Read this first
What this is built on
Every finding here comes from the shot-by-shot tagging behind the Goalie Efficiency Model: the pre-save movement chain, the direction of each movement, the save selection, the rebound, and the quality of the chance. Nothing was typed in; every number was computed from the tags and re-checked goalie by goalie.
The sample
4,555decided shots · 14 goalies · NHL and AHL · 2025–26
- Quality first. Every cut is conditioned on chance quality (Grade-A, mid, low), so "more movement" can never masquerade as "more dangerous shots".
- Posture at release means the last movement in the chain before the shot, with structural depth gains on the rush stripped out.
- Held goalie by goalie. A finding is only reported if it holds for most goalies individually, not just in the total.
How to read it
- Save % is on shots. Rebound % is the share of saves that left a rebound; attacked % the share where an opponent got to it.
- Where a chart says Grade-A, only the highest-quality chances are in it. Most of the movement story lives there; on routine shots posture barely matters.
- Each finding ends with the coaching read. The tags describe what happened; the read is what to do with it.
- Five findings carry a clip. Use the ½× button to slow it down.
The short version: movement volume is not the danger. Where the goalie finishes is. Feet beat ice, set beats moving, and the cost of movement shows up in the rebound before it shows up in the save.
01Posture at release
Where you finish matters far more than how much you moved
Within Grade-A chances, save percentage barely moves with the number of movements in the chain. It moves a lot with what the goalie was doing when the puck left the stick.
Grade-A save % by posture at release
On his feet.758 583
RVH / post seal.706 163
On the ice.676 460
Grade-A save % by chain length
1 movement.737
2 movements.722
3 or more.714
Feet vs ice: 82 points, z = 2.9. Feet beat ice for 9 of 12 goalies with enough Grade-A sample. On mid-quality chances the gap closes to 8 points (.944 vs .936): this is a Grade-A finding.
The coaching read
- Chain length is not a fault. A goalie who makes three reads and arrives on his feet is in better shape than one who makes one read and arrives sliding.
- The question to ask of any goal is not "how much did he move" but "what was he doing at the release".
- Train the finish: every drill that ends in a lateral arrival should be graded on whether the feet were under him at the shot.
Movement is fine. Arriving on the ice is what costs.
Top three on this finding · Best when he arrives on his feet (Grade-A)
| Goalie | on feet at release | Sv% on feet | Sv% on ice | Grade-A shots | |
|---|---|---|---|---|---|
| 1 | Juuse Saros NSH | 34% | .885 | .660 | 151 |
| 2 | Anton Forsberg LAK | 51% | .860 | .725 | 184 |
| 3 | Scott Wedgewood COL | 51% | .833 | .862 | 59 |
Ranked by save % when the last movement before the release was on his feet. Minimum 40 Grade-A chances.
02Forced lateral plays
On a pass across the midline, the slide is the last resort
Grade-A, pass across the midline · save % by how he arrived
Dynamic push on feet.883 77
Slide.674 141
Same play, split by how hard the arrival was (coach-rated 1–5)
Easy adjustment on feet 1–3.588 17
Dynamic push on feet 4.883 77
Slide 4.667 81
Slide to the post 5.692 52
RVH exit on the ice 5.556 27
Down or recovering.595 37
The hardest pushes hold best
Push after 2+ prior movements.914 35
Slide after 2+ prior movements.697 66
dynamic push on feetslideother arrivals
Dynamic push means T-push, step shuffle, double shuffle or euro step: the arrivals a coach rates a 4 for difficulty, the same rating as a butterfly slide. Like for like, the push holds .883 and the slide .667 (z = 3.2); 6 of 7 goalies with enough sample are better on the push, one is level. The easy adjustments are not a cheaper option on this play: 17 times a pass across drew only a tracking shuffle or a drift, and 7 went in, because he never arrived. One caveat stands: some slides were the only way to get there.
The read survives the caveat, and the split sharpens it: the pushes that follow two or more prior movements hold .914, so it is not that the push only wins on easy plays. If a goalie is sliding into a lot of Grade-A releases, the cause sits upstream: the read, the first step, or the depth he started from. Coach the moment before the slide, not the slide.
Clip 1Pass across on the rush, in overtime. A drift back, then a step shuffle: he arrives on his feet, set, and the glove is there. Held, no rebound.
SaveGrade-ARush · SlotDrift › Step shuffleButterfly glove · no rebound
Watch the feet: the drift keeps him square while the pass is travelling, so the final movement is a step, not a dive.
Top three on this finding · Reaches the most passes across on his feet
| Goalie | arrives with a push | Sv% on the push | Sv% on the slide | plays | |
|---|---|---|---|---|---|
| 1 | Charlie Lindgren WSH | 56% | .867 | .500 | 27 |
| 2 | Justus Annunen NSH | 50% | .882 | .706 | 34 |
| 3 | Scott Wedgewood COL | 47% | 1.000 | 1.000 | 15 |
Ranked by the share of passes across the midline he reaches with a push rather than a slide. Minimum 15 plays.
03Where the slide gets beaten
Sliding opens a hole on the side you are sliding toward
Goals scored while the goalie was travelling laterally · share beaten on the lead side
Last move was a slide76% 52/68
Last move was a push on feet48% 29/60
All lateral travel58% 170/292
Lead side means the side he was moving toward, read against his catch hand. Chance would be 50%. The overall 58% is z = 2.8 against chance; on feet there is no bias at all.
A push arrives sealed. A slide arrives with the lead pad and lead hand still opening. When a goalie is beaten on the side he was going to, the question is not "was he late", it is "what was still open when he got there".
Clip 2Pass across on the rush. A drift back, then a butterfly slide to his glove side. Beaten glove high: the side he was sliding to.
GoalGrade-ARush · SlotDrift › BT slideBeaten glove high
Freeze it at the release: the slide has carried the pads across but the glove is still travelling. That is the lead-side hole.
Top three on this finding · Best when he does have to slide (Grade-A)
| Goalie | Sv% arriving sliding | Grade-A slides | slide goals | |
|---|---|---|---|---|
| 1 | Anthony Stolarz TOR | .850 | 20 | 3 |
| 2 | Justus Annunen NSH | .767 | 30 | 7 |
| 3 | Sergei Bobrovsky FLA | .727 | 33 | 9 |
Ranked by Grade-A save % when the arrival was a slide. The slide costs everyone; these three lose least to it. Minimum 20.
04Rebound control
Movement taxes the rebound before it taxes the save
Save percentage is nearly flat across chain length once quality is held constant. Rebound control is not.
All saves · share leaving a rebound, by chain length
0–1 movements34.5% 1,207
3 or more42.3% 1,513
Grade-A saves · share where the rebound was attacked
1 movement16.0%
3 or more24.4%
Rebound rate z = 4.1; attacked rate z = 2.2. 11 of 12 goalies pay the tax, from 4 to 17 points. The one exception leaks more on short chains, which is his rush profile seen from another angle: rush chains average 1.35 movements.
The coaching read
- When a goalie is under movement his hands and pads absorb less. The save comes; the puck does not stay.
- A rebound-control session should be built on long-chain reps: two or three reads, then the shot. Static rebound drills measure a situation where nobody leaks.
- Rebound rate after movement is a better early-warning number than save percentage after movement, because it moves first.
The cost of movement shows up in the rebound first. Track it there.
Top three on this finding · Best rebound control under movement
| Goalie | rebound % · 3+ moves | rebound % · 0–1 moves | tax | saves | |
|---|---|---|---|---|---|
| 1 | Adin Hill VGK | 36% | 26% | 10 pts | 186 |
| 2 | Scott Wedgewood COL | 37% | 26% | 11 pts | 144 |
| 3 | Charlie Lindgren WSH | 38% | 31% | 8 pts | 280 |
Ranked by rebound rate on saves that followed three or more movements. The tax column is the change from short chains.
05The second shot
The rebound goal is a posture goal
Attacked rebound → next shot within 8 seconds · goal rate by posture at the second release
Faced from the ice30% 12/40
Faced from feet or post3% 2/58
Grade-A second shots only
From the ice50% 11/22
From feet8% 1/13
109 linked pairs, z = 3.7. Second shots overall go in at 13.8% against 10.4% for all shots, so the sequence is dangerous, but nearly all of that danger is carried by the ones faced from the ice. Saves with a tagged elevate-to-feet recovery: 0 goals on 18 follow-ups.
Getting up is the rebound save. The recovery is not what happens after the save; on a second shot it is the save.
Clip 3aPad save, rebound attacked. He goes for the second shot with a lateral recovery on the ice, and it goes in.
Clip 3bThe contrast. Save, rebound attacked, and he elevates to his feet at once. The Grade-A second shot is faced from his feet, and stopped.
3a · Goal from the iceLat on-ice recovery3b · Save from feetElevate to feetBoth second shots Grade-A
Top three on this finding · Most second shots met from his feet
| Goalie | from feet or post | second-shot goals | linked pairs | |
|---|---|---|---|---|
| 1 | Anton Forsberg LAK | 100% | 0 of 12 | 12 |
| 2 | Anthony Stolarz TOR | 73% | 1 of 11 | 11 |
| 3 | Carter Hart VGK | 66% | 4 of 29 | 29 |
Attacked rebounds linked to the next shot within 8 seconds, corpora with a game clock only. Ranked by the share of second shots met from his feet or a post. Minimum 8 pairs.
06 · 07Going down early · Screens
Two releases where the goalie should be still
06 · Grade-A goal rate when the last pre-shot movement was a butterfly
Dropped before the release40% 24/60
Everything else27% 315/1,149
z = 2.1, and the saves that do come leave a rebound 61% of the time. Dropping before the shooter has committed is the costliest single pre-shot decision in the corpus.
07 · Screened shots · save % by last movement
Step out and set.896 77
Double shuffle.897 29
Tracking shuffles.846 104
T-push at release.833 96
The extreme case: a T-push into a screened Grade-A release went in 14 of 18 times. Small sample, striking number.
On a screen: stop, set, find the puck. Lateral movement without a sightline is the one situation nothing rescues. And on any Grade-A release, the butterfly is a response to the shot, not a way to wait for it.
Clip 4Rush. A tracking shuffle, then he drops into the butterfly before the shooter has released, and the puck goes around him.
GoalGrade-ARushTracking shuffles › Butterfly (pre-shot)Beaten around
Freeze at the moment his knees touch: the shooter still has the puck. Everything the shooter does after that is against a goalie who has already chosen.
Top three on this finding · Least often down before a Grade-A release
| Goalie | dropped before release | Grade-A Sv% | Grade-A shots | |
|---|---|---|---|---|
| 1 | Darcy Kuemper LAK | 2% (2) | .659 | 129 |
| 2 | Sergei Bobrovsky FLA | 2% (2) | .660 | 103 |
| 3 | Anton Forsberg LAK | 3% (6) | .793 | 184 |
Ranked by how rarely the butterfly is the last thing he does before a Grade-A release. Minimum 40.
Top three on this finding · Best through traffic
| Goalie | Sv% screened | set at release | screened shots | |
|---|---|---|---|---|
| 1 | Scott Wedgewood COL | .947 | 53% | 38 |
| 2 | Logan Thompson WSH | .913 | 35% | 46 |
| 3 | Anton Forsberg LAK | .909 | 38% | 110 |
Ranked by save % on screened shots. Set at release means step-out-and-set, double shuffle, tracking shuffles or a positional adjustment. Minimum 30.
08Post play
RVH is a dead-angle tool that mostly faces slot shots
When RVH appears in the chain, where the shot came from
Dead angle287
Slot278
Dot lane188
Blue line126
Near parity. Half the time the post seal gets used, the release comes from the place it is not built for.
What it costs
| Cut | RVH | No RVH |
|---|---|---|
| Grade-A save % | .691 | .730 |
| Grade-A attacked rebound | 27.7% | 18.0% |
| RVH save vs blocking body save · rebound | 50.8% | 32.0% |
| Exit mid-shot · forced lateral · save % | .583 | .887 push |
Grade-A with RVH in the chain is worse for 9 of 12 goalies. Entering RVH for a true dead-angle shot is fine (.936 on 187). The cost is the exit.
The seal is not the concern; the exit is. Every RVH entry should be coached with its exit attached: what is the next pass, and can he be on his feet for it. On genuine sharp-angle plays, goalies who stayed on their feet at the post went 27 for 27; a small sample, but the direction is clear.
Clip 5Puck below the goal line. He enters RVH from his feet on the near post, the pass comes out to the slot, and he is exiting on the ice when the shot arrives.
GoalGrade-ASet possession · SlotPost entry › RVH entry › RVH exit on iceBeaten around
The exit is the whole clip: the puck moves faster than a goalie can get from a knee to his feet, and the top of the net is open for exactly that long.
Top three on this finding · Best with RVH in the chain (Grade-A)
| Goalie | Sv% RVH in chain | attacked rebound | Sv% shot mid-exit | Grade-A shots | |
|---|---|---|---|---|---|
| 1 | Justus Annunen NSH | .844 | 41% | .818 (11) | 32 |
| 2 | Scott Wedgewood COL | .813 | 15% | – | 16 |
| 3 | Anton Forsberg LAK | .767 | 39% | .815 (27) | 60 |
Ranked by Grade-A save % when RVH appears anywhere in the chain. Minimum 15.
09 · 10Save selection · Direction
The trailing pad, and which way he was going
09 · Share of saves leaving a rebound, by save selection
Butterfly glove24% 640
Blocking body save32% 397
Butterfly blocker38% 514
Lead pad51% 207
Pad save56% 346
Trailing pad79% 161
Trailing vs lead pad: z = 5.6.
A trailing-pad save is a diagnostic, not a technique. It says the goalie arrived square to where the puck was, and the puck got past the body line. Count them per goalie; the number is a position finding wearing a save-selection label.
10 · Grade-A save % by direction of the last lateral movement
Toward the glove.741 586
Toward the blocker.691 570
9 of 13 goalies better glove-ward, but only z = 1.9. Treat it as a lean, not a rule.
- The tagged direction is the direction travelled, not the pushing leg: 86.5% of last lateral movements go toward the side the shot came from. The drive leg is the mirror.
- Post choice in RVH runs 44% left / 56% right regardless of catch hand. The lean is to the right post, and there is no glove-post preference in this sample.
If a goalie's blocker-ward gap is wide, look at the glove-side drive leg and the blocker-side seal on arrival before anything else.
Top three on this finding · Cleanest on pad saves
| Goalie | rebound % on pad saves | trailing pad share | pad saves | |
|---|---|---|---|---|
| 1 | Adin Hill VGK | 35% | 10% | 52 |
| 2 | Scott Wedgewood COL | 53% | 8% | 40 |
| 3 | Logan Thompson WSH | 55% | 26% | 42 |
Ranked by rebound rate on pad-family saves (lead, trailing and pad save). Minimum 25.
Top three on this finding · Most even moving to either side (Grade-A)
| Goalie | Sv% toward glove | Sv% toward blocker | gap | catches | |
|---|---|---|---|---|---|
| 1 | Scott Wedgewood COL | .833 | .857 | −.024 | L |
| 2 | Anton Forsberg LAK | .775 | .807 | −.032 | L |
| 3 | Logan Thompson WSH | .700 | .667 | +.033 | R |
Ranked by the smallest gap between glove-ward and blocker-ward Grade-A save %. Minimum 25 each side.
11 · 13Situations · Load
The rush is a depth question, the set possession is a lateral one
11 · Average movements per shot, by how the play arrived
Rush1.35
Coming into the zone2.13
Quick attack off a turnover2.45
Offensive-zone possession3.09
Grade-A share is the same on all of them (29–31%), yet set possession runs .883 against .907 on the rush and leaves a rebound on 44% of saves against 39%. The gap is not shot mix; it is the movement the possession forces. Most tagged depth movement is retreat: 754 of 846 directed drifts go backward.
Rush work is about the depth line and the timing of the retreat. Possession work is about lateral arrival on feet. They are different sessions.
13 · Movements per shot when short-handed, relative to even strength
Highest1.60×
Typical range1.1–1.5×
Lowest1.04×
Every goalie in the sample skates more per shot on the kill than at even strength; the spread runs from 1.04× to 1.60×.
- The kill is where a goalie's movement budget is spent fastest: plan recovery around it.
- A goalie whose kill work barely rises above even strength is either exceptionally efficient or is letting the play come to him. The tags cannot tell which; video can.
Workload on the kill is a per-shot number, not a per-minute one. That is how it is built here, on purpose.
Top three on this finding · Best in set possession, the lateral situation
| Goalie | Sv% set possession | Sv% rush | moves / shot · OZP | OZP shots | |
|---|---|---|---|---|---|
| 1 | Scott Wedgewood COL | .939 | .966 | 3.08 | 99 |
| 2 | Justus Annunen NSH | .925 | .906 | 3.17 | 213 |
| 3 | Carter Hart VGK | .920 | .893 | 3.12 | 138 |
Ranked by save % in offensive-zone possession, the situation that forces the most movement. Minimum 60 possession shots.
Top three on this finding · Works hardest per shot on the kill
| Goalie | moves / shot · PK | moves / shot · even | ratio | PK shots | |
|---|---|---|---|---|---|
| 1 | Adin Hill VGK | 3.65 | 2.27 | 1.60× | 48 |
| 2 | Charlie Lindgren WSH | 3.62 | 2.47 | 1.46× | 86 |
| 3 | Logan Thompson WSH | 3.24 | 2.26 | 1.44× | 41 |
Movements per shot short-handed against even strength, raw chain length. Minimum 25 short-handed shots.
12Individual profiles
Every goalie has a movement fingerprint, and it is coachable
Two numbers describe most of the difference between goalies in this sample, and both follow directly from findings 01 to 03. Sorted, lowest first.
Share of forced lateral plays answered with a slide rather than a push
Charlie Lindgren WSH44%
Justus Annunen NSH50%
Scott Wedgewood COL53%
Darcy Kuemper LAK54%
Anton Forsberg LAK63%
Adin Hill VGK63%
Juuse Saros NSH67%
Artur Akhtyamov TOR67%
Logan Thompson WSH73%
Anthony Stolarz TOR76%
Carter Hart VGK77%
Sergei Bobrovsky FLA84%
From 44% to 84%. The goalie at the top of this list reaches twice as many of these plays on his feet as the one at the bottom.
Share of Grade-A releases faced from the ice or a post seal
Charlie Lindgren WSH43%
Logan Thompson WSH47%
Justus Annunen NSH49%
Scott Wedgewood COL49%
Darcy Kuemper LAK49%
Anton Forsberg LAK49%
Sergei Bobrovsky FLA52%
Adin Hill VGK54%
Artur Akhtyamov TOR54%
Anthony Stolarz TOR59%
Juuse Saros NSH64%
Carter Hart VGK66%
Most of the sample sits near half. The bottom three face six in ten of their hardest chances from the ice.
These two shares are development targets in themselves. Given 01 to 03, moving a goalie ten points down either list is worth more than any save-selection detail, and both can be measured every ten games.
In one page
The thirteen, and what to hold them against
The findings
| 01 | Where you finish matters more than how much you moved. Feet .758, ice .676 on Grade-A. |
| 02 | On a pass across, the slide is the last resort: push .883, slide .674. |
| 03 | A slide opens a hole on the lead side: 76% of slide goals, 48% of push goals. |
| 04 | Movement taxes rebound control before save %: 34.5% → 42.3% of saves leave a rebound. |
| 05 | The rebound goal is a posture goal: 30% from the ice, 3% from feet. |
| 06 | Dropping before the release: 40% goal rate on Grade-A, 61% rebound rate. |
| 07 | Never move through a screen: set .896, pushing .833, and 14 of 18 on Grade-A. |
| 08 | RVH faces the slot as often as the dead angle. The exit is the cost. |
| 09 | Trailing pad leaves a rebound 79% of the time. It is a position finding. |
| 10 | Glove-ward .741 vs blocker-ward .691: a lean. Post choice is right-biased regardless of catch hand. |
| 11 | Rush is depth (1.35 moves), set possession is lateral (3.09). Different sessions. |
| 12 | Slide share on forced laterals runs 44% to 86% between goalies. A target in itself. |
| 13 | The kill costs 1.04 to 1.60× the movement per shot, for every goalie. |
What the tags cannot see
- Movement is tagged around shots. Skating between whistles is invisible, so nothing here is an absolute workload.
- Slide vs push carries selection. Some slides were the only option. The per-goalie consistency and the lead-side finding are what make the read safe.
- Save selection is tagged on saves, not goals, so 09 is a rebound finding only.
- The second-shot link (05) uses the corpora with a game clock; one is excluded.
- 07's 14 of 18 and 08's 27 of 27 are small samples: striking, not settled. The top-three tables are ranked on one cut each and are gated on minimum samples, not adjusted for schedule or league.
Every number here can be re-run on the next ten games for any goalie in the model. That is the point of tagging movement: the fingerprint moves, and you can watch it move.