Goalie Efficiency Model · GEMGoalie Development Report
Washington Capitals
Sports Performance Analysis · G.E.M.Prepared for Washington Capitals Hockey Operations · Confidential — not for redistribution 01 /
Read This First

What this report is built on

Logan Thompson
G · Washington Capitals · NHL
12Games in sample 255Shots analyzed 248Decided 26GA .895SV% in sample
NHL sample · 02.05 – 04.02.2026 · 12 games
Charlie Lindgren
G · Washington Capitals · NHL
15Games in sample 397Shots analyzed 392Decided 50GA .872SV% in sample
NHL sample · 11.20.2025 – 04.05.2026 · 15 games
Every number in this report describes what happened in this analyzed sample — not the season, and not the goalie in general.
  • Lindgren’s sample grew 2.5× on 2026-08-25 — fifteen tagged games to Thompson’s twelve. One goal now moves his overall save % by about .003. Where the bigger sample moved a number, his slides say so rather than smoothing it over.
  • No single game carries his rates any more — 04.05 vs NYR (8 GA on 30 shots) held 38% of his goals in the six-game sample; at fifteen games it holds 16%, and excluding it he is .884 overall.
  • Two of his games arrived double-tagged at the source (two video uploads of the same periods). The duplicates — including one double-counted goal — were deleted in the source database on 2026-08-26, along with one save whose only video had died with the stale upload. Every number here is computed after that cleanup.
  • Baseline (“NHL pool”) = the other NHL goalies we track — 2,005 shots, includes backups — not a league average. Thompson’s rows always exclude his own shots from the pool.
  • The pool contains a 180-shot Thompson sample from GEM’s own earlier tagging (~9% of it), so Lindgren’s baselines are not fully independent of his tandem partner. It never mixes into either goalie’s own numbers on these slides.
Logan Thompson · games in the analyzed sample
DateOpponent
Charlie Lindgren · games in the analyzed sample
DateOpponent
Our Sample

Workload and quality faced

Logan Thompson
NHL · 255 shots analyzed · 26 goals against
Overall save %Overall
.000
222/248
High-danger save %High-danger
.000
52/75
20.7
shots faced per game
12 coded games
30%
of his shots were high-danger
NHL goalies we track: 31%
Charlie Lindgren
NHL · 397 shots analyzed · 50 goals against
Overall save %Overall
.000
342/392
High-danger save %High-danger
.000
79/118
26.1
shots faced per game
15 coded games
30%
of his shots were high-danger
NHL goalies we track: 31%
  • Lindgren carried the heavier per-game load — 26.1 shots per game to Thompson’s 20.7, across fifteen games to Thompson’s twelve, at an identical 30% high-danger share.
  • On high-danger itself both sat under the pool: Thompson .693 and Lindgren .669 against roughly .75 for the goalies we track — the tandem’s gap to the pool lives almost entirely in the high-danger tier.
  • Fifteen games is a real profile for Lindgren — the six-game caveat this deck first shipped with is retired.
What the movement data shows
Most-used chains share of all shots
Logan Thompson
Post Entry Feet Step Out - Set228.9%
Drift Post Entry Feet187.3%
Charlie Lindgren
Post Entry Feet Step Out - Set379.4%
Tracking Shuffles Step Shuffle266.6%
Save selection share of saves
Logan Thompson
Butterfly Glove2913.4%
Blocking Body Save2612.0%
Charlie Lindgren
Butterfly Glove6619.4%
Butterfly Blocker5516.1%
Beaten share of goals
Logan Thompson
Around623%
Blocker High519%
Charlie Lindgren
Glove High1428%
Glove Mid1122%
Chains are consecutive movement pairs, counted once per shot wherever they fall in the chain. GEM codes up to five movements, the save selection and the trajectory on every shot. The three blocks carry three different denominators, because the fields do: decided shots (248 · 392), saves carrying a technique tag (216 · 341), goals carrying a trajectory (26 · 50).
Logan Thompson · Profile

Traffic and the perimeter didn’t beat him

Overall save % · in these games
.000
222/248 in sample · 12 games · pool .906
30%high-danger share .693high-danger save % 11.5%should’ve-had goals
Identity
  • Sees it, stops it: .913 through screens (46 shots, pool .853) and .900 on deflections — obstructed sightlines are not where he gets beat.
  • Strong off the pass: .885 on one-timers (46/52) against a pool .850, and .955 from the dot lane.
  • Goal quality is not the story: 3 of 26 goals carry a should’ve-had qualifier (11.5%, pool 14.3%). 41% of his saves left a rebound, 27 of those attacked.
By situation — where he grades wellNHL poolThompson
Routine shots2 of his 26 goals.979.986 143/145
Dot-lane shots.907.955 64/67
Through screens46 shots.853.913 42/46
Deflections20 shots.818.900 18/20
Long possessions3+ movements before the shot.878.889 96/108
One-timersshot directly off the pass.850.885 46/52

Deflections and screens rest on 20 and 46 shots — comfortable margins on modest populations. Every row on this page excludes his own shots from the pool.

Logan Thompson · Analysis

Focus areas

By situationNHL poolThompson
Forced lateral movementGEM movement bucket · 8 of his 26 goals · −.141 against the pool.741.600 12/20
Slot shots17 of his 26 goals.855.764 55/72
High-danger save %23 of his 26 goals.752.693 52/75
Upper gloveglove high + mid · 7 goals — 12 of 26 finish high including blocker side.882.821 32/39
Penalty kill4v5 · 27 of his 36 PK shots came against set OZ structure.854.800 28/35
Controlled zone entriescoming into the offensive zone · 18 shots.953.889 16/18

The lead row was checked for game concentration: no single game carries more than one of the eight forced-lateral goals’ weight — without his heaviest game (03.31 PHI) the area is still .632. That game holds 4 of 26 goals on 8% of his shots, so no rate on this page leans on one night.

  • The theme is lateral: 8 of 26 goals on forced lateral movement, and Around leads his beaten chart (6 goals) — when he is beaten, the puck moves him and goes around, not through.
  • 17 of 26 goals come from the slot; outside the interior he is at or above the pool nearly everywhere.
  • In-tight chaos is a small population but a costly one: 4 goals on 5 such chances.
Logan Thompson · Development

Development keys

WT1 Forced lateral movement .600 (8 GA/20) · pool .741 · 8 of his 26 goals (31%) · clip-reviewed 2026-08-26: a perfect hold, every clip applies 20 clips On plays the tagging codes as a forced lateral — a low-to-slot feed, a net drive across him, a seam pass that moves him post to post — he is .600 against a pool .741, with eight of his 26 goals on those 20 shots. The coaches’ read (2026-08-26): the goals stem mainly from late arrivals and late reads on connected passes — when he is slow to identify the developing play, he shifts into chase mode instead of staying ahead of it; his relaxed style, a strength at times, contributes to those late reads, and on the rush his drift timing needs refinement to handle the lateral pass when it comes. His own clean lateral saves are the proof: set and square early, well before the release, the same urgency becomes an asset and he handles the movement cleanly. Development lever: sharpen set-play positional awareness to anticipate earlier and cut down the late reactions. Dev Clips
WT2 Upper glove .821 (7 GA/39) · pool .882 · add blocker high and 12 of his 26 goals finish high · clip-reviewed 2026-08-26: goals held 7 of 7 27 clips Beaten glove high or glove mid on 39 tracked shots at .821 against a pool .882 — with the five blocker-high goals added, nearly half his goals finish high. The coaches’ read (2026-08-26): this connects directly to the forced-lateral pattern — he is late to position and still in motion at the release, so the glove reacts lateral to his body instead of projecting forward to meet the trajectory line early. On clean glove saves the pattern reverses: at his target early, motion eliminated before the release, glove projecting forward into the line — minimum movement, maximum coverage. Identical filter to Lindgren’s WL1, and the two coaching reads now agree across the tandem: establish early and project forward. Dev Clips
WT3 Penalty kill .800 (7 GA/35) · pool .854 · 14% of his shots, 27% of his goals · clip-reviewed 2026-08-26: all goals apply 30 clips On the kill he is .800 against a pool .854, and 27 of the 36 PK shots are set OZ structure rather than scramble — seam play against an established unit. He also leaves the most second chances here: 14 of his 28 PK saves left a rebound, four attacked. The coaches’ read (2026-08-26): the same pattern as his lateral and glove-side play — late to position, still in motion at the release, with broader zone awareness in the mix. Seam pass, low-to-slot feed, or finding sight lines through traffic: delayed reads leave him reacting rather than anticipating. The three focus areas are one development chain — anticipate earlier, arrive earlier, and the release finds him set. Dev Clips

Clip counts are what each area’s playlist plays for Thompson: 20 · 27 · 30, 68 clips in the deduped union of the three (77 before dedupe). All three areas were clip-reviewed in full by Washington’s goalie staff on 2026-08-26: WT1 held perfectly — every goal and save applies, with the mechanism read as depth and timing under lateral movement — WT2’s and WT3’s goals all held, with 12 and 2 saves curated out of the playlists. Every shot still counts in the stats.

Charlie Lindgren · Profile

Forced movement and pace stayed his side of the net

Overall save % · in these games
.000
342/392 in sample · 15 games · pool .907
30%high-danger share .669high-danger save % 22.0%should’ve-had goals
Identity
  • The inverse of his partner’s profile: .780 in forced lateral movement (32/41, pool .738) and .932 off the rush (110/118) — the exposure that leads Thompson’s focus list is Lindgren’s best delta, and speed does not beat him.
  • The kill settled to the pool: .841 on 69 PK shots against a pool .850 — the six-game .906 was small-sample variance, not a trend.
  • Second chances stayed clean: 36% of saves left a rebound, under the pool’s 40%, with 29 attacked.
By situation — where he grades wellNHL poolLindgren
Routine shots5 of his 50 goals.981.976 201/206
Rush chances8 of his 50 goals.917.932 110/118
Dot-lane shots.938 at six games.917.910 91/100
Quick releasecatch-and-shoot.876.891 41/46
Penalty kill4v5 · .906 at six games.850.841 58/69
Forced lateral movementGEM movement bucket · 41 shots — the exposure that leads Thompson’s focus list.738.780 32/41

Fifteen games, 41–207 decided shots per row. Two six-game strengths — the kill (.906) and the dot lane (.938) — settled back to the pool as the sample grew; forced lateral movement, quick release and the rush held above it. The pool for these rows includes an earlier-tagged Thompson sample (~9%); see slide one.

Charlie Lindgren · Analysis

Focus areas

By situationNHL poolLindgren
Upper gloveglove high + mid · 25 of his 50 goals — 50% · clip-reviewed 2026-08-26: goals held 24 of 25.885.764 81/106
Through screens56 of the 73 against set OZ play.871.753 55/73
One-timersshot directly off the pass · .879 at six games, .779 at fifteen.866.779 60/77
Deflections32 shots — faded toward the pool as the sample grew.817.813 26/32
Controlled zone entriescoming into the offensive zone · 29 shots · goals in four different games.955.793 23/29
Two or more lateral movementsbefore the shot, excluding settled feet · non-routine — .842 at six games.816.752 85/113
Slot shots27 of his 50 goals.851.767 89/116
High-danger save %39 of his 50 goals.753.669 79/118
Should’ve-had goalsshare of goals against, lower is better · held as the sample grew 2.5×13.9%22.0% 11/50

Game concentration was re-checked on every row. 04.05 vs NYR (8 GA) now holds 16% of his goals; without it he is .884, and no row on this page leans on one night — upper glove is .772 without its heaviest game, and the zone-entry goals sit in four different games. One-timers are the row the sample changed most: .879 at six games, .779 at fifteen.

  • The organizing theme is sightline and the seam: screens and deflections together are 20 GA on 86 shots (.767, pool .865), five more goals come from the blue line through bodies (.896 vs .970), and set-play one-timers run .779 on 77.
  • Upper glove now carries half his goals (25 of 50) — the one exposure he and Thompson share, on an identical filter, and the rate held as the sample tripled.
  • His should’ve-had share, 22.0% against the pool’s 13.9%, held as the sample grew from six games to fifteen — no longer small-sample noise. Flagged for coach review.
Charlie Lindgren · Development

Development keys

WL1 Upper glove .764 (25 GA/106) · pool .885 · 25 of his 50 goals (50%) · clip-reviewed 2026-08-26: goals held 24 of 25 74 clips His most robust gap, and the first WSH area to be clip-verified: .764 on shots finishing glove high or glove mid, half his goals, with goals in 10 of his 15 games. The coaches’ read (2026-08-26): he is confident and consistent low glove, and that low-protect tendency means the glove is often not established early — it arrives late, through rounded, non-forward, windmill routes, with the eyes deprioritized. When his feet are set and his hands present forward, upper-region saves become direct and clean. Goals held 24 of 25 across both coaches’ reviews (one DNA was overturned on a second look); 32 saves the staff read as low-glove or off-mechanism were curated out of this playlist. Identical filter to Thompson’s WT2, so verdicts promote across the tandem. Dev Clips
WL2 Screens and deflections .767 (20 GA/86) · pool .865 · screened alone .753, deflections .813 · 40% of his goals 86 clips Two obstructed-sightline exposures grouped because they share a mechanism: shots through a screen (.753, 18 GA/73 — 56 of the 73 against set OZ play) and deflections (.813, 6 GA/32), still .792 without his heaviest game. The deflection leg faded toward the pool as the sample grew — the screened leg carries the area, itself a review question: if the clips say the deflection goals are a different mechanism, narrow the filter to screens. The tandem contrast sharpened: Lindgren now faces more screened shots than Thompson (73 to 46) and is .753 on them where Thompson is .913 — same structure, opposite results on the same tags. The clips must answer whether the difference is depth against the screen or how each finds the puck — and what each goalie’s game can lend the other. Dev Clips
WL3 Coming into the offensive zone .793 (6 GA/29) · pool .955 · goals in four different games — no longer a one-game artefact 29 clips On controlled entries that become set play he is .793 where the pool is .955 — still the widest delta in his profile after the bigger sample softened it. The one-game concentration that kept this area provisional is resolved: the six goals sit in four different games, and without the heaviest of those games it still runs .852. His rush number remains the mirror (.933, at the pool): pace on the entry is not the exposure; the pause between the entry and the shot might be. Kept, softened, for coach review — on those six goals, where is he when the puck crosses the line, and is he still tracking the carrier when the puck moves off him? Dev Clips
WL4 One-timers .779 (17 GA/77) · pool .866 · Thompson .885 · invisible at six games (.879 on 33) 76 clips The area the bigger sample surfaced: at six games his one-timer number was .879 — indistinguishable from the pool — and nine more games moved it to .779 on 77 shots, with goals in ten different games. It is a set-play pattern: 48 of the 77 come out of OZP structure, and ten of the 17 goals finish glove high or glove mid — the same ten sit inside the upper-glove area’s population. Only one of the 17 is tagged should’ve-had: earned goals off lateral puck movement, not soft ones. The clips must answer: is this its own pattern — set-play seams testing his set-and-release timing — or the upper glove arriving by another name? If the latter, it folds into WL1 rather than standing alone. Dev Clips

Clip counts are what each area’s playlist plays for Lindgren: 74 · 86 · 29 · 76, 209 clips in the deduped union of the four (265 before dedupe). WL1 was clip-reviewed in full by Washington’s goalie staff on 2026-08-26 — goals held 24 of 25 across both coaches — and its playlist reflects their verdicts; every shot still counts in the stats. WL2–WL4 await review — the player marks each with its open question.

Goalie Efficiency Model

Goalie Development Report

Logan Thompson Washington Capitals · NHL  ·  Charlie Lindgren Washington Capitals · NHL
Prepared for Washington Capitals Hockey Operations · Confidential — not for redistribution
Click to begin