How to Watch Hockey

Overview

Almost everything that decides a hockey game happens away from the puck, and the camera that follows the puck is training you to miss it.

Give your eyes one job per period. Watch a single player for a whole shift; or only the two defenders; or only the half of the ice the puck is not in; or the five seconds before each goal, rewound. 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. If a defending forward follows the player he is marking into the corner, that team is marking players; if he stays up near the blue line instead, it is marking areas of ice. Count how many attackers chase the puck deep in the attacking end and you have identified their pressure system; count how many stand up near the blue line during a penalty and you have the power-play formation.

Watch live for shape and television for technique. In the building, sit as high as you can, ideally high behind a goal — from up there the ice becomes a diagram. Television shows you hands and feet and hides four players.

Watch the blue line, the most under-watched place in hockey: tracked public research found that at 5-on-5 carrying the puck across it produced roughly twice the unblocked shot attempts and goals that shooting it in and chasing did.

Then the numbers, which are mostly misread. Shot-attempt counts — Corsi, Fenwick — measure territory, not the quality of what happened. Expected goals prices quality, but is built only on attempts nobody blocked and has no standard definition, so two sites will disagree about the same game. For goaltenders, goals saved above expected beats save percentage, which beats goals-against average, because only the first accounts for how hard the shots were. Every figure needs seven things before you believe it: its denominator, its strength state — all situations or 5-on-5 only — its sample size, who was on the ice with the player, what the score was, whose model it came from, and whether it survives the eye test. Without those, a hockey statistic is a rumour with decimal places.

---

Part 1: The Core Instruction — Stop Watching the Puck

Why your eyes are trained wrong

Broadcast hockey is produced for entertainment, not for education. The main camera is a tight-ish follow shot that keeps the puck roughly centred, which means the frame contains the puck carrier, the nearest defender, and perhaps two other players. The other six are off screen. Directors cut to replays during the exact moments — the seconds after a whistle, the seconds after a shot — when the tactical information is richest.

None of that is a criticism of broadcasters. It is simply a statement that the broadcast is optimised 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.

You cannot fix this by trying harder to concentrate. Puck-following is a reflex — the eye is drawn to the fastest-moving high-contrast object in the frame. You fix it by giving your eyes a different job.

The four exercises

Do these one at a time. Each is deliberately narrow. The point is not to enjoy the game; the point is to build a habit that later runs in the background while you do enjoy it.

Exercise 1 — Watch one player for a complete shift.

Pick a player. Watch only them from the moment their skate hits the ice to the moment they sit down. Do not look at the puck. The measured range for a shift is roughly 30 to 80 seconds (measured by Lignell et al. 2018, Brocherie et al. 2018 and Bracko et al. 1998; Vigh-Larsen and Mohr's 2024 physiology review collects the band rather than having measured it — see Conditioning and Recovery), with a well-managed one nearer 30 to 45, 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 single moment separates good players from everyone else. Do they stop and watch, or do they immediately turn and skate to a place that will matter?
• How often their head turns. Good players scan constantly — but "constantly" is less often than you would guess. Hockey coaching resources commonly suggest a shoulder check every one to two seconds off the puck as a target; the one study that has counted contradicts that. Berg et al. (2025) coded 2,545 puck receptions by 88 SHL and SDHL professionals and measured a mean of 1.48 scans in the five seconds before receiving, with at least one scan in 61% of those receptions. Those two are the paper's own published figures. The two that are easier to carry onto the ice are arithmetic on them rather than quotation, and the paper prints neither: 1.48 in five seconds works out at one scan roughly every 3.4 seconds, and the complement of the 61% is no scan at all before 39% of receptions. So set your expectations at roughly a look every three or four seconds, and note that even professionals frequently take none. One limit on that comparison, in the coaching benchmark's favour: Berg counted only the five seconds before a reception, while "every one to two seconds" describes general off-puck play, so the two are not counting quite the same thing. What separates the professionals is not volume: scanning first was worth 78.8% action success against 70.1% — "success" being the study's own criterion set, formulated in consultation with the Swedish Ice Hockey Association, under which a pass counts if it was precise enough to be received and a shot if it was fired towards the net, rather than a count of goals. See Scanning and Anticipation.
• Whether they ever stop moving. Amateur players are often said to glide for around a third of their shift — a coaching observation, not a measurement. Do not assume professionals do not. Only one study has counted it, and it puts two-foot glide first among NHL forwards' skating characteristics — but that study has been restated in two published places that print two different figures, and nothing held here settles which is the paper's own. Bracko's own 2004 article gives 39% of total time on the ice, without a decimal, in its prose and in its table; the Hockey Institute page, which is also Bracko's, gives 40%. The 1998 paper the tables come from is paywalled, so neither can be checked against it. (Skating carries the study, its sample, both restatements and its limits.) That is a balance position, not a rest: Bracko records the gliding position as rarely sustained, interspersed with strides and turns, and every other characteristic both starts from it and returns to it. So the thing to watch is not whether a player glides. It is whether the glide is taking them somewhere.
• How they get off the ice. Do they change on the correct side of the rink, at a moment when the change is safe?

Exercise 2 — Watch the defencemen only.

Watch the two defencemen of one team. Ignore all six forwards and the puck. This is the fastest way to learn defensive structure because defencemen are almost never near the puck, so everything they do is positional.

What you are looking for:

• Gap — the distance a defender keeps between themselves and an attacker coming at them. Watch how it closes as the attacker crosses centre, and whether it is a stick-length or three car lengths. See Defending the Rush.
• Which one goes to the puck and which one goes to the net. Every defensive pair has this conversation on every puck retrieval, silently.
• Pinching — a defenceman stepping down from the offensive blue line to keep a puck in the zone. Note that this is a different play from stepping up, which is challenging an attacker in the neutral zone. Watch how often they pinch, and what has to be true before they do.
• Where they are on the power play compared with even strength.

Exercise 3 — Watch the weak side.

The strong side is the side of the ice the puck is on; the weak side is the other side. Deliberately watch the half of the ice the puck is not on. On a broadcast you will lose it off frame constantly — that is the exercise. Every time the camera pans away from the weak side, hold in your head what was happening there.

The two labels swap the instant the puck crosses the middle of the ice, which is the part that is hard to hold in your head while you watch. The picture below is borrowed from the Rink Map and Glossary; its spatial description calls that a half of the attacking end, but nothing in the picture depends on which end it is — strong and weak side attach to the puck rather than to a zone, and they read the same way in the defensive zone the first two bullets below are about.

Diagram from rink map and glossary

What you are looking for:

• The weak-side winger in the defensive zone. Are they high near the point, low near the net, or somewhere in between? This one player tells you almost everything about which coverage the team plays.
• The weak-side defenceman. How far off the post are they? Are they facing the puck or facing their check?
• The backdoor. The far post is where a great many goals are scored, and nobody watches it until the puck is already there. Treat the size of that share as a coaching commonplace rather than a counted one: shot-location data is public and plentiful, but this document has no count of goals broken out by which post they went in at.

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. Then answer one question: what happened five seconds earlier that made this possible?

Almost always it is one of a small number of things — a lost battle on the wall, a failed clear, a defenceman caught flat-footed, a forward who did not get back, a bad line change, a screen. The shot itself is usually the least interesting part. If you can name the cause of a goal rather than the finish, you have started watching the game properly.

The same exercise works in reverse for chances that were stopped. Ask what the defending team did in the five seconds before the save 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 on the point all night" is a finding. Vague impressions evaporate.
• Expect it to be boring at first. Watching a fourth-line winger for forty seconds feels like nothing is happening. Nothing is happening — that is the point. 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 start reading what the teams are actually doing. Almost all of it is visible from a seat, and none of it needs coaching access.

A caution before the recipes. Every system below is a coaching choice, not a rule of hockey. Teams change them by score, by period, by opponent and by which players are on the ice — a team that traps with a two-goal lead may forecheck aggressively when tied. Read a team over a full period before deciding what they run, 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 defensive zone when the opponent has the puck. The naming convention is a count of 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 so that the two can be set side by side. Two forwards are below the top of the circles in the first picture and one is in the second, and that count is what the two names record. (If the symbols are unfamiliar, Reading Ice Hockey Diagrams is the key: a circle is a forward, a triangle a defenceman, a bare G the goaltender, and the unfilled shapes are the team the picture is drawn for — which is not the same team in every picture below.)

The next two captions are written for the player arriving on the forecheck, not for you in the stands. The body-checking law they carry belongs to Forechecking Systems rather than here. From a seat you are reading only the count and the depth.

Diagram from forechecking systems

Diagram from forechecking systems

Remember that F1, F2 and F3 are roles defined by order of arrival, not by position. Whichever forward gets to the puck first is F1, even if that is the left winger. Do not try to work out the system by watching which named position goes where. Count bodies and count depth. See Forechecking Systems.

A second tell: watch what F1 does when they arrive. If F1 takes a straight line at the puck carrier's body, the team is hunting turnovers. If F1 angles across to take away one side and steer the carrier towards the boards, the team is playing containment and waiting for F2 to seal.

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

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

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 — up on the opposing point, the area just inside the blue line — and never follows a body below the hash marks, the short lines outside each end-zone faceoff circle that coaches use as the zone's high/low depth reference — they play a zone system, most commonly a low zone collapse. (Be precise about "high" here: it means on the point, not down at the faceoff dot. On an NHL sheet those are a little under 44 feet apart — the blue line is 64 feet from the goal line and the end-zone dots are 20 feet from it, and the point sits just inside the line rather than on it, so the gap is a shade under the difference between the two rather than a distance the rule book states directly; Rink Map and Glossary carries both figures, and the IIHF book that is the operative one in Britain gives them in metres. Confusing the two places is how people misread the system.) In a collapse the low three players (usually both defencemen and the centre) sink towards the net and defend the space in front of it, while the wingers guard the points and the high slot. Nobody is "assigned" to an opponent; you defend the area you are standing in. This is the house default assumed throughout. It is also widely described as the most common system in amateur hockey, on the reasoning that it is the easiest to teach — but that is coaching consensus, not a measured prevalence, and no survey of amateur defensive-zone 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 them off and picks up whoever arrives — they play a hybrid, sometimes called man-on, zone-off, or a collapsing man system.

The first two of those answers are drawn side by side in Defensive Zone Coverage. The moment drawn is not the one described above — it is the opposing point man walking the puck down from the blue line rather than a puck sent into the corner — but the winger is still the player to watch, and what separates the two pictures is whether he releases him or goes with him.

Diagram from defensive zone coverage

Diagram from defensive zone coverage

Second confirming tell: watch the strong-side defenceman when the puck is behind their own net. In a collapse, they will hold the net-front and let the puck carrier come out. In man-to-man, they will pursue 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 surrenders the opponent's own end and instead builds a wall across the middle of the ice, forcing the attacking team to either dump the puck in without possession or turn it over 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 in the offensive zone. A single forechecker goes, applies token pressure, and retreats.
• The defending team's players stop in the neutral zone rather than skating at anyone. Trapping teams look passive, and they are meant to.
• 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

An honest note: the pure 1-3-1 neutral-zone trap is far less common now than its reputation suggests, and most modern teams use trap elements — a 1-2-2 with a heavy middle-lane emphasis — rather than a full trap. Watch for the behaviour rather than trying to match a diagram. See Neutral Zone Systems.

Two more pictures make that instruction concrete rather than contradicting it. Both are a 1-2-2 — the same five players, the same name — and what separates them is depth and intent. In the first, the two middle forwards are up near the opposing blue line. In the second, they have dropped back to the centre red line, and the structure has given up on stopping the attack — the carrier is allowed to skate, and steered to a wall. That much you can read from a seat. What neither picture can show you is the part that actually decides which one you are watching — whether that lone forechecker is closing or gliding — so set the two shapes against each other, then go back to watching the players.

Diagram from neutral zone systems

Diagram from neutral zone systems

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

The point is the area just inside the blue line. The half-wall is the boards roughly level with the faceoff dot. The bumper is the player in the middle of the ice 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, one bumper between them, one at the net front. First units are widely described as running four forwards and one defenceman, and the 1-3-1 is routinely called the dominant modern formation. ⚠️ The personnel half of that has been counted; the shape half has not. Matt Cane, writing on Hockey Graphs in March 2017, measured four-forward units at "roughly 56% of the 5-on-4 ice-time this season, up 4% from last year and more than 15% from 5 years ago" — so the four-forward power play was already the majority choice across the league nearly a decade ago, and it was rising. Read that figure for what it is. It is a share of ice time across the whole league, not a count of teams, and it is 2016-17 data — and it is not a first-unit figure at all. What Cane offers on first units is a separate finding: that since 2015-16 teams were "1.2 times as likely to use a 3 forward unit on shifts that did not start a power play". He flags that one's limit himself, in a footnote attached to it — the method measures "only when the shift starts and not which units are on the ice", so it "may be understating the difference in 4 forward usage between 1st and 2nd units." No count of which shape teams run — 1-3-1 against umbrella against overload — was found for this guide, so "most teams use a 1-3-1" is coaching-material consensus rather than a measured prevalence. Special Teams sets the figure out in full with its qualifications.
• Three high — an umbrella: one in the middle at the point and two on the flanks below him, forming an arc rather than a flat line, with two low. How far the flanks drop varies — barely, in the classic version with both defencemen out wide; all the way to the tops of the circles in the modern one. A true umbrella has no bumper; that middle high player is at the top of the umbrella, not in the slot. See Special Teams.
• Four players loaded onto one side of the ice — an overload, working the puck along one wall.

All three of those are drawn in Special Teams, with the four penalty killers left out of each so that the shape reads on its own.

Diagram from special teams

Diagram from special teams

Diagram from special teams

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

• Two high and two low, in a rectangle — a box. The two high players are the forwards, the two low are the defencemen. A box has no single top player.
• A single player at the top, two in the middle, one low — a diamond. Teams switch to a diamond against a 1-3-1 specifically because a box leaves the bumper unmarked in the middle of the ice.
• Three killers in a tight triangle near the net plus one who pressures whoever has the puck — a wedge+1. Watch for the handover rather than the chase: when the puck moves the pressuring player stops and the nearest triangle player steps out in his place. One killer following the puck all the way around the perimeter is not the shape working — it is the shape broken. Expect the fourth player to look different from team to team: some send him straight out from the net at the puck, some sweep him across the top of the zone to keep the power play on one side, some let him roam out to the wall and the point.

All three are drawn in Special Teams, each against the same 1-3-1, so that they can be set side by side. The fill has swapped since the pictures above: here the four killers are the unfilled team and the power play is the filled one. The third caption is written for the four killers, not for you in the stands — the shot-blocking technique in it belongs to Special Teams rather than here. From a seat you are reading the triangle, the fourth player, and whether the handover happens when the puck moves.

Diagram from special teams

Diagram from special teams

Diagram from special teams

See Special Teams.

---

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 attempt to move the puck out of your own zone. When a defenceman goes back for the puck, immediately look away from them and find the other four skaters.

The five jobs are drawn as a shape in Breakouts, frozen at the instant the puck is retrieved and with no forecheck in the picture at all — the retriever, and the four skaters the bullets below ask you to find.

Diagram from breakouts

• Where is the strong-side winger? On the wall between the hash marks and the goal line is the classic outlet spot, and the landmark is routinely misread: that band is below the hash marks, not level with them. A winger higher than that, up near the blue line, is not automatically out of position — holding high is what a flat or spread breakout asks for, and what a stretch asks for. It can equally be a winger who never made the trip down from point coverage, and the pass to a winger up at the blue line is long, flat and arrives in their back. A winger right down behind the goal line, or in the corner, usually means the breakout is already broken. See Breakouts.
• Where is the centre? The centre is the swing option, usually curling low-to-high through the middle. If the centre is standing still, the breakout has one fewer option than it should.
• 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? Watch whether they present a target on the far side or stay glued to the net.

That first landmark is the routinely misread one, and the reason is that the right answer moves with the structure. Winger draws the same player three times on the same wall — deep between the hash marks and the goal line, higher up toward his own blue line, and gone up ice altogether — one for each of three breakout structures, with only one of the three on the ice at any moment. Use it as a list of the places a winger can legitimately be, not as a target he is missing.

Diagram from winger

Then watch what the retriever chooses: a rim (a hard puck sent onward around the boards), a reverse (a puck sent back against the flow), a D-to-D, a direct pass to the wall, or a skate-out. Ask which forechecker made that choice necessary. See Breakouts.

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

This is where games are decided and almost nobody watches it. Public tracking work by Eric Tulsky and colleagues, presented at the 2013 MIT Sloan Sports Analytics Conference, found that at 5-on-5 carrying the puck into the offensive zone is worth roughly twice as much offence as dumping it in — see Zone Entries, which covers the per-entry figures, and note there that "shots" in that study means unblocked attempts rather than shots on goal.

So when the puck approaches the blue line, watch:

• Did they carry it in, dump it in, or turn it over? Keep a mental tally for one period. You will be surprised how 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 out at the red line, already shrinking the gap, has usually won. Defending the Rush carries the distances, and they are not one number — roughly two to three stick lengths through the neutral zone in open ice, around a stick and a half as the carrier crosses the centre red line, and roughly a stick length at your own blue line. Be precise about what stands behind each rung, because the published coaching pages disagree with each other. The neutral-zone figure is published, and so is the stick length at the blue line — but a second page anchored to that same line prints two stick lengths rather than one, and the only page found that anchors a distance to the centre red line prints two stick lengths there rather than a stick and a half. Some coaching pages give distances in stick lengths but attach them to zones rather than to lines, so they settle none of this. Defending the Rush also says in terms that all three are a guide, not a law: they move with the defender's backward speed and the attacker's skill, so a wider gap on a fast attacker is not automatically a mistake.
• On a dump-in, how many attackers actually chase? A dump with one chaser is a change of possession dressed up as a play.
• Offside requires both skates to have completely crossed the blue line before the puck completely crosses it. One skate on or behind the line keeps you onside, and a skate lifted in the air over the neutral zone still counts as onside — in the NHL and IIHF; under USA Hockey Rule 630(a) and Hockey Canada Rule 6.11 skate contact is required, so the lifted skate is offside in two books of the four. Watch how attacking wingers deliberately drag a trailing skate.

Two pictures, one for each side of the line. The first is the attack arriving — three lanes, the two wide players already on the blue line, the carrier a stride behind them, and the wide players onside because a skate is still on the line — which is the stronger reason of the two, since a skate on or behind the line keeps you onside whatever the puck does. The second is the defender's gap, drawn at the centre red line, which is where the second bullet above says it is usually won. The two are drawn from opposite benches: in the first the unfilled team is the one attacking the line, in the second it is the one defending it.

Diagram from zone entries

Diagram from defending the rush

See Zone Entries.

Faceoff alignments — read them before the drop

There are around 56 faceoffs in an average NHL game — Faceoffs covers that figure and the computation behind it — so this happens constantly. Look at the four non-centres on each team before the puck drops.

• 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, work out which defenceman is the shooter. In the common default alignment both defencemen stand at the point — the area just inside the blue line — and the one on the draw side is the shooter, with feet already set and stick already loaded before the puck drops, while the far-side one holds the line as the safety against a counterattack. Offensive-zone alignments differ from team to team, so take that as an opening read and then watch where the puck is actually pulled on a clean win. See Faceoffs.
• In the defensive zone, watch which winger is on the boards side. That is the player who has to beat the opposing defenceman to a puck that gets back to the point.

Both of those alignments are drawn in Faceoffs, each at the right-hand circle and each frozen before the drop, and in both the unfilled five are the team the picture is drawn for. Take them as the common default and then check them against what this team actually does — the captions say so themselves. The defensive-zone one also shades the circle that the players not taking the draw have to keep clear of — and its caption sets out how the four rule books differ on whether touching the line counts, which is worth knowing before you decide from a seat that you have seen an encroachment.

Diagram from faceoffs

Diagram from faceoffs

One rules detail worth knowing while you watch: faceoff ejections are caused by moving before the drop, improper stick placement, or encroachment by a teammate — a winger creeping into the circle gets the centre thrown out of the draw. A second violation by the same team on the same draw is a bench minor in the NHL — an individual minor under USA Hockey Rule 613(d). The IIHF answer to the first violation changed with the 2026/27 edition. The IIHF Official Rule Book 2026/27, published June 2026, has the offending team's draw-taker "ejected from the face-off circle" (IIHF 2026/27 Rule 76.6) — the NHL's answer. The 2025/26 edition it replaced ejected nobody: it gave the offending team a warning instead. Under either edition a second violation by the same team on the same draw is a bench minor for delay of game (IIHF Rule 76.7), so it was never free. Closing your hand on the puck is a minor penalty rather than a faceoff violation under the NHL and IIHF books (Rule 67.2) — and a penalty shot rather than a minor when the puck is inside the offending team's own goal crease, or an awarded goal if that team's net is empty (Rules 67.2(ii), 67.4, 67.5, 63.6). Under USA Hockey Rule 618(a) it is not a penalty at all outside the crease — "Play shall be stopped and a last play face-off shall occur" — though that book reaches a penalty shot or optional minor inside it (618(a), 614(b)). See Faceoffs.

Line changes — where structure collapses, and nobody is watching

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

Rules that shape what you are watching, from NHL Official Rules 2025-2026, Rule 82:

• Following a stoppage the visiting team must place its line-up first, and only then may the home team change. In practice this gives the home coach the last change and therefore control of matchups. Watch a home coach hold a line at the bench door until the away team commits.
• After an icing, only the team that iced it is barred from changing. Rule 82.1's first paragraph is often quoted short — the home team may substitute "except in cases following an icing" — which makes it sound as though nobody changes after any icing. The full sentence is "The home team may then make any desired substitution, except in cases following an icing, which does not result in the delay of the game," and the rule tells you later exactly who the exception catches: a team in violation of Rule 63.8 (Delaying the Game) or Rule 81 (Icing), which is also what Rule 81.4 says. So the offending bench is frozen and the other one changes freely — including the home bench, which keeps last change after a visitor's icing. Watch for it: the team that iced it stands still while the other team pours fresh legs over the boards. See Faceoffs.
• Rule 82.2 runs to eight steps, and the first four are the ones you can see at the benches. (i) The referee gives the visiting team up to five seconds to change. (ii) The referee then raises their hand — that signal both closes the visitors' window and starts the home team's. (iii) The home team gets up to eight seconds. (iv) The referee lowers their hand and no further changes are allowed. If you are watching for the last change, the raised hand is the moment to look at the home bench. One thing to watch for after an icing: the right to last change comes from Rule 82.1, which grants it "except in cases following an icing, which does not result in the delay of the game" — an exception that catches only the team that iced the puck, so if the home side iced it there is nothing to see at the raised hand.
• The other four steps are how it is enforced — and the last of them changes what you are watching at the end of a game. (v) A change attempted after the referee's signal, or too many men sent out for it, "shall not be permitted and the Referee will send the players who have attempted to change back to their players' bench"; the coach is then warned, and any further violation for the rest of the game, overtime included, is a "Bench Minor Penalty for Delay of Game – Improper Line Change." (vi) Once the hand is lowered, the linesperson taking the draw "will blow his whistle … to indicate that all players must be in position and on-side for the face-off within five (5) seconds." (vii) Players who are slow to the dot after that whistle, or standing in an offside position for the draw, are "warned once in the game by the Referee" — the coach is told as well, and in that situation the centre is not thrown out of the draw — with any later violation a bench minor for delay of game. (viii) Then the exception that matters most to a viewer: "In the last two (2) minutes of regulation time and any time in the overtime period(s), points (vi) and (vii) above are not applicable. The Linesperson shall give the teams a reasonable amount of time to set up for the ensuing face-off after points (1) through (5) above have been enforced." So in the last two minutes and in overtime the five-second whistle and the slow-player warning both switch off, and the teams get "a reasonable amount of time" instead. The long, unhurried set-up you are watching in the endgame is the rule working, not the officials losing hold of it. (The rule letters its own steps with roman numerals, one to eight. Its closing sentence then switches to arabic digits — "points (1) through (5)" — which is the book's own numbering slipping, and on its face points back at the first five steps.)
• A team that ices the puck cannot change lines (Rule 81.4) and cannot use a timeout — that second one is a different rule, Rule 87.1: "No time-out shall be granted to the defensive team following an icing, following the goalkeeper causing a stoppage of play from a shoot-in from beyond the center red line, or when a defending player accidentally dislodges the net causing a stoppage of play." Note the third item in that quotation, because it is a second route to the same trapped bench: under Rule 63.8 a net "displaced accidentally by a defending skater or goalkeeper" freezes that team's bench exactly as an icing does, and sends the draw to their own end at the dot the attacking team picks — so a net knocked off its pegs is worth watching for the same reason an icing is. Watch a tired line trapped in their own end after an icing — coaches name it as one of the game's reliable ways to concede, and it is why the change ban bites. That it is a large share of goals is coaching consensus rather than anything counted here. Both bars are the NHL's and the IIHF's; if you are watching a game played under USA Hockey or Hockey Canada rules, expect the benches to behave differently, because USA Hockey imposes none of them and Hockey Canada's line-change bar reaches only U18AAA, Junior and Senior at the Member's option (Rule 6.7(d) for the icing, Rule 6.1(g) for the dislodged net).

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 gets to the bench before the player going on leaves it. Too many men is almost always a communication failure, not a counting failure. See On-Ice Communication.
• Which forward stays out when only two change. Usually the one deepest in the zone stays, and the two nearest the bench go.

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. Force yourself to keep watching the live play rather than the replay.

• Where is the net-front attacker? Are they stopping at the top of the crease with their stick on the ice, or drifting past?
• Does the shooter follow their shot? Point shots that are not followed by a forward arriving are free clears for the defending team.
• Which team reacts to the rebound first? Rebound goals go to whoever moved before the puck landed.
• What does the weak-side defenceman do? Under the house-default low zone collapse the backdoor is theirs; under man-on-man it belongs to whoever's check went there — the person changes with the system and the requirement does not. So watch whether the far post is covered rather than whether that particular player is covering it, and note that the discipline to not go to the puck is a learned skill. See Defensive Zone Coverage, which sets the two systems side by side.
• On a save, does the goaltender freeze, direct it to the corner, or leave it in the slot? Directing rebounds is a deliberate skill, not luck. See Goaltender.

That last question has a geography, and Goaltender draws it: the shaded area is the one a rebound must not land in, with an attacker already standing in it waiting. That caption is written for the goaltender rather than for you — from a seat you are watching only where the puck ends up and who moved first.

Diagram from goaltender

---

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. This is the whole advantage and it is enormous. Every tactical read in Part 2 is easy live and hard on TV, because systems are patterns made of five players and TV shows you two.

Live also gives you:

• Real speed and real distance. Television flattens both. The first time you see live hockey at a good level, the thing that shocks you is how fast a two-line pass arrives and how little time anyone has on the puck. See Time and Space.
• Sound. You can hear which passes are hard, which shots are one-timed, whether players are actually talking, and how physical a battle really was.
• Benches and warm-ups. You can watch line combinations, who talks to whom, and how a team sets up its power play in warm-ups.

Sit as high as you can. This is not a consolation prize for cheap seats; a high seat is genuinely the better seat for learning. From high up the ice becomes a diagram and you can see spacing, layers and the shape of a forecheck. From the front row you see speed and skill and almost no structure. If you have a choice between the tenth row and the top row, take the top row when you are there to learn and the tenth row when you are there for fun.

Best of all, if the rink allows it, is a seat high behind the net rather than at centre ice — that view compresses the neutral zone and makes gaps, layers and defensive depth extremely visible.

Television is far better for detail

• Replays. You get to see the five seconds before the goal again, from another angle, which is exactly Exercise 4.
• Slow motion. The only realistic way to study hand position on a shot, a goaltender's push-off, or the exact moment a defender's hips turn.
• Pause and rewind. The single most useful button in hockey education. Pause the frame at the moment a pass is released and ask what the passer could see.
• Camera angles. Overhead and end-zone cameras occasionally give you something close to the live view.
• Commentary and telestration — sometimes genuinely instructive, sometimes not; treat it as one opinion.

The weakness of TV is structural and permanent: the main camera cannot show you five players at once and stay watchable. Some leagues and services offer wider tactical or "all-22"-style feeds borrowed from the football idea of a camera that keeps all twenty-two players in frame. In hockey this is usually called an overhead, tactical or coaching feed, and availability varies enormously by league and season. If you can get one, it is worth more than any other single 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 teaches you what good looks like. Watching yourself teaches you what you look like, and the gap between the two is almost always larger than you think.

Getting the footage

You do not need a rig. A phone on a tripod at the top of the stands, wide enough to see the whole ice, is better than a close-up shot of the puck. 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 available, not the "highlights". Highlights are the least useful footage in existence for player development.

The protocol for reviewing your own shifts

1. Watch the whole shift, not the moment you touched the puck. You will be tempted to skip to your touches. They are a far smaller slice than they feel: NHL tracking — a dataset the league itself calls unofficial — puts a player's puck possession at roughly 3–4% of 5-on-5 ice time, so on a forty-second shift your touches are somewhere between one and two seconds. The other thirty-eight-plus are where the improvement is. (Playing Without the Puck covers this figure and gives the study and its limits.)
2. Watch each shift twice — once following yourself, once following the play — and compare. The question is whether you were in the place the play needed you to be.
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? Could the puck carrier actually have passed to me — was there a clean lane, was my stick on the ice, was I moving to a place the pass could reach? See Puck Support and Spacing.
4. Count something. Counting beats impressions. Count your touches, your completed passes, your times entering the offensive zone with the puck, your times back below the puck on the backcheck. Track the same count over three games.
5. Pick exactly one thing to change for the next game. One. A review that produces six things to fix produces none.

The discipline of watching yourself away from the puck

This is uncomfortable, which is why almost nobody does it. Watching your own play with the puck is enjoyable — you get to relive the good pass. Watching yourself standing still on the weak side while a goal goes in is not.

Do it anyway, and do it specifically:

• Watch the three seconds after every one of your passes. Did you move afterwards, or admire it? Passing and standing still is among the habits coaches name most often at amateur level — an observation, not a counted ranking.
• Watch yourself on every goal against while you were on the ice. Not to assign blame, but to find out whether you were where your team's system says you should be.
• Watch your changes. Are you leaving the ice at safe moments, and are your shifts the length you think they are? Time three of them. That most players' shifts run ten to twenty seconds longer than they believe is a coaching observation rather than a measured finding, so time yours rather than trusting the figure.
• Watch your head. Count your looks over your shoulder in a shift, and compare the count with what you thought you were doing. That amateur players scan far less than they believe is a coaching observation rather than a measured finding — the only published count is Berg's, and it is of professionals — so treat it as a thing to check on your own video rather than as a fact about you.

Watching your own team as a team

• Work out your own team's systems using Part 2 and check your answer with your coach. If your read and your coach's answer differ, you have learned something valuable about how clearly the system has been communicated.
• Watch how the team behaves in the last five minutes of a one-goal game compared with the first period. If nothing changes, that is a finding. See Game Management.
• Watch the opposition's next opponent if you can. Scouting a team you are about to play is the most immediately practical watching you will ever do — and the two tells to look for are which forecheck they run and whether their wingers follow into the corner.

---

Part 6: Learning by Watching a Position

Pick the position you play and spend a whole period on it. Then pick a different one, because understanding what the player next to you is being asked to do is what makes you easy to play with.

Watching a centre

The centre's job spans all three zones and the full width of the rink, more of it than any other position's — coaching craft rather than a measured comparison, because no measurement of how much of the ice a position covers was found for this guide. The centre is also the connective tissue of every system. (A different belief is often said in the same breath: that a centre skates the furthest. That one is coaching craft as well, and the nearest thing to a test of it points the other way. No study separates centres from wingers on skating distance — the external-load literature splits players only into forwards and defence — but that forwards-versus-defence comparison has been measured: Lignell et al. (2018), tracking 36 NHL players through one game, found defencemen covered 29% more skating in total and were on the ice 47% longer, with forwards doing 54% more high-intensity skating per minute. Defencemen also take more shifts and more ice time. Center covers this and carries the citation.)

Watch for:

• Low support in the defensive zone. In a low zone collapse the centre is the low forward, working between the net front and the corners. Watch how they never fully commit to either.
• The curl on the breakout. Watch the centre swing low, receive with speed, and where they look before they get the puck.
• Faceoff technique, and then what they do in the second after the draw — the follow-up is a bigger part of a centre's faceoff value than the draw itself.
• The backcheck. A centre is usually the first forward back. Watch which lane they take — but do not grade it, because which backchecker owns which lane is a coaching choice, not a law. Under the standard positional assignment, which is the default assumed throughout this guide, the centre takes the middle lane and the wingers take the outside ones; other teams give the middle to whoever is first back, and others pick up man-on-man from the puck outward. So a centre coming back along the boards may be doing exactly their job. What is invariant under every version is that somebody owns the middle — so watch whether anybody does. See Center and Defending the Rush, which sets out the named team backchecking systems that decide which assignment applies.

The three lanes are drawn in Defending the Rush, with the middle one shaded and all three backcheckers drawn. The most dangerous player in it is not the puck carrier — it is the attacker driving the middle lane without the puck, who gets missed precisely because the puck is somewhere else. He is the clearest case in the game of what the camera will not be showing you.

Diagram from defending the rush

See Center.

Watching a winger

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

Watch for:

• How high they stay in the defensive zone, which as above tells you the team's coverage.
• Their position on the wall in the breakout, and whether they present a stick target or wait passively.
• What they do after they dump the puck in — do they become F1, or do they let the centre chase?
• Whether they support the puck or drift wide. Wide wingers look like they are in position and are actually 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. Watch the moment they stop skating backwards and start closing.
• Stick position. An active stick in the passing lane takes away options without any contact at all.
• How they take away time and space on a retrieval — do they check over their shoulder before they arrive at the puck?
• The decision to pinch or retreat. Watch what they check first: usually the position of their own centre and whether a teammate is high enough to cover.
• How deep they go in the offensive zone. Depth is the clearest read there is on how committed a defenceman is, and there are three depths that matter. Just inside the blue line — the point — is the default: at once the low-to-high outlet, the shooting threat that stops the defence collapsing, and the safety net against a counterattack. Down at the top of the circle, meaning the edge of the faceoff circle nearest the blue line, is the activated position, taken to shoot from a better angle, to walk the line toward the middle or to receive in the high slot — and still recoverable. Below the circles is always a commitment, and below the hash marks a defenceman is effectively a fourth forward who cannot get back. So watch whether a forward has taken the blue line for that defenceman before they go down there rather than after. How deep a team lets its defencemen go is a coaching choice — some give both a standing green light, some the strong-side defenceman only, some forbid going below the top of the circles — so read what this team allows rather than grading it against what you would do. See Offensive Zone Play, which sets out the three depths and who covers the vacated line.
• Net-front play. Body position, stick on the opponent's stick, eyes on the puck rather than the player.

The three depths are drawn in Defender, with the band below the circles left empty on purpose. Read it as the three places to check rather than as a ranking of them: which of the three this team allows is the coaching choice, and working out where their line falls is the thing you are watching for.

Diagram from defender

See Defender.

Watching a goaltender

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

Watch for:

• Depth. How far out of the crease do they start, and how does that change with the distance of the shooter?
• Tracking. Their head should be still and their eyes locked on the puck — you can see this clearly on replays.
• Movement type. Shuffles for small adjustments, T-pushes for larger ones, and whether they are set and stopped when the shot is released. A goalie moving at the moment of release is beaten far more often than a goalie who is set.
• Rebound control. Do pucks come back into the slot, or get directed to the corners or frozen?
• Puck-handling behind the net. In the NHL and IIHF play, the trapezoid (goalkeeper's restricted area, NHL Rule 27.8; IIHF Rule 27.7) limits where a goalie may play the puck behind the goal line; the only exception is a goalie who maintains skate contact with the crease. The KHL is commonly said to have it too, though that one is unverified here: no KHL rulebook was available to check it against. Neither North American governing body's book has one. The words "trapezoid", "restricted area" and "designated area" appear nowhere in USA Hockey's playing rules or its Casebook, and Hockey Canada's playing rules print "designated area" once only, for the spot Rule 5.4(b) stations a Goal Judge in. How much rec, beer-league and youth hockey that adds up to is a question about leagues rather than about books, and this document has no count for it. Under either of those books a puck-handling goalie has far more freedom than the one you are watching on TV — though no trapezoid is not no restriction. USA Hockey Rule 614(c) makes it a delay-of-game minor for a goalkeeper who falls on or gathers the puck and causes a stoppage where "(1) The puck is behind the goal line and their body is entirely outside of the goal crease" or where "(3) The puck is outside the boundaries of the 'goalkeeper's privileged' area" — that book's own restricted area, which its Note to 614(c) draws by "connecting the end zone face-off spots with an imaginary line and imaginary lines from each face-off spot running perpendicular to the end boards", so the corners sit outside it. Hockey Canada's Interpretation 3 to Rule 10.1(a), at its fifth item, is blunter: "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." So under those books their goalie may go and fetch the puck out there but may not sit on it. 🇬🇧 In England and Wales, treat this as unsettled. The IIHF book carries the rule (Rule 27.7, not the NHL's 27.8), but England Ice Hockey's Rules & Regulations 22.3 say the restricted area is "not currently enforced" — that is the 2024-25 edition — and neither the 2025-26 nor the 2026-27 In-House Rules mention it at all, both editions having been searched for it, though their Introduction says "Unless otherwise stated, all rules will be enforced in line with the IIHF Rule Book", which points the other way. Assume it binds, ask your league, and look behind the net to see whether the lines are painted at all. Treat an unmarked sheet as a hint rather than a ruling: IIHF 27.7 turns on "the position of the puck", not on the markings.

The shape itself is drawn in Rink Map and Glossary, whose caption carries the NHL and IIHF dimensions and the same book-by-book scope as the paragraph above — so read the picture as the NHL and IIHF one, and then look behind the net in front of you.

Diagram from rink map and glossary

See Goaltender.

---

Part 7: Statistics Literacy

Numbers are a way of watching hockey when you cannot be at the rink. They are also the fastest way to be confidently wrong, because almost every hockey statistic measures something slightly different from what its name implies.

A blunt warning before any of this. Public hockey analytics is a contested and evolving field, not settled science. Different sites compute the same metric differently, models are rebuilt every few years, the underlying event data contains real errors — arena scorekeepers have historically differed in how they record shots and hits — and reasonable analysts disagree about what most of it means. Treat every number below as evidence, never as a verdict, and check the definition used by whichever site you are reading, because it will not always match the one here.

Statistics literacy: the basic statistics

Goals. Credited to the player who propelled the puck into the opponent's goal. Per NHL Official Rules 2025-2026, Rule 78.2, only one point can be credited to any one player on a goal.

Assists. Per Rule 78.3, an assist is credited to the player or players — maximum two — who touch the puck prior to the goal scorer, provided no defender plays or has control of the puck in between.

• Good for: identifying who is involved in offence.
• Limits: the two-assist maximum is arbitrary, so the pass that actually broke the defence often goes uncredited. Assists are also awarded by the in-arena scorer, and there is a human judgement in every one. Rule 78.1 makes that decision final "notwithstanding the report of the Referee or any other game official", and obvious errors "should be corrected promptly" — but note the exception the rule carries: changes should not be made to the official scoring summary after the referee has signed the game report "except by the League's Chief Statistician." So "final" is not quite final; there is one route to a later correction.

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

• Limits: takes no account of ice time, linemates, power-play usage, or anything defensive. A player getting 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 on the ice for the conceding team gets a minus. Power-play goals award no plus or minus to either team. Empty-net goals are treated as even strength unless the scoring team is on the power play, and penalty-shot goals are excluded. It was used earlier by the Montreal Canadiens for internal evaluation before the league adopted it. The adoption date is contested: the Wikipedia article cited below gives 1959-60, but 1967-68 is at least as widely quoted, and no primary NHL source settles it. Treat the date as unresolved rather than quoting either figure with confidence.

• Why it is so widely criticised: it weights every goal equally regardless of situation; it depends heavily on your teammates, your goaltender and your opposition; it ignores power-play play entirely while counting shorthanded play; it is enormously affected by shooting and save percentage luck; and it does not distinguish a player who was the cause of a goal from one who was on the ice for it. A fourth-liner on a great team can post a better plus/minus than a first-line player carrying a bad one.
• Use it for: essentially nothing on its own. If you want on-ice goal differential, use a rate-based, situation-filtered version rather than the raw season total.

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

• Limits: conflates two completely different things — undisciplined stick fouls that cost your team, and fighting majors that may not. A high PIM total tells you almost nothing without knowing the composition. See Risk Management.

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

• Good for: it is arguably the most honest single number in hockey, because it is the coach's own revealed opinion of the player. Rising TOI means trust.
• Essential as a denominator. Comparing raw totals between a player getting 20 minutes and one getting 11 is meaningless. Convert to per-60-minutes 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 much narrower than people assume — at team level it is under nine percentage points from the league's best faceoff club to its worst, and much wider between individual centres. Faceoffs covers both distributions and the computations behind them, and explains why the two are not comparable. Faceoff wins are also credited crudely — a "win" that goes to a spot where nobody is standing is recorded identically to one that starts a breakout. Zone and situation matter enormously; winning defensive-zone draws shorthanded is a different job from winning offensive-zone draws on the power play.

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

• Good for: identifying finishing ability over very large samples.
• Limits: extremely noisy over short samples. A player who scores on 25% of their shots over twenty games is not a 25% shooter; they are a normal shooter having a good month. Individual shooting talent is real but takes several hundred shots to establish.

Save percentage (SV%). Saves divided by shots on goal faced, conventionally written as a decimal in North America — .920 means the goalie stopped 92.0% of shots — and as a true percentage in IIHF competition.

• Good for: a much better goaltending measure than goals against average, because it accounts for workload.
• Limits: treats every shot as equal. A point shot through no traffic and a backdoor tap-in count the same. This is the flaw that expected-goals models exist to fix.

Goals against average (GAA). Goals allowed per 60 minutes of play: goals against multiplied by 60, divided by minutes played. Overtime goals and time count; empty-net and shootout goals do not.

• Limits: heavily dependent on the team in front of the goalie. A goalie on a defensively strong team will post a good GAA while facing easy shots. Save percentage is generally considered the better of the two, and expected-goals-based measures are better again.

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). A CF% of 55% means that while you were on the ice, 55% of all shot attempts by either team belonged to your team. 50% is exactly break-even. Most players sit between about 40% and 60%, and above roughly 55% is often called elite.
• FF% is the same calculation using unblocked attempts.

The names are historical accidents rather than acronyms. Corsi was named by Tim Barnes, a Chicago financial analyst writing as "Vic Ferrari". He had wanted to name it after the Sabres' general manager, then after their head coach, and settled instead on Jim Corsi — Buffalo's goaltending coach from 2001 to 2014 — after finding a picture of him among the club's staff and liking his moustache. Fenwick was named after Matt Fenwick, an Alberta blogger and Calgary Flames fan, who in a November 2007 post proposed excluding blocked shots on the argument that a blocked shot is either not a genuine chance or is on average a worse one. (The further point often made alongside it — that blocking shots is itself a defensive skill — is not part of the summary of that post relied on here.)

• Why they became popular: they are a proxy for possession. You cannot attempt shots without the puck, so a team consistently out-attempting opponents is usually spending more time in the offensive zone. Crucially, shot attempts accumulate roughly eighteen times faster than goals — that ratio is an approximation from recent league scoring and attempt rates rather than a published constant, and the widely repeated "twenty times" is a rounding of it — so they stabilise into a meaningful signal over a handful of games where goals take most of a season. In the 2000s and 2010s, shot-attempt differential was shown to predict future results better than past goal differential did over short samples.
• Limits: every attempt counts the same, so a harmless point shot equals a slot chance; they are heavily distorted by score effects (see below) and by zone starts; they are strongly influenced by teammates, so a weak player alongside strong linemates inherits their numbers; and they systematically undervalue elite finishers, whose value comes partly from converting attempts at a higher rate than average. They are a team measurement being attributed to individuals.

Treat Corsi and Fenwick as what they are: a rough measure of territorial control, superseded for most purposes by expected goals, but still useful because they are simple, transparent and available everywhere.

Expected goals (xG)

An expected goals model assigns a probability of becoming a goal to each unblocked shot attempt, based on the characteristics of that attempt. Summing those probabilities gives an expected goal total: "this team generated 2.7 expected goals" means their chances would, on average, produce 2.7 goals.

The word "unblocked" is load-bearing, and it is the part most explanations of xG leave out. MoneyPuck's glossary defines xGoals as "The chance of an unblocked shot attempt being a goal," and states plainly that "blocked shot attempts are valued at 0 xGoals." So xG is built on the Fenwick event set, not the Corsi one: shots on goal plus misses, with blocks excluded rather than downweighted. That matters for the argument in the next paragraph — xG does not fix Corsi's flaw by pricing every attempt correctly, it fixes it by pricing the unblocked ones and discarding the blocked ones entirely. Other providers may treat blocks differently; check the glossary.

Typical model inputs include distance from the net, angle, shot type, whether it was a rebound, whether it came off the rush, the previous event and how long ago it happened, and the strength state. MoneyPuck's public model, for example, uses around fifteen such variables and was trained on hundreds of thousands of historical NHL shots using gradient boosting; other public models use logistic regression or other machine-learning methods.

• Why it improved on raw shot counts: it fixes Corsi's central flaw for the attempts it counts. A shot from the slot and a shot from the point are no longer the same event. This gives you a shot-quality-aware measure that still accumulates fast enough to be useful over a small number of games. The trade is that blocked attempts, which Corsi does count, drop out of xG altogether.
• How models differ: substantially, and this matters. Providers use different event definitions, different input variables, different algorithms and different training data, so two sites will give different xG values for the same game. There is no standard xG. Some models adjust for shot flurries and rebounds; some incorporate pre-shot passing or player tracking data; most public models still work from the league's event data alone, which does not know where the goalie was, where the defenders were, or whether the shooter was screened.
• Other limits: single-game xG is volatile and should not be read as a verdict on who deserved to win. Player-level over- or under-performance against xG is noisy and model-dependent. And the input data itself is imperfect — 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

Three contexts change what a number means. Ignore them and you will misjudge players constantly.

Zone starts. Which zone a player's shifts begin in — usually reported as an offensive zone start percentage. Check the denominator before you read anything into it, because sites do not define it the same way and the gap between the definitions is wide. Evolving Hockey publishes two versions on one table: one treats a change on the fly as a kind of zone start and divides by every shift you took, so nobody's figure is large; the other counts faceoffs only, dividing your offensive-zone draws by your draws in all three zones. A general-reference formula uses a third denominator again — offensive plus defensive zone starts alone, leaving out both neutral-zone draws and every shift that began on the fly — and the same player scores far higher under it. So a player quoted at 60% has not necessarily begun 60% of their shifts anywhere: read the glossary of whichever site you are on, and never compare a figure from one site against a figure from another. Game Management sets those definitions out with their sources and the limits on them. Coaches deliberately start skilled offensive players in the offensive 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.

Two important caveats about zone starts. First, the effect is real but smaller than early analysis assumed, because most shifts do not start with a faceoff at all. Micah Blake McCurdy, writing on HockeyViz in 2015, sorted roughly a million NHL shifts from the 2014-15 season by how they began and found that 57% of them started on the fly rather than from a draw; because those shifts run slightly longer, they accounted for 58% of all the hockey played that season. The two are not the same quantity: the 57% is a share of shifts and the 58% a share of ice time. McCurdy's count covers a single season and was published on his own site rather than in a journal, and that page is now reachable only through the Internet Archive.

The arithmetic points the same way and you can run it yourself: five skaters on the ice for sixty minutes is 18,000 skater-seconds per team per game, which at a mean shift of 47.8 seconds is a little over 375 skater-shifts a side. There are only about 56 draws in a game, and each draw can begin at most five of those shifts — 280 in all. So even if a team changed all five skaters at every single faceoff, which no team does, a quarter of its shifts would still have begun in open play — and McCurdy's 57% is what that gap looks like measured rather than merely bounded. (The arithmetic uses two figures from other documents: Faceoffs for draws per game, and Conditioning and Recovery, which covers shift length, for the mean. Carry that figure's qualifications with it: it is not a published statistic but a mean computed from the NHL's own public time-on-ice report — the 2024-25 regular season, averaged across the 597 skaters who played 41 or more games — and it carries no strength-state qualification, so it is all situations rather than 5-on-5. That report is a live statistics endpoint rather than a fixed publication, so the figure belongs to the day it was run, 2 September 2026, and a later run can return a different one.)

Second, the specific thing zone-start percentage measures is a third of the draws, not a majority of them. Faceoffs covers the zone split and the computation behind it: end-zone draws are the large majority of all faceoffs, but offensive-zone draws — the thing "sheltered usage" actually refers to — are only about a third of the total. A coach loading a player with offensive-zone starts is therefore redistributing within that third, spread across a whole bench, which is why the effect on any one player's season numbers is modest.

Quality of competition (QoC) and quality of teammates (QoT). Weighted averages of the opponents (or teammates) a player shares the ice with — as Evolving Hockey describes it, each metric is a weighted average of a player's teammates' or opponents' metric, weighted by the share of time spent with or against them.

• Limits: over a full season, quality of competition compresses heavily, because everyone plays everyone and the range between the toughest and easiest schedules is narrow. QoC is more useful over short samples and in playoff series where matching is aggressive. Quality of teammates is usually the bigger effect and the one people forget.

Score effects. Teams behave differently depending on the score. A trailing team takes more risk and generates more shot attempts; a leading team retreats into a defensive shell and concedes them. This is a large, well-documented distortion: a team that spends a season leading will look worse by shot attempts than it really is, and vice versa. This is why analysts use score-adjusted metrics, which weight events differently depending on the score state — Evolving Hockey, for instance, publishes weights that differ by score state and by home or away, using the score-adjustment method developed by Micah Blake McCurdy. (Evolving Hockey publishes the tables; the method is McCurdy's, and its own glossary says so.)

If you are reading raw, unadjusted possession numbers from games with lopsided scores, you are largely reading the scoreboard back to yourself. 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's definition: "Expected Goals against minus the actual number of goals the goalie has let in. A positive number means the goalie is stopping more goals than an average goalie would." A GSAx of +8 over a season means the goalie prevented roughly eight more goals than a league-average goalie facing that exact set of shots would have.

• Why it is much better: it separates the goalie from the team in front of them. A goalie facing thirty low-danger point shots a night and a goalie facing twenty slot chances can post identical save percentages while performing very differently. GSAx asks the only fair question: given the shots you actually faced, did you do better or worse than average?
• Related measures: expected save percentage — the save percentage an average goalie would post against the shots this goalie faced — makes the same comparison in percentage form. An older measure, goals saved above average (GSAA), compares against a league-average save percentage without adjusting for shot quality, and is a weaker version of the same idea.
• Limits: GSAx inherits every weakness of the underlying xG model. Public models generally do not know about screens, deflections that were not recorded as such, or the goalie's position, all of which matter enormously in goaltending specifically. It also needs a large sample; goaltending is the noisiest position in hockey and a season is not a long time.

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 a team's or player's on-ice shooting percentage plus on-ice save percentage. The NHL publishes it as SPSV%. It is expressed either on a 100 scale (multiply the sum by 100) or as a decimal around 1.000, depending on the site.

The name is not an acronym: it comes from the online handle of Brian King, the analyst who first proposed it.

The logic is that the two components sum, across the league, to almost exactly 100, and that neither on-ice shooting percentage nor on-ice save percentage is very repeatable at the individual level. So PDO regresses hard towards 100. The standard reading is that a team or player above about 102 is probably not as good as their results look, and one below about 98 is probably better.

• Good for: one thing only — a sanity check. If a team is winning while getting outplayed by every possession and xG measure, look at PDO. If it is 103, you have your explanation and you can expect the results to fall off.
• Limits: it is not purely luck. Elite finishers do genuinely sustain higher on-ice shooting percentages, and an elite goaltender genuinely sustains a higher save percentage, so a team with both can run above 100 for years. Calling PDO "luck" is a useful simplification, not a truth. It also says nothing about why — 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

Specific sites appear, change hands, go behind paywalls and disappear regularly, so learn the categories rather than memorising a link list. Anything named here may have changed, moved or shut down since this was written — verify before relying on it.

Public analytics sites. Free or cheap sites publishing possession, expected goals and goaltending numbers, usually with a glossary that tells you exactly how each metric is computed. Natural Stat Trick, MoneyPuck, Evolving Hockey and Hockey Reference are long-running examples. Always read the glossary first — the definitions differ between sites and knowing which one you are reading is the whole skill.

Coaching video breakdowns. Coaches and analysts who take clips and draw on them, explaining what each player was doing and why. This is the fastest way to convert watching into understanding, because someone is pointing at the thing you would not have noticed. The Coaches Site, Ice Hockey Systems and How To Hockey are established examples of coaching resources in this space; there is also a large amount of good short-form breakdown content published by individual coaches and analysts.

Wide-angle and all-22-style footage. Any feed that keeps all ten skaters in frame. Some leagues sell tactical camera angles; some teams' developmental and junior affiliates stream from a single fixed high camera, which is accidentally perfect for this purpose. Your own team's game film, shot wide from the top of the stands, is in this category and is more valuable to you than any professional footage.

Original research. Conference papers and public research blogs — the MIT Sloan Sports Analytics Conference has published a good deal of hockey work, including the zone-entry research cited earlier. Read these when you want to know why a metric exists rather than what its current value is.

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?" Those two questions resolve more confusion than any amount of watching.

How to judge a source. Good ones publish their definitions, say what their model does not know, and disagree with each other in public. Be sceptical of anyone who presents a single number as a final judgement on a player, and of anyone who dismisses all analytics on the grounds that the eye test is sufficient. Both errors come from the same place — wanting the answer to be simpler than it is.

---

Part 9: A Practical Weekly Routine

One hour a week, spent deliberately, will do more for your hockey brain than several hours of ordinary spectating — a coaching judgement rather than a measured ratio. Here is a realistic split.

20 minutes — one period of professional hockey, one exercise. Pick one exercise from Part 1 and do only that. Rotate weekly: week 1 one player, week 2 defencemen only, week 3 weak side, week 4 the five seconds before each goal. Watch live if you can, and sit high.

15 minutes — your own footage. Three or four of your own shifts, each watched twice, with the three questions from Part 5 written down. If you have no footage of yourself, get some — a phone at the top of the stands is enough, and asking a parent or a spare player to film one period is not a big favour.

10 minutes — one systems read. Take one team, professional or your next opponent, and answer two questions in writing: which forecheck do they run, and do their wingers follow into the corner? That is it. Two answers, ten minutes.

10 minutes — one statistic, properly understood. Pick a single metric from Part 7 and look up its definition on the site that publishes it. Then look at its actual values for a handful of players you know well and ask whether the numbers match what you have seen. You will learn more from the disagreements than the agreements.

5 minutes — write down one thing to try. The whole hour is wasted if it does not change what you do on the ice. One sentence: "This week I stop and look over my shoulder before every retrieval." Take it to practice. See Practice and Development.

If you only have half an hour, cut the professional hockey and keep the footage of yourself. Watching yourself is uncomfortable and it is also where the improvement is.

---

Common Mistakes

• Watching the puck and calling it studying. Three hours of following the puck teaches you nothing you did not know at the start. If your eyes did not do a specific job, you were a spectator.
• Trying to watch everything at once. Attempting to track systems, one player, the weak side and the statistics simultaneously means you retain none of it. One exercise per period.
• Only watching the highlights. Highlight reels are the least educational hockey footage that exists. They show you finishes, which are the last and least transferable link in a chain.
• Watching only players who play like you already play. You learn most from watching a player whose game you do not understand.
• Sitting as close to the ice as possible. Great for atmosphere, poor for learning. Height beats proximity when the goal is structure.
• Assuming what a professional team does is what your team wants. Almost every "always do X" you see on TV is "in this team's system, in this score state, do X." Find out what your team wants.
• Skipping to your own touches on video. Your touches are roughly 3–4% of your 5-on-5 ice time, not a tenth of it. That is NHL tracking data the league itself calls unofficial, and 5-on-5 is the basis it reports — see Playing Without the Puck, which covers the figure and its limits. The other ninety-six per cent is the part that is actually holding you back.
• Quoting a statistic without knowing its definition. Every site computes things slightly differently, and a great many hockey statistics arguments turn out to be two people using the same word for different numbers.
• Treating plus/minus as a measure of defensive ability. It is not, it never was, and it does not even include power-play goals against.
• Reading a single game's expected goals as a verdict. One game of xG is noise wearing a lab coat. Use several games and know whose model it is.
• Believing an unusually good stretch is a new level. If shooting or save percentage jumped and possession numbers did not, check PDO before you draw a conclusion.
• Treating analytics as either gospel or garbage. It is an evolving, contested field of imperfect measurements that are still much better than guessing.

Check yourself

Answer each question before you go on — a real attempt, not a guess. Then go and check, because the checking is the part that does the work. Producing an answer and getting it wrong is worth more than recognising a right one, but only when the right answer follows: the section named with each question, and the Key Takeaways after them, are where you get it. Where the answer lives in another document, that is named too.

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 goaltenders 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. The broadcast is optimised to show you the puck, and the puck is the least informative object on the ice. Watch one player for a full shift, or the defencemen only, or the weak side, or the five seconds before each goal — one at a time, never all four.
2. The players furthest from the puck tell you what a team is doing — but every system is a coaching choice, not a rule of hockey. A defending winger who follows their man into the corner is marking players; one who stays high on the point (not down at the faceoff dot, a little under 44 feet away on an NHL sheet — Rink Map and Glossary carries that measurement and what it is derived from) is marking areas; one who hands off when the man leaves their area is playing a hybrid. Read over five or six sequences and expect the answer to change with the score.
3. Count, to identify the rest. How many forwards go deep gives you their likely forecheck; how many stand near the blue line once a power play is set up gives you the formation. F1, F2 and F3 are roles defined by order of arrival, not by position.
4. Watch live for systems and TV for technique. When you are there to learn, sit as high as you can — high behind a net if the rink allows it, and the ice becomes a diagram. Take the tenth row when you are there for fun.
5. The blue line is the most under-watched place in hockey. Tulsky and colleagues' public tracking work (MIT Sloan, 2013) found that at 5-on-5 carrying the puck into the zone produced roughly twice the unblocked shot attempts and goals of dumping it in — Zone Entries covers the per-entry figures, the sample and the definition of "shots" used there.
6. Watch your whole shift, not your touches. They are a far smaller slice than they feel — and what you do in the two seconds after your team loses the puck is what separates good players from everyone else.
7. Time on ice is arguably the most honest single number in hockey, because it is the coach's own revealed opinion of the player. It is also the denominator you need: convert totals to per-60 rates before comparing anyone.
8. Corsi and Fenwick are a rough measure of territory, not quality; expected goals prices quality — but xG has no standard definition, so two sites will disagree about the same game. MoneyPuck values blocked attempts at zero; other providers may treat blocks differently, so check the glossary.
9. For goaltenders, GSAx beats save percentage beats goals against average, because only the first accounts for how hard the shots were — though it inherits every weakness of the underlying xG model and needs a large sample. PDO regresses hard towards 100, but calling it luck is a useful simplification rather than a truth: elite finishers and elite goaltenders genuinely sustain it. Treat it as a flag, not a verdict.
10. Every number needs its context — denominator, strength state, sample size, teammates, score state, whose model, and whether it passes the eye test. Without those, a hockey statistic is a rumour with decimal places. A deliberate hour a week beats several hours of ordinary spectating — a coaching judgement rather than a measured ratio — and it only counts if it ends with one specific thing you are going 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(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 · 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 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.

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 (3.08 goals for per team per game, 27.8 shots on goal per team per game, 11.1% team shooting percentage, power-play percentage mean 21.1% and penalty-kill mean 78.9%) were computed from the NHL's public team-summary data for the 2025-26 regular season, available via NHL Stats. 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}
}