How to Watch Hockey

Key focus

(This document is about watching rather than playing: where to put your eyes, and how to carry what you see back onto the ice.)

Stop watching the puck and give your eyes one job per period. Watch one player for a whole shift, or the two defencemen, or the weak side, or the five seconds before a goal.

Count, don't guess, to identify a system. Forwards going deep give you the forecheck; whether the weak-side winger follows into the corner gives you the coverage; players at the point give you the power-play shape.

Read a team over five or six sequences before deciding. Systems are coaching choices, and teams change them by score, period and opponent.

Watch every entry at the blue line. Carried in, dumped in or turned over, and where was the defender's gap? Games are decided there and almost nobody watches it.

After a shot, watch the net front, not the replay. The player who stopped just outside the crease, stick down and clear of the goalie, and moved first usually gets the rebound.

Watch live for structure and television for technique. When you are there to learn, sit as high as you can.

Film yourself wide and watch whole shifts, not your touches. The two seconds after your team loses the puck are where the improvement is.

End every session with one thing to try at practice. Watching that changes nothing on the ice was spectating.

Treat any statistic as a rumour until you know its context: denominator, strength state, sample size, linemates, score and whose model it is.

Overview

Almost everything that decides a hockey game happens away from the puck, and the camera that follows the puck trains you to miss it. So give your eyes one narrow job at a time. Trying to watch everything is how you end up watching the puck again.

The players furthest from the puck tell you what a team is doing. A defending forward who follows their man into the corner is marking players; one who stays up near the blue line is marking areas of ice. Counting bodies gives you the forecheck and the power-play shape.

The blue line deserves special attention: public tracking research found that at 5-on-5 carrying the puck in produced roughly twice the unblocked shot attempts and goals that dumping it in did. And the half-second after a shot is when the rebound is decided, which the broadcast often cuts away from.

Live, sit high and the ice becomes a diagram; television shows hands and feet and hides four players. Then film yourself and watch what you do without the puck, and take one thing to practice.

Statistics come last, because they are mostly misread. Shot-attempt counts measure territory, not quality; expected goals prices quality but has no standard definition; and for goalies, goals saved above expected beats save percentage, which beats goals-against average.

---

Part 1: The Core Instruction — Stop Watching the Puck

Why your eyes are trained wrong

Broadcast hockey is made for entertainment, not education. The main camera keeps the puck centred, so the frame holds the puck carrier, the nearest defender and perhaps two others. The other six are off screen. And directors cut to replays at exactly the moments — after a whistle, after a shot — when the tactical information is richest.

The broadcast is built to show you the puck, and the puck is the least informative object on the ice. The puck tells you what happened. The players away from it tell you why.

Puck-following is a reflex: the eye goes to the fastest, highest-contrast object in the frame. Concentrating harder will not fix it. Giving your eyes a different job will.

The four exercises

In short: Practise one at a time: watch one player, then the defencemen, then the weak side, then the five seconds before a goal.

Do these one at a time. Each is deliberately narrow: the point is to build a habit that later runs in the background while you enjoy the game.

Exercise 1 — Watch one player for a complete shift.

Pick a player. Watch only them from the moment they step on to the moment they sit down. Do not look at the puck. Shifts measure roughly 30 to 80 seconds, a well-managed one nearer 30 to 45 (Conditioning and Recovery carries the studies), so this is a short exercise repeated many times.

What you are looking for:

• What they do in the two seconds after their team loses the puck. This moment separates good players from everyone else. Do they stop and watch, or turn and skate somewhere that matters?
• How often their head turns. Coaching resources often suggest a shoulder check every one to two seconds off the puck. The one study that has counted found far fewer. Berg et al. (2025) coded 2,545 puck receptions by Swedish professionals: a mean of 1.48 scans in the five seconds before receiving, and at least one scan in 61% of receptions. On arithmetic the paper itself does not print, that is one scan every 3.4 seconds or so, and no scan at all before 39% of receptions. In fairness to the coaching figure, Berg counted only the five seconds before a reception, not general off-puck play. What separated players was not volume: scanning first was worth 78.8% action success against 70.1%, on the study's own success criteria rather than goals. See Scanning and Anticipation.
• Whether they ever stop moving. Amateurs are often said to glide for about a third of a shift — a coaching observation, not a measurement. Do not assume professionals do not. The one study that counted puts two-foot glide first among NHL forwards' skating characteristics, at 39% or 40% of ice time: Bracko himself has published both figures, and the paywalled 1998 paper that would settle it could not be checked (Skating carries the study and its limits). That glide is a balance position between strides and turns, not a rest. So watch whether the glide is taking the player somewhere.
• How they get off the ice. On the correct side, at a safe moment?

Exercise 2 — Watch the defencemen only.

Watch one team's two defencemen and ignore everything else. It is the fastest way to learn defensive structure, because defencemen are rarely near the puck, so everything they do is positional.

What you are looking for:

• Gap — the distance between the defender and an attacker coming at them. Watch how it closes as the attacker crosses centre. See Defending the Rush.
• Which one goes to the puck and which one goes to the net, on every retrieval.
• Pinching — stepping down from the offensive blue line to keep a puck in the zone (not the same as stepping up on an attacker in the neutral zone). How often, and what has to be true first?
• Where they are on the power play compared with even strength.

Exercise 3 — Watch the weak side.

The strong side is the side the puck is on; the weak side is the other. Watch the half the puck is not on. On a broadcast you will keep losing it off frame — hold in your head what was happening there.

The two labels swap the instant the puck crosses the middle. The picture is borrowed from the Rink Map and drawn in the attacking end, but strong and weak side attach to the puck, not the zone, so they read the same way in the defensive zone.

Diagram from rink map · What the shapes and fills mean

What you are looking for:

• The weak-side winger in the defensive zone. High near the point, low near the net, or between? This one player tells you which coverage the team plays.
• The weak-side defenceman. How far off the post, and facing the puck or their check?
• The backdoor. Many goals are scored at the far post, and nobody watches it until the puck is there. That is a coaching commonplace: this document has no count of goals by post.

Exercise 4 — Watch the five seconds before a goal, not the goal.

When a goal is scored, ignore the celebration and wait for the replay. Ask one question: what happened five seconds earlier that made this possible?

It is usually one of a few things: a lost wall battle, a failed clear, a flat-footed defenceman, a forward who did not get back, a bad change, a screen. If you can name the cause rather than the finish, you are watching properly.

It works in reverse for saves too: what did the defence do in the five seconds before, that meant the shot came from where it did?

Making these stick

• Do one exercise per period, not per game. Trying to watch everything means watching nothing.
• Say it out loud, or write one line. "Their weak-side winger stayed high all night" is a finding. Vague impressions evaporate.
• Expect it to be boring at first. Positional hockey is mostly quiet, and the quiet is what you need to learn.

---

Part 2: Seeing the Systems

Once your eyes leave the puck, you can read what the teams are doing. Almost all of it is visible from a seat.

A caution first. Every system below is a coaching choice, not a rule of hockey. Teams change them by score, period, opponent and personnel — a team that traps with a two-goal lead may forecheck hard when tied. Read a full period before deciding, and expect the answer to change. See Game Management.

Which forecheck are they running? Count the forwards going deep.

Forechecking is the pressure a team applies in the opponent's end when the opponent has the puck. The names count how many forwards go below the top of the faceoff circles.

Sit and count, over five or six sequences:

Two of those systems are drawn in Forechecking Systems, against the same puck carrier: two forwards below the top of the circles in the first, one in the second. (In the diagram key, a circle is a forward, a triangle a defenceman, G the goalie, and the unfilled shapes are the team the picture is drawn for — not the same team in every picture below.)

The next two captions are written for the forechecker, not for you in the stands; their body-checking law belongs to Forechecking Systems. From a seat, read only the count and the depth.

Diagram from forechecking systems · What the shapes and fills mean

Diagram from forechecking systems · What the shapes and fills mean

F1, F2 and F3 are roles defined by order of arrival, not by position. Whoever gets to the puck first is F1, even the left winger. So count bodies and depth, not named positions.

A second tell: watch what F1 does on arrival. A straight line at the carrier's body means the team is hunting turnovers. Angling across to steer the carrier to the boards means containment, waiting for F2 to seal.

Which defensive-zone coverage do they play? Watch the winger.

This is the most reliable read in hockey, and it takes about ninety seconds.

The question: when the puck goes into the corner in their own end, does the defending winger follow their man down into the corner?

• If the winger stays high, on the opposing point just inside the blue line, and never follows a body below the hash marks — they play a zone system, most commonly a low zone collapse. ("High" means on the point, not down at the faceoff dot, a shade under 44 feet away on an NHL sheet; Rink Map has the figures. Confusing the two is how people misread the system.) In a collapse the low three — usually both defencemen and the centre — sink to defend the front of the net, while the wingers guard the points and the weak-side winger drops into the high slot whenever the centre goes low. You defend the area you are standing in. This is the standard coverage assumed throughout. It is often called the most common amateur system because it is easiest to teach, but that is coaching consensus; no survey of amateur systems was found.
• If the winger follows their man everywhere, including into the corner and behind the net — they play man-to-man.
• If the winger follows their man only until the man leaves the winger's area, then hands off — they play a hybrid, sometimes called man-on, zone-off or collapsing man.

The first two are drawn side by side in Defensive Zone Coverage. The moment drawn is the point man walking the puck down rather than a puck in the corner, but the winger is still the one to watch: do they release their man or go with that player?

Diagram from defensive zone coverage · What the shapes and fills mean

Diagram from defensive zone coverage · What the shapes and fills mean

Second tell: watch the strong-side defenceman when the puck is behind their net. In a collapse they hold the net front; in man-to-man they chase behind the goal line. See Defensive Zone Coverage.

Are they trapping? Watch where they stand when they have nothing to do.

A trap is a neutral-zone system that gives up the opponent's end and builds a wall across the middle, so the attack must dump the puck in or lose it trying to carry it through.

The tells:

• A line of three across the neutral zone, usually around their own blue line or the far side of centre — a 1-3-1, one forward pressuring, three across, one defenceman deepest.
• Nobody chases into the corners. A single forechecker applies token pressure and retreats.
• The defenders stop in the neutral zone rather than skating at anyone. Trapping teams look passive on purpose.
• The puck carrier is repeatedly steered towards the boards in the neutral zone and never gets to the middle lane.

The first of those tells is drawn in Neutral Zone Systems.

Diagram from neutral zone systems · What the shapes and fills mean

The pure 1-3-1 trap is far rarer now than its reputation suggests; most teams use trap elements, such as a 1-2-2 with a heavy middle. Watch for the behaviour rather than matching a diagram. See Neutral Zone Systems.

The next two pictures are both a 1-2-2, separated by depth and intent. In the first, the two middle forwards are up near the opposing blue line. In the second, they have dropped to the red line: the carrier is allowed to skate and steered to a wall. What no picture can show is whether the lone forechecker is closing or gliding, and that is what decides which one you are watching.

Diagram from neutral zone systems · What the shapes and fills mean

Diagram from neutral zone systems · What the shapes and fills mean

What is their power-play formation? Count the players at the point.

In short: Count the players at the point to name the power play, and the killers' shape to name the penalty kill.

The point is just inside the blue line. The half-wall is the boards level with the faceoff dot. The bumper stands in the middle between the half-walls, roughly in the high slot.

Count the players standing near the blue line once the power play is set up:

• One at the point — a 1-3-1: one point, two half-walls, a bumper and a net-front player. It is routinely called the dominant modern formation, and first units are widely said to run four forwards. ⚠️ The personnel has been counted; the shape has not. Matt Cane (Hockey Graphs, 2017) measured four-forward units at "roughly 56% of the 5-on-4 ice-time this season" — a league-wide share of ice time from 2016-17, not a count of teams or of first units. No count of which shape teams run was found, so "most teams use a 1-3-1" is coaching consensus. Special Teams sets out the figure with its qualifications.
• Three high — an umbrella: one in the middle at the point and two on the flanks slightly below the middle player, in an arc, with two low. The flanks drop further in the modern version. A true umbrella has no bumper. See Special Teams.
• Four players loaded onto one side of the ice — an overload, working the puck along one wall.

All three are drawn in Special Teams, with the penalty killers left out.

Diagram from special teams · What the shapes and fills mean

Diagram from special teams · What the shapes and fills mean

Diagram from special teams · What the shapes and fills mean

For the penalty kill, count how the four killers are shaped:

• Two high and two low, in a rectangle — a box: forwards high, defencemen low.
• One at the top, two in the middle, one low — a diamond. Teams switch to it against a 1-3-1 because a box leaves the bumper unmarked.
• Three killers in a tight triangle near the net plus one pressuring the puck — a wedge+1. Watch the handover: when the puck moves, the presser stops and the nearest triangle player steps out. One killer chasing the puck all the way round is the shape broken. Teams use the fourth player differently, so expect variation.

All three are drawn in Special Teams against the same 1-3-1. Here the killers are the unfilled team. The third caption is written for the killers, not for you; its shot-blocking technique belongs to Special Teams. From a seat, read the triangle, the fourth player and the handover.

Diagram from special teams · What the shapes and fills mean

Diagram from special teams · What the shapes and fills mean

Diagram from special teams · What the shapes and fills mean

---

Part 3: What to Watch at Each Part of the Game

Breakouts — watch the three supports, not the puck retriever

A breakout is the organised move of the puck out of your own zone. When a defenceman goes back for the puck, look away from them and find the other four skaters.

The five jobs are drawn in Breakouts, frozen at the moment of retrieval with no forecheck shown.

Diagram from breakouts · What the shapes and fills mean

• Where is the strong-side winger? The classic outlet spot is on the wall between the hash marks and the goal line — below the hash marks, not level with them. A winger up near the blue line may be playing a flat, spread or stretch breakout, or may never have come down from point coverage; either way, that pass is long and arrives in their back. A winger behind the goal line or in the corner usually means the breakout is already broken. See Breakouts.
• Where is the centre? Usually curling low-to-high through the middle. A centre standing still is one option fewer.
• Is the weak-side winger heading up ice early? That is a stretch — trading support for speed.
• Does the second defenceman offer a D-to-D pass, or stay glued to the net?

The right spot for that winger moves with the structure. Winger draws the same player at three depths on the wall, one for each of three breakouts. Read it as the places a winger can legitimately be, not a target they are missing.

Diagram from winger · What the shapes and fills mean

Then watch what the retriever chooses: a rim (hard around the boards), a reverse (back against the flow), a D-to-D, a pass to the wall, or a skate-out. Which forechecker forced that choice? See Breakouts.

Entries at the blue line — the most under-watched thirty feet in hockey

Games are decided here and almost nobody watches it. Public tracking work by Eric Tulsky and colleagues (MIT Sloan, 2013) found that at 5-on-5 carrying the puck in is worth roughly twice as much offence as dumping it in. Zone Entries has the figures; note that "shots" there means unblocked attempts.

So when the puck approaches the blue line, watch:

• Did they carry it in, dump it in, or turn it over? Keep a tally for one period. Many entries fail entirely.
• Where is the defender's gap? A defender standing on their own blue line as the carrier arrives has already lost; one who met them at the red line has usually won. Defending the Rush gives rough distances — two to three stick lengths in open neutral ice, about one and a half at the red line, about one at the blue line. Published coaching pages disagree on the last two, and all three are a guide, not a law: a wider gap on a fast attacker is not automatically a mistake.
• On a dump-in, how many attackers chase? A dump with one chaser is a change of possession dressed up as a play.
• Offside needs both skates completely over the blue line before the puck completely crosses it. Watch attacking wingers drag a trailing skate to stay on. In the NHL, IIHF and PWHL a skate lifted over the neutral zone still counts as onside; under USA Hockey and Hockey Canada rules it needs ice contact, so the lifted skate is offside there.

Two pictures, one for each side of the line, drawn from opposite benches. The first is the attack arriving: three lanes, the wide players onside because a skate is still on the line. The second is the defender's gap at the red line, where it is usually won. In the first the unfilled team attacks; in the second it defends.

Diagram from zone entries · What the shapes and fills mean

Diagram from defending the rush · What the shapes and fills mean

See Zone Entries.

Faceoff alignments — read them before the drop

In short: Before each draw, read the four players who are not taking it: their body angles tell you where the puck is meant to go.

There are around 56 faceoffs in an average NHL game (see Faceoffs), so look at the four non-centres on each team before the drop.

• The wingers' body angle tells you the plan. A winger squared to the boards is expecting the puck there. A winger cheating towards the middle is expecting a tie-up and a scrum.
• At an offensive-zone draw, find the shooter. In the common default both defencemen stand at the point; the one on the draw side is the shooter, stick loaded, and the far one holds the line as the safety. Alignments vary, so check where the puck is actually pulled on a clean win. See Faceoffs.
• In the defensive zone, watch the boards-side winger. They must beat the opposing defenceman to a puck that goes back to the point.

Both alignments are drawn in Faceoffs at the right-hand circle, before the drop: the common default, so check it against this team. The defensive-zone caption also explains how the books differ on whether touching the circle line counts — worth knowing before you call an encroachment from a seat.

Diagram from faceoffs · What the shapes and fills mean

Diagram from faceoffs · What the shapes and fills mean

What the rules change at the dot. A centre is usually thrown out of the draw for moving early, bad stick placement, or a teammate encroaching — a winger creeping into the circle costs the centre the draw.

• A second violation by the same team on the same draw is a bench minor for delay of game in the NHL and IIHF (IIHF 76.7), and an individual minor under USA Hockey (613(d)).
• The IIHF first violation changed in 2026/27. The new book ejects the offending draw-taker, as the NHL does (76.6); the 2025/26 book gave a warning instead.
• Batting the puck with a hand to win the draw is a minor in the NHL and IIHF, until a third player has touched it. Closing a hand on the puck is a minor too — a penalty shot if the puck is in the offending team's own crease, or an awarded goal if their net is empty (Rules 67.2, 67.4, 67.5, 63.6).
• USA Hockey splits the act. Catching it in the air and holding it costs no penalty: play stops for a faceoff (618(a)). Covering it on the ice is a delay-of-game minor (614(a)'s Note: "should be penalized promptly"), and picking it up off the ice is a minor (618(a)). In the crease, a penalty shot or optional minor, or an awarded goal if the net is empty and it stopped an obvious goal (618(a), 614(b)).

See Faceoffs.

Line changes — where structure collapses, and nobody is watching

In short: Watch the benches for a full stretch: changes are where structure collapses and where the home coach controls the matchups.

Watch the benches, not the ice, for one full stretch of play. Line changes are where structure collapses.

On the fly, watch for:

• Changing while the puck is in your own end — the classic mistake, and the one that produces odd-man rushes.
• Whether the player coming off reaches the bench before the player going on leaves it. Too many men is almost always a communication failure, not a counting one. See On-Ice Communication.
• Which forward stays out when only two change. Usually the deepest one stays and the two nearest the bench go.

At a stoppage, watch the home bench. Under the NHL's line-change rule (Rule 82) the visitors must put their players out first and then the home team may change, so the home coach gets the last change and controls the matchups. Watch them hold a line at the door until the away team commits.

What the rules change at the bench (NHL Rules 82, 81.4, 87.1 and 63.8 unless stated):

• The sequence. The visitors get up to five seconds to change. The referee then raises a hand, closing their window and opening the home team's eight seconds. The hand drops and changes end. So the raised hand is the moment to look at the home bench.
• Enforcement. A late change at a stoppage, or too many players sent out for it, is sent back to the bench and the coach is warned. For the visiting team, any change attempted after the referee's signal is a bench minor straight away. Any later violation that game, overtime included, is a "Bench Minor Penalty for Delay of Game – Improper Line Change." The linesperson then whistles for everyone to be in position within five seconds. Players slow to the dot, or offside for the draw, are warned once a game — the centre is not thrown out of the draw for it — and after that it is a bench minor.
• The endgame. In the last two minutes of regulation and in overtime, the five-second whistle and the slow-player warning both switch off, and the teams get "a reasonable amount of time" to set up. The long, unhurried set-up you see late in a game is the rule working.
• After an icing, only the team that iced it is frozen. Icing is shooting the puck from your own half across the opponents' goal line untouched (Language and Glossary has the versions). The icing team cannot change, except to replace an injured player, put a pulled goalie back, or after a penalty that changes either team's strength. The other team changes freely, so the home bench keeps last change after a visitor's icing. Watch a tired line trapped in its own end: coaches name it as a reliable way to concede, though that it is a large share of goals is not counted here.
• No time-out after your own icing either (Rule 87.1). The same rule bars it after the goalie stops play on a long shoot-in, and after a defender accidentally dislodges the net. A net knocked off by a defender also freezes that team's bench, with the same exceptions, and the draw comes back to their end at the attacking team's choice of dot (NHL and IIHF Rule 63.8).
• Other books. These bars are the NHL's and the IIHF's. USA Hockey imposes none of them, so expect its benches to behave differently. Hockey Canada's line-change bar is mandatory in U18AAA and Junior and optional in Senior (Rules 6.7(d) and 6.1(g)); that book lets either team take its time-out after an icing, but the icing team still may not change during it.

The moments after a shot — where rebounds are decided

The half-second after a shot is the highest-value moment in the game, and the one broadcasts most often cut away from. Keep watching the live play, not the replay.

• Where is the net-front attacker? Stopped just outside the crease with their stick on the ice, or drifting past?
• Does anyone follow the shot? A point shot with no forward arriving is a free clear for the defence.
• Which team reacts to the rebound first? Rebound goals go to whoever moved before the puck landed.
• Is the far post covered? Under the standard low zone collapse it is the weak-side defenceman's; under man-to-man it belongs to whoever's check went there. The person changes with the system; the requirement does not. Not chasing the puck is a learned skill. See Defensive Zone Coverage.
• On a save, does the goalie freeze it, direct it to the corner, or leave it in the slot? Directing rebounds is a skill, not luck. See Goaltender.

Goaltender draws where a rebound must not land. That caption is written for the goalie; from a seat, watch where the puck ends up and who moved first.

Diagram from goaltender · What the shapes and fills mean

---

Part 4: Live Versus Television

They teach different things. Use both, and know which one you are using.

Live is far better for structure

You see all ten skaters at once. That is the whole advantage. Every read in Part 2 is easy live and hard on TV, because systems are patterns of five players and TV shows you two.

Live also gives you:

• Real speed and distance. Television flattens both. Live, you see how fast a pass arrives and how little time anyone has. See Time and Space.
• Sound. Hard passes, one-timers, players talking, and how physical a battle really was.
• Benches and warm-ups. Line combinations, who talks to whom, and power-play set-ups in warm-ups.

Sit as high as you can. From high up the ice becomes a diagram: spacing, layers and the shape of a forecheck. From the front row you see speed and skill and almost no structure. Take the top row when you are there to learn, and the tenth row when you are there for fun.

Best of all is high behind the net rather than at centre ice, which makes gaps, layers and defensive depth very easy to see.

Television is far better for detail

• Replays. The five seconds before the goal, again, from another angle — Exercise 4.
• Slow motion. For hands on a shot, a goalie's push, or when a defender's hips turn.
• Pause and rewind. Pause as a pass is released and ask what the passer could see.
• Camera angles. Overhead and end-zone cameras sometimes come close to the live view.
• Commentary and telestration — sometimes instructive; treat it as one opinion.

TV's weakness is permanent: the main camera cannot show five players and stay watchable. Some leagues and services offer a wide overhead, tactical or coaching feed (the football idea of an "all-22" camera). Availability varies by league and season. If you can get one, it is worth more than any other resource.

Live versus television: the practical answer

Watch live for systems, watch TV for technique, and if you attend live games, spend at least one full period high up watching nothing but the five players of one team.

---

Part 5: Watching Your Own Team and Yourself

Watching professionals shows you what good looks like. Watching yourself shows you what you look like, and the gap is almost always bigger than you think.

Getting the footage

You do not need a rig. A phone on a tripod at the top of the stands is enough. Frame the whole sheet, or at least a full zone plus the neutral zone. Zooming in destroys the value.

If someone films your team, ask for the widest angle, not the highlights. Highlights are the least useful footage for development.

The protocol for reviewing your own shifts

1. Watch the whole shift, not your touches. They are a far smaller slice than they feel: NHL tracking, which the league calls unofficial, puts puck possession at roughly 3.3% to 4.2% of 5-on-5 ice time — one to two seconds of a forty-second shift. (Playing Without the Puck gives the study and its limits.)
2. Watch each shift twice — once following yourself, once following the play. Were you where the play needed you?
3. Ask three questions per shift, and write the answers down:
   • Where was I when we lost the puck, and where did I go in the next two seconds?
   • Did I ever stop skating?
   • Was I an option? A clean lane, stick on the ice, moving somewhere a pass could reach? See Puck Support and Spacing.
4. Count something. Touches, completed passes, entries with the puck, times back below the puck. Track the same count over three games.
5. Pick exactly one thing to change for the next game. A review that produces six fixes produces none.

The discipline of watching yourself away from the puck

Watching your good passes is enjoyable. Watching yourself stand still on the weak side while a goal goes in is not, which is why almost nobody does it. Do it anyway:

• Watch the three seconds after each of your passes. Did you move, or admire it? Coaches name passing and standing still among the commonest amateur habits — an observation, not a count.
• Watch yourself on every goal against. Not to assign blame, but to see whether you were where your system says.
• Watch your changes. Are you leaving at safe moments? Time three shifts: coaches say most players' shifts run ten to twenty seconds longer than they believe, but that is an observation, not a measurement.
• Watch your head. Count your shoulder checks in a shift. Coaches say amateurs scan far less than they believe; the only published count is of professionals, so check it on your own video.

Watching your own team as a team

• Work out your own team's systems using Part 2, and check with your coach. If your read differs from the coach's answer, that tells you how clearly the system has been communicated.
• Watch your team in the last five minutes of a one-goal game against the first period. If nothing changes, that is a finding. See Game Management.
• Scout your next opponent if you can: which forecheck do they run, and do their wingers follow into the corner?

---

Part 6: Learning by Watching a Position

Spend a period on your own position. Then pick another, because knowing what the player next to you is asked to do makes you easy to play with.

Watching a centre

The centre's job spans all three zones and the full width of the rink, and connects every system — coaching craft, since no measurement of ice coverage by position was found. (Centres are also often said to skate furthest. No study splits centres from wingers, but the forwards-versus-defence comparison points the other way: Lignell et al. (2018), tracking 36 NHL players through one game, found defencemen skated 29% more in total and were on the ice 47% longer, with forwards doing 54% more high-intensity skating per minute. Centre carries the citation.)

Watch for:

• Low support in the defensive zone. In a low zone collapse the centre works between the net front and the corners, never fully committing to either.
• The curl on the breakout. Low swing, receive with speed — and where do they look first?
• The second after the draw. The follow-up is a bigger part of faceoff value than the draw itself.
• The backcheck. The centre is usually the first forward back. Watch the lane, but do not grade it: which backchecker owns which lane is a coaching choice. The default in this guide gives the centre the middle; other teams give it to whoever is first back, or go man-on-man. What every version needs is that somebody owns the middle, so watch whether anybody does. See Centre and Defending the Rush.

The three lanes are drawn in Defending the Rush. The most dangerous attacker is not the carrier but the one driving the middle lane without the puck, who gets missed because the puck is elsewhere — the clearest case of what the camera will not show you.

Diagram from defending the rush · What the shapes and fills mean

See Centre.

Watching a winger

The winger's job is the most system-dependent on the ice, which makes them the best diagnostic.

Watch for:

• How high they stay in the defensive zone, which as above tells you the team's coverage.
• Their spot on the wall in the breakout, and whether they show a stick target or wait.
• After a dump-in, do they become F1, or let the centre chase?
• Do they support the puck or drift wide? Wide wingers look in position and are unavailable for a pass.

See Winger.

Watching a defenceman

Almost everything a defenceman does is positional and happens before the puck arrives.

Watch for:

• Gap control on every rush: the moment they stop skating backwards and start closing.
• Stick position. An active stick in the lane takes away options without contact.
• On a retrieval, do they check over their shoulder before reaching the puck?
• Pinch or retreat. What do they check first? Usually their centre's position and whether someone is high enough to cover.
• How deep they go in the offensive zone. There are three depths. The point, just inside the blue line, is the default: outlet, shooting threat and safety. The top of the circle is the activated position, for a better shot or a walk to the middle, and still recoverable. Below the circles is a commitment; below the hash marks a defenceman is a fourth forward who cannot get back. Watch whether a forward takes the blue line before the defenceman goes down. How deep a team lets them go is a coaching choice, so read what this team allows. See Offensive Zone Play.
• Net-front play. Body position, stick on the opponent's stick, eyes on the puck.

The three depths are drawn in Defender, with the band below the circles left empty on purpose. They are places to check, not a ranking: which one this team allows is what you are watching for.

Diagram from defender · What the shapes and fills mean

See Defender.

Watching a goalie

In short: Watch what the goalie does before the shot: depth, tracking, being set at release, and where the rebounds go.

Goaltending is the hardest position to watch, because the interesting part happens before the shot.

Watch for:

• Depth. How far out do they start, and how does it change with the shooter's distance?
• Tracking. Head still, eyes locked on the puck — clear on replays.
• Movement. Shuffles for small moves, T-pushes for larger ones — and are they set when the shot is released? A goalie still moving at release cannot push anywhere, which is why coaches rate being set above being perfectly deep (a coaching judgement).
• Rebound control. Into the slot, or to the corners, or frozen?
• Puck-handling behind the net. Does the goalie come out to play the puck, and where? Whether that area is restricted depends on the book — see below. Even where it is not, a goalie who leaves the crease and falls on the puck can be given a minor.

The trapezoid is drawn in Rink Map, whose caption carries the NHL dimensions.

Diagram from rink map · What the shapes and fills mean

What the books say about the trapezoid.

• NHL and IIHF. The trapezoid (NHL Rule 27.8; IIHF Rule 27.7) limits where a goalie may play the puck behind the goal line, unless the goalie keeps skate contact with the crease. The KHL is commonly said to have it too, but no KHL rulebook was available to check.
• USA Hockey, Hockey Canada and CARHA have none. Neither "trapezoid", "restricted area" nor "designated area" appears in USA Hockey's rules or Casebook, and Hockey Canada's only "designated area" is a Goal Judge's spot. How much rec, beer-league and youth hockey that covers is a question about leagues, and this document has no count.
• The PWHL contradicts itself. It switches the trapezoid off in Rules 1.8 and 27.8, yet Rule 65.2(viii) still writes a minor for a goalie who plays the puck "outside of the designated area behind the net".
• No trapezoid is not no restriction. In all three trapezoid-free books a goalie may fetch the puck from the corner but may not sit on it. USA Hockey Rule 614(c) makes it a delay-of-game minor for a goalie who falls on or gathers the puck behind the goal line with their body entirely outside the crease, or with the puck outside their "privileged" area — drawn from the end-zone faceoff spots, so the corners are outside it. Even inside it, freezing the puck is a minor if the goalie had time to play it with the stick before being pressured, or held it against the frame or boards, or dropped it on the back of the net. Hockey Canada (Interpretation 3 to Rule 10.1(a)): "When a goaltender leaves their crease, they will not be allowed to freeze the puck. If they do, they will be assessed a Minor penalty. No warning will be issued." CARHA Rule 58(b) is a minor for a goalie who falls on, gathers or pins the puck behind the goal line with their body entirely outside the crease.
• 🇬🇧 In Britain, watch it as though it binds. The Elite League's Casebook falls back on the 2026/27 IIHF book, which carries it at Rule 27.7. England Ice Hockey's Rules & Regulations 22.3 (2024-25) call it "not currently enforced", but neither the 2025-26 nor the 2026-27 In-House Rules mention it, and they enforce the IIHF book "Unless otherwise stated" — so whether that suspension still stands is unsettled. Ask your league, and look for painted lines behind the net. An unmarked sheet is a hint, not a ruling: IIHF 27.7 turns on "the position of the puck", not the markings.

See Goaltender.

---

Part 7: Statistics Literacy

Numbers let you watch hockey when you cannot be at the rink. They are also the fastest way to be confidently wrong, because most hockey statistics measure something slightly different from what their name implies.

A blunt warning first. Public hockey analytics is a contested, evolving field, not settled science. Sites compute the same metric differently, models get rebuilt, the event data has real errors — arena scorekeepers record shots and hits differently — and analysts disagree. Treat every number as evidence, not a verdict, and check the definition on the site you are reading.

Statistics literacy: the basic statistics

In short: Read every basic number with its limits: time on ice is arguably the most honest, and plus/minus alone tells you little.

Goals. Credited to the player who put the puck in; only one point can go to any one player on a goal (NHL Rule 78.2).

Assists. Up to two players who touched the puck before the scorer, provided no defender played or controlled it in between (Rule 78.3).

• Good for: identifying who is involved in offence.
• Limits: the two-assist cap is arbitrary, so the pass that broke the defence often goes uncredited. Assists are also a judgement by the in-arena scorer, whose decision Rule 78.1 makes final — though after the referee signs the game report, the League's Chief Statistician can still correct it.

Points. Goals plus assists. The default measure of offensive production.

• Limits: ignores ice time, linemates, power-play usage and defence. Five extra minutes a night with better teammates will out-point a better player.

Plus/minus. A player's on-ice goal differential. On an even-strength or shorthanded goal, every skater on the ice for the scoring team gets a plus and every skater for the conceding team a minus. Power-play goals count for neither side. Empty-net goals count as even strength unless the scorers are on the power play; penalty-shot goals are excluded. The Montreal Canadiens used it internally before the league adopted it, and the adoption date is contested — 1959-60 in the Wikipedia article cited below, 1967-68 elsewhere, with no primary NHL source to settle it.

• Why it is criticised: every goal weighs the same; it depends heavily on teammates, goalie and opposition; it ignores power-play play but counts shorthanded play; it swings with shooting and save-percentage luck; and it cannot tell the player who caused a goal from one who was merely on the ice. A fourth-liner on a great team can beat a first-liner carrying a bad one.
• Use it for: essentially nothing on its own. For on-ice goal differential, use a rate-based, situation-filtered version.

Penalty minutes (PIM). Total minutes of penalties assessed.

• Limits: lumps together stick fouls that give the other team a power play and fighting majors, which usually do not, because both fighters go off. Without the breakdown it tells you almost nothing. See Risk Management.

Time on ice (TOI). Total minutes played, usually split by strength state (even strength, power play, shorthanded).

• Good for: arguably the most honest number in hockey, because it is the coach's own opinion of the player. Rising TOI means trust.
• Essential as a denominator. Raw totals for a 20-minute player and an 11-minute player are not comparable; convert to per-60 rates.

Faceoff percentage. Draws won divided by draws taken.

• Good for: comparing centres against each other on a single, well-defined task.
• Limits: the spread is narrower than people assume — under nine percentage points between the best and worst NHL teams, much wider between individual centres (Faceoffs explains why the two are not comparable). A "win" to an empty spot counts the same as one that starts a breakout, and a shorthanded defensive-zone draw is a different job from a power-play offensive one.

Shooting percentage. Goals divided by shots on goal. In the 2025-26 NHL, teams scored on about 11.1% of shots on goal in all situations (about 27.8 shots and 3.08 goals per team per game).

• Good for: identifying finishing ability over very large samples.
• Limits: very noisy over short samples. A player scoring on 25% of shots over twenty games is a normal shooter having a good month. Real shooting talent takes several hundred shots to show.

Save percentage (SV%). Saves divided by shots on goal faced. North America writes it as a decimal (.920 means 92.0%); IIHF competition writes a true percentage.

• Good for: better than goals-against average, because it accounts for workload.
• Limits: every shot counts the same, so a point shot and a backdoor tap-in are equal. Expected-goals models exist to fix that.

Goals against average (GAA). Goals allowed per 60 minutes. Overtime counts; empty-net and shootout goals do not.

• Limits: depends heavily on the team in front. A goalie behind a strong defence posts a good GAA facing easy shots.

Possession statistics: Corsi and Fenwick

Corsi counts all shot attempts: shots on goal, plus shots that missed the net, plus blocked shots. Fenwick counts unblocked shot attempts: shots on goal plus missed shots, excluding blocks.

They are normally reported as differentials or as percentages:

• CF = Corsi For (your team's attempts while you are on the ice). CA = Corsi Against.
• CF% = CF ÷ (CF + CA). 55% means your team took 55% of all attempts while you were on the ice; 50% is break-even. Most players sit between about 40% and 60%, and above about 55% is often called elite.
• FF% is the same calculation using unblocked attempts.

The names are not acronyms. Corsi was named by Tim Barnes, a Chicago financial analyst writing as "Vic Ferrari", after Jim Corsi, Buffalo's goaltending coach from 2001 to 2014 — chosen, after two Sabres staff names were discarded, from a picture of him whose moustache Barnes liked. Fenwick is named after Matt Fenwick, an Alberta blogger and Flames fan, who in November 2007 proposed excluding blocked shots because a blocked shot is not a genuine chance, or on average a worse one.

• Why they became popular: they proxy for possession — you cannot shoot without the puck. And shot attempts pile up roughly eighteen times faster than goals (an approximation from recent rates, not a published constant; the common "twenty times" rounds it), so they become meaningful within a handful of games where goals take most of a season. Over short samples, attempt differential predicted future results better than past goal differential did.
• Limits: a harmless point shot equals a slot chance; score effects and zone starts distort them; linemates drive them; and they undervalue elite finishers. They are a team measure attributed to individuals.

Treat them as a rough measure of territory: superseded by expected goals for most purposes, but simple, transparent and available everywhere.

Expected goals (xG)

An expected goals model gives each unblocked shot attempt a probability of becoming a goal. Summed, "2.7 expected goals" means those chances would produce 2.7 goals on average.

"Unblocked" matters. MoneyPuck defines xGoals as "The chance of an unblocked shot attempt being a goal" and states that "blocked shot attempts are valued at 0 xGoals." So xG is built on Fenwick's events, not Corsi's: it fixes Corsi's flaw by pricing unblocked attempts and dropping blocked ones entirely. Other providers may differ; check the glossary.

Typical inputs are distance, angle, shot type, rebound, rush, the previous event and its timing, and strength state. MoneyPuck's public model uses around fifteen such variables, trained on hundreds of thousands of NHL shots with gradient boosting; others use logistic regression or other methods.

• Why it improved on raw shot counts: a slot shot and a point shot are no longer the same event, and it still accumulates fast. The trade is that blocked attempts drop out.
• How models differ: substantially. Different events, inputs, algorithms and training data mean two sites will give different xG for the same game; there is no standard xG. Most public models use only league event data, which does not know where the goalie or defenders were, or whether the shooter was screened.
• Other limits: single-game xG is volatile and no verdict on who deserved to win; player over- or under-performance against xG is noisy; and recorded shot locations have known biases.

Rule of thumb: use xG over Corsi, use several games rather than one, and never quote an xG figure without knowing which model produced it.

Context statistics — why raw numbers mislead

In short: Before trusting a number, check zone starts, teammates and the score, and never compare zone-start figures across sites.

Three contexts change what a number means.

Zone starts. Which zone a player's shifts begin in, usually an offensive-zone start percentage. Coaches start offensive players in the attacking zone and defensive specialists in their own end, so a player buried with defensive-zone starts will have worse possession numbers than an identical player who is not.

Check the denominator first, because sites define it very differently. Evolving Hockey publishes two versions: one counts on-the-fly changes as zone starts and divides by every shift, so nobody's figure is large; the other counts faceoffs only. A general-reference formula uses offensive plus defensive zone starts alone, and scores the same player far higher. So read the site's glossary, and never compare figures across sites. Game Management sets out the definitions and their sources.

Two caveats. First, the effect is smaller than early analysis assumed, because most shifts do not start with a faceoff. Micah Blake McCurdy (HockeyViz, 2015) sorted roughly a million 2014-15 NHL shifts and found 57% started on the fly; being slightly longer, they made up 58% of all ice time. That is one analyst's count of one season on his own site, now reachable only through the Internet Archive.

You can check the direction yourself. Five skaters for sixty minutes is 18,000 skater-seconds a side; at a mean shift of 47.8 seconds that is about 375 shifts a side. With about 56 draws a game, each starting at most five shifts, faceoffs can start at most 280. So even changing all five at every draw, a quarter of shifts would begin in open play. (Draws per game come from Faceoffs. The 47.8 s is not a published statistic but a mean computed from the NHL's live time-on-ice report for 2024-25, all situations, run on 2 September 2026; Conditioning and Recovery sets out the method.)

Second, offensive-zone draws are only about a third of all faceoffs (Faceoffs has the split). A coach sheltering a player is redistributing within that third across a whole bench, so the effect on one player's season is modest.

Quality of competition (QoC) and quality of teammates (QoT). Averages of the opponents' or teammates' metrics, weighted by time shared with each (Evolving Hockey's description).

• Limits: over a season, QoC compresses because everyone plays everyone; it matters more in short samples and playoff series. Quality of teammates is usually the bigger effect, and the one people forget.

Score effects. A trailing team takes more risk and more shots; a leading team sits back and concedes them. It is a large, well-documented distortion, so a team that leads all season looks worse by shot attempts than it is. Analysts use score-adjusted metrics; Evolving Hockey publishes weights by score state and home or away, using Micah Blake McCurdy's method, as its glossary says.

Raw possession numbers from lopsided games mostly read the scoreboard back to you. See Game Management.

Goaltending: save percentage versus goals saved above expected

Save percentage treats every shot as identical. Goals saved above expected (GSAx) does not.

GSAx is expected goals against minus actual goals allowed. MoneyPuck: "A positive number means the goalie is stopping more goals than an average goalie would." +8 over a season means about eight goals fewer than an average goalie facing the same shots.

• Why it is better: it separates the goalie from the team. Thirty easy point shots a night and twenty slot chances can produce the same save percentage. GSAx asks: given the shots you faced, were you better or worse than average?
• Related measures: expected save percentage makes the same comparison as a percentage. The older goals saved above average (GSAA) ignores shot quality and is weaker.
• Limits: it inherits the xG model's weaknesses — public models miss screens, unrecorded deflections and the goalie's position — and needs a large sample, because goaltending is the noisiest position.

Use GSAx over save percentage, and use save percentage over goals against average, but do not treat any of them as final.

PDO — the luck detector

PDO is on-ice shooting percentage plus on-ice save percentage (the NHL calls it SPSV%), shown on a 100 scale or around 1.000. The name is the online handle of Brian King, who proposed it.

Across the league the two sum to almost exactly 100, and neither repeats well for individuals, so PDO regresses hard towards 100. Above about 102, a team is probably not as good as its results; below about 98, probably better.

• Good for: a sanity check. A team winning while outplayed on every measure with a PDO of 103 has its explanation, and its results will probably fall off.
• Limits: not purely luck. Elite finishers and elite goalies genuinely sustain it, so a team with both can run above 100 for years. It is a flag to investigate, not a conclusion.

A checklist for reading any hockey statistic

Before you believe a number, ask:

1. What is the denominator? Per game, per 60 minutes, or a raw total? Raw totals mostly measure ice time.
2. Which strength state? All situations, or 5-on-5 only? Power-play production hides inside "all situations" totals.
3. How big is the sample? Goals, shooting percentage and save percentage need enormous samples. Shot attempts need fewer.
4. Who was on the ice with them? Quality of teammates is usually the largest hidden variable.
5. What was the score? Score effects distort every possession measure.
6. Whose model is it? For anything expected-goals-based, the provider matters.
7. Does it pass the eye test, and if not, which one is wrong? Sometimes the number. Sometimes you. The interesting cases are the ones where they disagree — that disagreement is where you learn something.

---

Part 8: Where to Find Good Analysis

Sites appear, change hands, go behind paywalls and disappear, so learn the categories rather than a link list. Anything named here may have changed since this was written — check before relying on it.

Public analytics sites. Free or cheap sites publishing possession, expected goals and goaltending numbers, usually with a glossary. Natural Stat Trick, MoneyPuck, Evolving Hockey and Hockey Reference are long-running examples. Read the glossary first: definitions differ, and knowing which one you are reading is the whole skill.

Coaching video breakdowns. Coaches and analysts drawing on clips to explain what each player was doing and why — the fastest way to turn watching into understanding. The Coaches Site, Ice Hockey Systems and How To Hockey are established examples, and many individual coaches publish good short breakdowns.

Wide-angle footage. Any feed that keeps all ten skaters in frame. Some leagues sell tactical angles, and some junior and development teams stream from one fixed high camera, which is ideal. Your own team's film, shot wide from the top of the stands, belongs here and is worth more to you than any professional footage.

Original research. Conference papers and research blogs — the MIT Sloan Sports Analytics Conference has published much hockey work, including the zone-entry research above. Read these to learn why a metric exists.

Your own coach. Underrated and free. Ask "what forecheck are we running and when do we change it?" and "in the d-zone, do you want my winger following into the corner?"

How to judge a source. Good ones publish their definitions, say what their model does not know, and disagree in public. Be sceptical of anyone who treats one number as a final verdict, and of anyone who dismisses all analytics. Both want the answer to be simpler than it is.

---

Part 9: A Practical Weekly Routine

One deliberate hour a week will do more for your hockey brain than several hours of ordinary spectating — a coaching judgement, not a measured ratio. A realistic split:

20 minutes — one period of professional hockey, one exercise. One exercise from Part 1, rotated weekly: one player, then defencemen only, then the weak side, then the five seconds before each goal. Live if you can, sitting high.

15 minutes — your own footage. Three or four of your shifts, each watched twice, with the three questions from Part 5 written down. No footage? A phone at the top of the stands is enough; ask a parent or spare player to film one period.

10 minutes — one systems read. One team, professional or your next opponent. Two written answers: which forecheck, and do their wingers follow into the corner?

10 minutes — one statistic, properly understood. One metric from Part 7. Read its definition on the site that publishes it, then check its values for players you know well. The disagreements teach you most.

5 minutes — write down one thing to try. One sentence, such as "This week I look over my shoulder before every retrieval." Take it to practice. See Practice and Development.

A retrieval puck is often on the boards, so meet them with skates parallel, forearm and hip to the wall, head up and chin off your chest; never turn your back to anyone coming at you and never duck, even going slowly. If you are already facing the boards and cannot turn, get your stick and gloves up and spread on the glass, head off it. And if someone gets there first, play the puck, not them: whether you may hit at all is your league's call, never hit a player who cannot see you coming, never from behind, and if you can see their numbers, pull up. See Body Contact and Battles.

Only half an hour? Cut the professional hockey and keep your own footage. That is where the improvement is.

---

Common Mistakes

• Watching the puck and calling it studying. Pick one job before the drop — one player, the defencemen, the weak side, or the five seconds before each goal — and hold it for the period.
• Trying to watch everything at once. Do one exercise per period, or you retain none of it.
• Only watching the highlights. Watch whole shifts. When a highlight comes up, rewind to the pass before the pass and to whoever got open without the puck.
• Watching only players who play like you. Spend a period on a player in your position whose choices puzzle you, then one on the position beside yours.
• Sitting as close to the ice as possible. When you are there to learn, sit high, ideally behind a net. The front row is for atmosphere.
• Grading a player against a system you have assumed. Work out the system first, then judge the player. A centre coming back along the boards, a winger holding high, or a defenceman who never goes below the circles may each be doing what their team asks.
• Looking up at the replay the moment a shot is taken. Keep your eyes on the net front. In your own game, be that player: arrive under control, stop just outside the crease, stick on the ice and clear of the goalie, head up and knowing where the post is, first to the puck when it comes out. Never go into the goalie; contact with one can draw a penalty, or get you thrown out.
• Assuming what a professional team does is what your team wants. Ask. Almost every "always do X" on TV is "in this team's system, in this score state, do X."
• Skipping to your own touches on video. Watch each shift from step-on to sit-down. Your touches are only about 3.3% to 4.2% of your 5-on-5 ice time, by NHL tracking the league calls unofficial.
• Watching without taking anything to practice. End every session with one sentence: the one thing you will try next.
• Quoting a statistic without knowing its definition. Look it up on the site that publishes it first. Sites compute things differently.
• Treating plus/minus as a defensive measure. Watch the defender instead: the gap, the stick, and where they were before each goal against.
• Reading one game's expected goals as a verdict. Use several games and know whose model it is.
• Believing a hot stretch is a new level. If shooting or save percentage jumped and possession did not, check PDO.
• Treating analytics as gospel or garbage. Let a number tell you what to watch, then let the watching decide what it means.

Check yourself

Answer each question before you go on, then check it in the section named. Getting an answer wrong and then checking it is worth more than recognising a right one.

1. You have one period and you want to know which defensive-zone coverage a team plays. Which single player do you watch, and what are the three things they might do? (§Which defensive-zone coverage do they play? Watch the winger)

2. A goal goes in and everyone is watching the celebration. What should you be watching instead, and what one question should you be answering? (§The four exercises, Exercise 4)

3. You can have a tenth-row seat or the top row. Which do you take when you are there to learn, why — and where is the best seat of all? (§Live is far better for structure)

4. Somebody quotes a player's plus/minus as evidence they are defensively poor. Give two reasons that is not what the number measures. (§Statistics literacy: the basic statistics — Plus/minus)

5. What does expected goals price that Corsi does not — and what does xG discard that Corsi keeps? Why can two sites give different xG for the same game? (§Possession statistics: Corsi and Fenwick, §Expected goals (xG))

6. A team is winning while being outplayed on every possession and expected-goals measure. Which single number do you check, what would it have to say, and why is calling that number "luck" a simplification? (§PDO — the luck detector)

7. You are reviewing your own footage and you want to skip to your touches. Roughly what share of your ice time are they, on what basis is that share measured — and what are the three questions to ask of each shift instead? (§The protocol for reviewing your own shifts)

8. Two goalies post identical save percentages. Which measure would tell you which of them actually played better, and what does that measure not know? (§Goaltending: save percentage versus goals saved above expected)

Key Takeaways

1. Stop watching the puck, and give your eyes one job per period: one player, the defencemen, the weak side, or the five seconds before each goal.
2. To read defensive-zone coverage, watch the defending winger when the puck goes into the corner. Follows their man: man-to-man. Stays high on the point: zone. Hands off: hybrid. Read five or six sequences, because teams change it.
3. Count to identify the rest. Forwards going deep give the forecheck; players near the blue line give the power-play shape. F1, F2 and F3 are order of arrival, not position.
4. Watch live for systems and TV for technique. To learn, sit as high as you can, ideally behind a net.
5. Watch every blue-line entry. Carried, dumped or turned over? Where was the defender's gap? At 5-on-5, carrying it in produced roughly twice the offence of dumping it.
6. On a breakout, look away from the retriever and find the other four. Then ask of your own game: when your defenceman goes back for the puck, are you a pass they can make?
7. After a shot, watch the net front, not the replay. In your own game, get there under control, stop just outside the crease, stick down and clear of the goalie, head up and knowing where the post is, and be first to the puck. Never go into the goalie.
8. Watch your whole shift, not your touches. The two seconds after your team loses the puck are what separate good players.
9. Watch your own position for a period, then the one beside it. What a team asks of each player is a coaching choice, so learn what this team wants before you judge.
10. Convert every total to a per-60 rate before comparing players. Time on ice is the coach's own opinion of a player, and the denominator every other number needs.
11. Never judge a defender by plus/minus. It depends on teammates and goaltending, ignores power-play goals, and cannot tell who caused a goal.
12. Prefer expected goals to Corsi, and check whose model it is. Two sites will give different xG for the same game.
13. Judge goalies by GSAx, then save percentage, then goals-against average. Before believing a hot streak, check PDO, and treat it as a flag, not a verdict.
14. Give every number its context: denominator, strength state, sample, teammates, score, model, and the eye test.
15. Spend a deliberate hour a week, and end it with one thing to try at practice.

Sources — retrieved 27 July 2026

Rules (primary): NHL Official Rules 2025-2026 (PDF) — Rule 78 (Goals and Assists; 78.1 scorer's authority and the League's Chief Statistician exception to the signed game report, 78.2 crediting goals, 78.3 maximum two assists), Rule 82 (Line Changes; 82.1 visiting team first / home team last change except following an icing of its own, and the three exceptions to the post-icing substitution ban; 82.2 the line-change procedure — eight sub-clauses, (i) to (viii); the five- and eight-second windows and the referee's raised hand at step (ii) are the first four, and the other four are (v) the improper-change warning, (vi) the linesperson's five-second whistle, (vii) the once-in-the-game warning for players "slow (after the five-second warning whistle given by the Linesperson) getting to the face-off location", in which "the offending team's center is not ejected from the face-off", and (viii) the suspension of both (vi) and (vii) in the last two minutes of regulation and at any time in overtime), Rule 81.4 (no line change after your own icing), Rule 63.8 (the same bar after a net "displaced accidentally by a defending skater or goalkeeper", with the draw in that team's own end at the attacking team's choice of dot), Rule 27.8 (goalkeeper's restricted area / trapezoid), Rule 76 (face-offs and face-off violations), Rule 67 (Handling Puck — 67.2 and the note to 67.2(ii), 67.4 Penalty Shot, 67.5 Awarded Goal) and Rule 63.6 (the puck-location test for that penalty shot) · USA Hockey Official Playing Rules 2025-29 (PDF) — Rule 630(a), which makes a player offside when they do not "have skate contact with any part of the Neutral Zone or the blue line", and Rules 618(a) and 614(b), for the closed hand being a stoppage and a last-play faceoff outside the crease and a penalty shot or optional minor inside it, and Rule 614(a), the delay-of-game minor for a player who "deliberately falls on or gathers the puck into their body", with its Note that "the use of hands to make the puck unplayable should be penalized promptly", and 618(a)'s delay-of-game minor for a player who "picks the puck up from the ice with their hand(s) while play is in progress", and Rule 614(c), whose trigger (1) is a delay-of-game minor for a goalkeeper who falls on or gathers the puck where "⁠[t]he puck is behind the goal line and their body is entirely outside of the goal crease". The playing rules were searched in full for trapezoid, restricted area and designated area: no hits. · the same three searched in the separate USA Hockey Official Rules and Casebook 2025-29 (PDF), also with no hits, and whose Rule 614 Situation 1 — "A defending player slides and covers the puck with their hand" outside the crease — answers "The Referee must assess a minor penalty" · Hockey Canada Playing Rules 2026-2028 (PDF) — Rule 6.11 Off-side, whose preamble counts only "the player's skate(s) that are in physical contact with the ice surface". Those two are the books in which a lifted trail skate is offside. Also Rules 6.7(d) and 6.1(g), the no-change bars after an icing and after a dislodged net, both scoped as 6.7(d) words it, "In categories of U18AAA, Junior, and at the option of the Member in Senior" — mandatory in the first two, optional only in Senior — and 6.7(d)(iii), under which "Either team is permitted to take their time-out … following an icing" but the icing team may not substitute during it. Also Interpretation 3 to Rule 10.1(a), clause v: "When a goaltender leaves their crease, they will not be allowed to freeze the puck. If they do, they will be assessed a Minor penalty. No warning will be issued." The book was searched for a goalkeeper's restricted area and has none — its only "designated area" is the one Rule 5.4(b) stations a Goal Judge in. · IIHF Official Rule Book 2025/26 (PDF) — Rule 76.6 Face-off Procedure – Other Players and Rule 76.7 Face-off Procedure – Violations, the two the faceoff-ejection note above rests on, and Rule 67.2(II), whose crease escalation is the NHL's in its own words: "When this is done in their team's goal crease area, a 'Penalty Shot' shall be assessed Rule 67.4 – Penalty Shot, or Rule 67.5 – Goal Awarded" · IIHF Official Rule Book 2026/27 (Version 1.0, June 2026) (PDF) — Rules 76.6 and 76.7, read on 29 August 2026 against the 2025/26 edition above. 76.6 now has the offending team's draw-taker "ejected from the face-off circle" where the older book gave the team a warning; 76.7's bench minor for a second violation by the same team on the same draw is unchanged. Only those two rules were compared, so every other IIHF citation in this document is the 2025/26 edition's and has not been re-read in the newer book.

Women's professional, read in the primary text on 26 September 2026: PWHL Official Rule Book 2025-2026 (PDF) — Rule 85.1 (Off-side), which carries the NHL's vertical-plane test word for word: "If a player’s skate has yet to break the “plane” prior to the puck completely crossing the leading edge, she is deemed to be on–side for the purpose of the off–side rule." ⚠️ The book renumbers from Rule 52 onward, so the NHL's 83.1 will not find it.

Rules (Britain — the trapezoid's status): IHUK In-House Rules 2026-27, Revision 1 (published 25 July 2026) (PDF) — the current British document, adopted by IHUK and the British Universities Ice Hockey Association for "England, Wales, Scotland and Northern Ireland (the Home Countries)" · IHUK In-House Rules 2025-26, Revision 1 (published 27/10/25) (PDF) — superseded, and scoped only to England and Scotland. Both editions were searched for the goalkeeper's restricted area on 30 August 2026 and neither mentions it — neither amends Rule 27 at all, and both close their Introduction "Unless otherwise stated, all rules will be enforced in line with the IIHF Rule Book." · England Ice Hockey Rules & Regulations 2024-2025 (PDF) — Rule 22.3, which says of the goalkeepers' restricted area that it is "not currently enforced to allow ice rinks the necessary time to make the changes to implement this rule", read with Rule 1.1, which makes England Ice Hockey the National Governing Body "in England and Wales" and so sets how far 22.3 reaches. Whether that suspension still stands is not resolved by anything published, and no IHUK Rule Bulletin has been obtained here.

Statistical definitions: MoneyPuck – Glossary (Corsi, Fenwick, expected goals, goals saved above expected, expected save percentage, high-danger shots, flurry adjustment) · MoneyPuck – About / model description (xG model inputs and training data) · Wikipedia – Corsi (statistic) — read 2 September 2026; source for the naming story, the discarded Regier and Ruff names, and the moustache, but it says nothing about what job Jim Corsi held · Wikipedia – Jim Corsi (ice hockey) — source for the Buffalo Sabres goaltending-coach tenure, 2001–2014 · Wikipedia – Fenwick (statistic) — read 2 September 2026; "a blogger from Alberta and fan of the Calgary Flames", and the November 2007 post · Wikipedia – Analytics (ice hockey) (PDO / SPSV%, CF% ranges, zone starts, score effects) — read 2 September 2026; the third of the zone-start formulas given above, "offensive zone starts / (offensive + defensive zone starts)", is this article's, which attributes it in turn to an NHL club piece not obtained here · Evolving Hockey – Standard Skater Tables glossary — rendered page read 2 September 2026; source for the other two zone-start definitions, which it prints under separate names on the same table, and for its on-the-fly category of zone start. Only two published definitions of a zone-start percentage were obtained. Natural Stat Trick's glossary returned HTTP 403 to every attempt on 2 September 2026 and was not read, and MoneyPuck's and Hockey Reference's were not checked, so the paragraph above is evidence that at least two published definitions differ widely rather than a survey of what every site does. The definitions are carried in full, with their wording, in Game Management · Wikipedia – Plus-minus (ice hockey) · Wikipedia – Save percentage · Wikipedia – Goals against average · Wikipedia – Expected goals (model construction, provider variation, limitations) · Evolving Hockey – Quality of Teammate and Quality of Competition · Evolving Hockey – Score Adjustments

Power-play personnel: Matt Cane — "Second Units and Zone Entries: Why teams should go all-in on the 4 forward power play", Hockey Graphs, 7 March 2017 — refetched at that canonical address on 2 September 2026, HTTP 200; source for the four-forward share of 5-on-4 ice time, quoted verbatim. It measures ice time across the league, not a count of teams, and it is 2016-17 data; the 56% share is not a first-unit figure at all. The first-unit point rests on Cane's separate finding that teams were "1.2 times as likely to use a 3 forward unit on shifts that did not start a power play", and it is that finding his footnote qualifies: "because we're measuring only when the shift starts and not which units are on the ice we may be understating the difference in 4 forward usage between 1st and 2nd units." It counts nothing about the shape a unit plays. The figure is owned by Special Teams, with its qualifications; this document recaps it.

Zone entry research: the Tulsky et al. (MIT Sloan 2013) per-entry figures are owned by Zone Entries, which carries the full citation, the sample, and the definition of "shots" used there. This document only recaps the headline.

Scanning research: Berg, Malmsten, Lind, Mannix, Sjösten, Josefsson, Gredin, Stenling & Ivarsson (2025) — Scanning is associated with better performance in professional ice hockey, Journal of Sports Sciences 43(2):145–150, DOI 10.1080/02640414.2024.2433899 — open access; 2,545 puck receptions by 88 SHL and SDHL professionals, mean 1.48 scans (SD 1.01) in the five seconds before receiving, at least one scan in 61% of receptions, and 78.8% vs 70.1% action success on the study's own criteria, "formulated in consultation with SIHA" (the Swedish Ice Hockey Association). Source for the scanning-rate expectation in Exercise 1. Verification note: the paper prints no per-second rate and no no-scan share — the one scan per 3.4 seconds and the 39% used in Exercise 1 are arithmetic on its published 1.48 and 61%, and are labelled as such where they appear; discussed in full in Scanning and Anticipation.

League figures: the 2025-26 NHL regular-season numbers quoted in the body above (3.08 goals for per team per game, 27.8 shots on goal per team per game, 11.1% team shooting percentage) were computed from the NHL's public team-summary data for the 2025-26 regular season, available via NHL Stats. Two further figures were computed from the same data and are recorded here rather than quoted in the body: a power-play percentage mean of 21.1% and a penalty-kill mean of 78.9%. Each is the unweighted mean of the 32 clubs' own percentages — every club counted once, whatever its number of opportunities. That is not the same statistic as the 21.1% in Risk Management, which owns power-play conversion for this guide: that one is the league aggregate, 1,595 power-play goals on 7,555 opportunities, weighting every power play equally instead of every club. Recomputed 22 September 2026, the two come to 21.10% as a mean of clubs against 21.11% as a league aggregate: they round alike this season by coincidence, not by construction. Say which one you are using, and use the aggregate when the question is what a single power play is worth. These are team-level, all-situations figures and will change every season — check current values rather than quoting these. No faceoff statistic is computed in this document. Faceoffs per game, the team faceoff win range, the skater-level range and the offensive/defensive/neutral zone split all live in Faceoffs, which covers every faceoff figure along with its qualifications. The zone split and the team win range were computed from the same NHL public team-summary and team-faceoff data and live in that document, where their qualifiers are recorded in full.

Shift starts: Micah Blake McCurdy — "Shift Starts and Ends, Part 1", HockeyViz, 27 August 2015 — read through the Internet Archive on 2 September 2026, the live hockeyviz.com page having gone. Source for the on-the-fly share, which it gives as "In 2014-2015 there were roughly a million shifts, of which 57% were started on-the-fly. The on-the-fly shifts are very slightly longer, accounting for 58% of all hockey that was played last year." It is one analyst's count of a single NHL season on his own site rather than in a journal, and it states no strength-state filter for that count. The Wikipedia analytics article cited above points at this post and gives the 58% as a share of shifts; that figure is the share of ice time, and the share of shifts is 57%. The same author's score-adjustment method is cited above for score effects.

James Bloom, "How to Watch Hockey", Learning to Play the Game. https://learn-ice-hockey.com/off-the-ice/how_to_watch_hockey/ — licensed CC BY-NC 4.0.

<a href="https://learn-ice-hockey.com/off-the-ice/how_to_watch_hockey/">How to Watch Hockey</a> by <a href="https://github.com/jamesdbloom">James Bloom</a>, licensed under <a href="https://creativecommons.org/licenses/by-nc/4.0/">CC BY-NC 4.0</a>.

@misc{off-the-ice-how-to-watch-hockey,
  author  = {James Bloom},
  title   = {How to Watch Hockey},
  howpublished = {Learning to Play the Game},
  url     = {https://learn-ice-hockey.com/off-the-ice/how_to_watch_hockey/},
  note    = {Licensed CC BY-NC 4.0}
}