bet is not always a good one.

The diamond: paying the flanks to close the middle

The diamond exists because the box has a hole in the middle of it.

A rectangle has four corners and an empty interior. Put an attacker in that interior, standing in the slot with his stick on the ice, and the box faces a question it cannot answer: collapse inward to reach him and the half-walls are free, stay at the corners and he is alone in the most dangerous square metre on the ice.

The diamond rearranges the same four bodies to remove the interior. One killer high on the point. Two at the sides, roughly level with the hash marks. One low, in front of the net. Written out as positions, the shape is a rhombus with a player at the top, a player at the bottom, and one on each side.

What that buys is the middle. The high killer takes the point and the top of the zone. The low killer takes the player standing in the slot, and takes him permanently rather than reaching for him. The interior is no longer empty, which is the entire reason the shape exists.

What it pays is the flanks. The two side killers are now responsible for a larger arc each, and the lanes from the half-wall to the net are wider than they were in a box. Shots arrive from wider angles more often, and a power play with a left-shot on the right wall and a right-shot on the left wall can get the puck to the flank with time more easily than it could against a box.

A kill that chooses a diamond has decided that wide-angle shots are a better thing to concede than a free man in the slot, which is almost certainly correct and is still a concession.

25%25%25%25%

• Structures that protect the slot and concede the perimeter1
• Structures that close the middle and concede the flanks1
• Structures that fix a triangle and concede the point almost entirely1
• Structures that attack the puck and concede the free man more often1

A structural comparison of trade-offs rather than a count of outcomes. All four shapes use the same four skaters.

Show the numbers

What each structure concedes to get what it wants

Item	Value
Structures that protect the slot and concede the perimeter	1
Structures that close the middle and concede the flanks	1
Structures that fix a triangle and concede the point almost entirely	1
Structures that attack the puck and concede the free man more often	1

The wedge plus one: guaranteeing the middle, abandoning the point

The wedge plus one is the most structurally opinionated of the four.

Three killers form a tight triangle around the slot and in front of the net, and they do not leave it. Not for a puck on the wall, not for a point shot, not for anything. The fourth killer is a rover who chases the puck wherever it goes, pressuring the carrier on the half-wall, following it down to the goal line, coming back up to the point.

What it guarantees is that the middle is never empty and never thin. Three players in a fixed triangle in the slot is more coverage of the dangerous ice than any other arrangement of four bodies produces, and because the triangle never moves for the puck, it cannot be pulled out of position by puck movement.

What it abandons is the point, almost entirely. One rover cannot cover the point and the half-wall at the same time, and the triangle is not coming out to help. A power play against a wedge gets more or less unlimited shots from the top of the zone.

That is a deliberate and defensible trade, and the reasoning behind it is worth stating properly rather than assuming. A shot from the point has to travel the length of the zone, through three stationary bodies in the slot, to reach the goaltender. Most of those shots do not arrive. The ones that do arrive late and with their angle already read. So the wedge is conceding a large number of low-value shots in order to concede almost no high-value ones, and a shot count will make it look like the worst kill in the league while the goaltender has a quiet night.

The cost is to the rover, who skates the whole penalty, and to the triangle's discipline. The moment one of the three steps out to help on the wall, the shape is a bad box with a rover attached and it has the weaknesses of both.

The rotating box, and the kill that attacks the half-wall

The fourth family refuses to sit at all.

An aggressive or rotating kill attacks the puck the moment it arrives on a half-wall, sending the nearest killer out hard and rotating the other three to cover behind him. The puck is pressured again on the next pass, and the next, with a different killer leaving each time. Nothing is held.

What it buys is rhythm, and rhythm is the thing a power play actually runs on. A set unit has a sequence in mind: puck to the half-wall, look at the bumper, swing to the point, cross to the far flank, one-timer. Every one of those reads assumes a certain amount of time. A killer arriving half a second early does not need to win the puck; he needs only to make the receiver's first option unavailable, and the sequence collapses into a safe pass.

What it concedes is the free man, and more of him than a passive box concedes. By definition, a killer who has left the shape to pressure is a killer who is not covering anybody. The remaining three have to rotate to fill a hole they can see opening, and three players rotating always arrive later than one pass travelling.

Aggressive kills therefore live and die on the quality of the first pressure. Pressure that forces a hurried pass up the wall is a win. Pressure that arrives late, after the puck has already gone, is a killer out of the play for nothing, and the power play now has a four-on-three in the offensive zone.

This is the structure most dependent on personnel. It needs four killers who read passes rather than chase pucks, and it needs legs. Coaches use it in bursts, after a face-off win or after the power play changes personnel, rather than for a full two minutes.

Choosing between them, and switching inside one penalty

No team runs one structure for a whole season, and the good ones switch inside a single penalty. The cues are visible if you know them.

What the power play is set up in. A shape with a player standing in the slot demands a diamond or a wedge, because a box cannot reach him. A shape with three players strung high across the top of the zone is asking the kill to come out, and a box that stays home takes the whole arrangement away.

Where the puck is. Most kills are hybrids by location. Box when the puck is high, collapse to something closer to a triangle when it goes below the goal line, rotate aggressively when it arrives on a half-wall where the kill has numbers.

How long is left. A kill with ninety seconds to serve wants the puck and will pressure for it. A kill with twelve seconds left wants nothing except for nothing to happen, and will sit in the most passive shape it has.

Who is on the ice. A power play that has just changed personnel is a power play that is not yet organised, and the two or three seconds after a change is the cheapest moment in the penalty to pressure.

Whether the kill has already been burned. A kill that has conceded from the slot will tighten the middle and live with the wide shots. A kill that has conceded from a one-timer on the flank will push its side killers wider and live with a bit more traffic inside. Both adjustments are reactive and both are correct.

1. 1The whistleThe penalised player goes to the box and the clock is set at two minutes. The kill sends out its first pair of forwards and its best defensive pairing, usually its most reliable face-off man.
2. 2The face-offThe draw is taken in the penalised team's own end zone, which means the kill starts the penalty defending rather than in the neutral zone. Win it and the first clear is available immediately.
3. 3Win the drawThe puck is sent the length of the ice. Icing is not called on a short-handed team, so the clear is legal however far it travels, and ten or fifteen seconds come off the clock while the power play retrieves it.
4. 4Lose the drawThe power play is set up inside the zone with the kill already boxed in, and the first in-zone sequence begins with no entry needed and the full penalty remaining.
5. 5The neutral zone layerAs the power play comes back up the ice, the kill decides how far to contest. A 1-3 or a 1-1-2 tries to force the puck in without possession. A retreating kill concedes the entry and sets up at home.
6. 6The in-zone structureBox to protect the slot and concede the perimeter. Diamond to close the middle and concede the flanks. Wedge plus one to fix a triangle and concede the point. Rotating pressure to break the unit's rhythm and concede the free man.
7. 7Sticks before bodiesKillers prioritise taking away the pass that creates the dangerous shot rather than blocking the shot itself. A stick laid flat in a seam removes a shot that has not been struck.
8. 8The touchThe kill gets a stick or a skate on the puck. This is what the structure exists to produce, and it is the moment the penalty is actually being killed.
9. 9The clearThe puck goes down the ice, legally and without icing. The power play retrieves, reorganises and travels the length of the sheet again, and the clock runs throughout.
10. 10The changeA stoppage, or a long clear, lets four tired killers come off. A goaltender who freezes a reachable puck is buying that change at the cost of a fresh face-off for the power play.
11. 11ExpiryThe penalty runs out, the player leaves the box, and the kill's privileges go with him. The icing exemption ends the instant numerical strength is restored.

The rule-governed and tactical sequence in order. No timings, save rates or success percentages are asserted, only the order in which the decisions arise.

Face-offs while short-handed

The draw at the start of a penalty is taken in the penalised team's own end zone, which is a rule, and it means a kill begins every penalty defending rather than in transit.

That makes the first face-off the single most valuable puck in the sequence, and it is why coaches will send out a specialist centre for one draw even if he is not part of the kill unit they want on the ice, then swap him at the first stoppage. One clean win converts into a clear, and a clear converts into ten or fifteen seconds. A lost draw hands the power play a set-up with the full two minutes available and no entry to make.

The alignment on a short-handed defensive draw differs from the even-strength version in one important way. With four skaters there is one fewer body to place, so somebody's usual assignment goes unfilled. Most kills prioritise the two most direct routes to a dangerous shot, which means covering the near flank and the net-front area, and accept that the far point is reachable by the power play if it wins the puck back cleanly.

There is a second-order consideration that only applies while short-handed. A centre taking a defensive draw at even strength may be happy to tie the puck up and accept a scrum, because a scrum is a neutral outcome. Short-handed, a scrum is not neutral. It is five attackers and four defenders fighting over a loose puck in the defensive zone, which is a situation the kill loses more often than it wins. Killing centres are therefore coached to win the puck somewhere rather than to deny it everywhere.

The goaltender's job on a kill, including the shot he lets through

A goaltender on a penalty kill is doing a different job from the one he does at even strength, and the differences are not only about volume.

He sees less, because there is a body in front of him whose assignment is to remove his view of the release. He moves more, because the attacking shape is built on lateral passes and every lateral pass is a demand to cross the crease and re-set. And he is defending against a structure that has an extra passing option by construction.

The adjustment that follows is visible from the stands. Goaltenders generally play deeper against a power play than against even-strength possession. Coming out to challenge a shooter on the half-wall takes away that shooter's angle and leaves nothing at all if the puck goes cross-ice, because a goaltender high in his crease has further to travel to reach the far post. Sitting deeper concedes more of the net to the shot he can see in order to be able to arrive on the shot he cannot.

Then there is the shot a kill lets through on purpose, which sounds like a contradiction and is not.

A killer in a shooting lane has two options: get a stick or a body into the lane, or clear the lane so his goaltender can see. Blocking a shot from the point is not always the better play, because a blocked shot is a loose puck with four killers facing the wrong way and five attackers around it. A point shot that the goaltender sees cleanly from forty-five feet, with no deflection and no screen, is one of the least dangerous events in hockey. So killers are coached to take away the seam and the tip while leaving the long shot visible, which from the outside looks like four players standing aside and is in fact the structure working.

The freeze decision sits on top of all of it. A goaltender who covers a reachable puck stops the clock, which is time the kill no longer has to defend, and lets four exhausted killers change. He also hands the power play a fresh offensive-zone face-off with its best unit already in position. That trade, made in a fraction of a second several times a night, is one of the least discussed judgements in the sport. The full version of the goaltender's problem from the attacking side is set out in the account of how a power play works.

Killing a five-on-three is a different job

Two men down is not a worse version of one man down. It is a different task with different objectives.

Three skaters cannot form a box, because a box needs four corners. They form a triangle, and a triangle in front of the net accounts for two fewer attackers rather than one. Run the count: one killer goes to the puck, two remain, and there are four attackers away from the puck. Two free men, and the goaltender now has to respect a second option if he cheats towards the first.

Possession stops being a realistic objective. Three skaters cannot both hold the middle and send somebody to win a puck on the wall, so the kill stops trying to win the puck and starts trying to make it go away. Anything that moves the puck two hundred feet is a win. A blocked shot that ends up in the corner is a win. A puck frozen against the boards for a stoppage is a win.

The structure narrows to three jobs. One killer takes the net front and does not leave it. The other two take the two most direct passing lanes into the slot and keep their sticks flat. Nobody chases the point, because the point shot has to travel through three bodies and is the least of the available problems.

The one rule detail that changes the arithmetic is the expiry order. When the attacking team scores during a two-man advantage created by two minors, the penalty with the least time remaining ends. So the kill goes from three skaters to four, and it is still short-handed, and it has just conceded. The relief is partial by design.

The duration is usually short, because a five-on-three only lasts as long as the overlap between two penalties. A constructed example with round numbers: a minor is called, forty seconds later a second minor is called against the same team, and the two-man disadvantage therefore lasts eighty seconds before reverting to four skaters for the remaining forty seconds of the second penalty. The two-man portion is the overlap and nothing more.

Short-handed chances and why they are worth more than their frequency

Short-handed goals are rare enough to be treated as noise. They are not noise, and the reasons are structural.

The first is the exposure a power play voluntarily accepts. A power play sets up with both defencemen at the attacking blue line and three forwards below the tops of the circles. That is the most stretched formation any team adopts on purpose. A blocked shot, a mishit one-timer or an intercepted pass at the line all produce the same thing: a loose puck with the attacking team's two rearmost players standing at the wrong blue line.

So the chances a kill generates are few and they are of a quality almost nothing else in the game matches. A two-on-one going the other way with the goaltender having had no warning is a better chance than most even-strength possessions produce in a full shift. Counting short-handed goals and concluding they are rare is exactly the kind of error that chance-quality measurement exists to catch, and the reasoning is set out at length in the piece on measuring chance quality rather than chance volume.

The second reason is that a short-handed goal does not end the penalty. Scoring while a man down discharges nothing, because the box time was imposed on the other team for its own offence. The side that has just conceded remains a man up, now chasing a goal back on a power play it no longer wants, which is an awkward state to be in and an easy one to play badly.

The third reason is deterrence, and it is probably worth more than the goals. A power play that has been burned once stops pinching. Its defencemen hold the line instead of jumping into the zone, its half-wall player hesitates before the cross-ice pass, and the whole unit becomes measurably more conservative for the rest of the night. None of that appears in any statistic, and all of it makes the next kill easier.

How kill percentage is calculated, and what it hides

Penalty kill percentage is the share of times a team was short-handed in which it did not concede a power play goal. Kill eighty of a hundred and the figure is eighty per cent.

It is a reasonable first number and it misleads in at least five specific ways, all of which follow from the fact that it counts events rather than exposure.

It ignores length. A penalty that begins with fifteen seconds left in a period is counted as one kill, exactly like a full two minutes. A team whose opponents take late penalties will have a flattering number.

It treats a five-on-three like a five-on-four. Two minutes of two-man disadvantage is a far harder task than two minutes of one-man disadvantage, and the measure cannot tell them apart.

It ignores short-handed goals entirely. A kill that concedes nothing and scores twice has the same percentage as a kill that concedes nothing and never touches the puck. The first team gained two goals from being penalised.

It does not care how the kill was survived. A unit that cleared four times and faced three shots scores the same as a unit that was pinned for two minutes and saved by its goaltender. Those are different teams with different futures.

It is sensitive to how often a team is penalised. A disciplined team with a mediocre kill can post a respectable percentage over a small number of penalties, and an undisciplined team with an excellent kill can look worse while conceding fewer power play goals per sixty minutes of actual short-handed time.

• Clears per penalty
• Shots faced per penalty

Kill A, nine of ten43

Kill B, nine of ten19

A constructed example with invented round numbers, not measured data. Both teams killed nine of ten penalties, which is ninety per cent. No real team is being described.

Show the numbers

Invented worked example: two kills with the same percentage

Item	Clears per penalty	Shots faced per penalty
Kill A, nine of ten	4	3
Kill B, nine of ten	1	9

Same percentage, and the second team has been rescued nine times by a goaltender while the first barely needed one. Which of them is the better penalty kill is not a question the percentage can answer, and it is the question that matters for the next ten penalties.

What to watch on the next penalty kill

Five things are visible from the sofa and together they tell you how the two minutes will go.

How far up the ice the kill meets the power play. Four killers strung across the neutral zone are playing to deny the entry, and the first thirty seconds will be a fight before the attacking blue line. Four killers already retreating have conceded the zone and will defend in front of the net.

Whether anybody is standing in the middle of the kill's shape. If a power play has put a player in the interior of the structure with his stick flat, watch whether a killer picks him up. A low killer who takes him permanently means a diamond or a wedge. Nobody taking him means a box, and the whole penalty will be about whether the half-wall can find him.

Whether the triangle in the slot ever moves. Three killers who stay fixed around the slot while a fourth chases the puck are running a wedge plus one, and the point shot is being conceded on purpose. Count how many of those long shots actually reach the goaltender.

Which pass the sticks are taking away. Watch the sticks rather than the bodies. Four sticks laid flat in the same lane means the kill has decided the cross-ice pass is the thing that beats it and has accepted everything else.

How many times the kill touches the puck. This is the real scoreboard. Four touches and four clears is a comfortable penalty however alarming it looked. Zero touches and eleven blocked shots is a kill that survived rather than one that worked, and it will not survive the next one.

Do that for a penalty and the structure stops being four dots on a whiteboard and becomes a set of deliberate concessions, each chosen because the alternative concession was worse. The rest of the hockey explainers here work the same way, starting from what the rule requires before asking what coaches have built on top of it, and the rules governing overtime and the shootout are the natural next step if you want to see a kill asked to do all of this at three skaters against four on open ice.