Bands show position percentiles; axis labels show current rolling values.
Dashed outline: peer average in the fixed reference. Outside = above average; inside = below average. All axes point outward for stronger results.
Player value: 1.99 Peer range: 1.68 – 4.20 xGF/60
MeanTyson Kozak
Lower Higher (better)
Player 1.99 · mean 2.54. Higher raw values rank higher.
Rolling profile against a fixed reference group
Fixed rolling-profile reference · 430 players. Values use the same history and smoothing as the radar.
History, sample sizes & calculation
Higher radar percentiles mean stronger recorded results against the fixed reference, including older seasons. These are separate from this season’s league percentile. Each player has a different amount of covered history. Weighted minutes and attempts give recent games more influence. “Prior” is the share contributed by a frozen league baseline: it is higher when a player has less evidence, and differs by metric because each uses its own smoothing strength. It is not a confidence score.
On-ice offense · xGF/60
350.5 weighted minutes · 36% prior
On-ice defense · xGA/60
350.5 weighted minutes · 53% prior
Chance generation · ixG/60
350.5 weighted minutes · 36% prior
Shot quality · ixG/attempt
32.7 weighted attempts · 43% prior
Finishing · goals above expected/60
32.7 weighted attempts · 55% prior
History through Oct 12, 2026. Recent games count more than older games. History carries across seasons and stays unchanged while inactive. On-ice results are shared; finishing describes results rather than established talent.
Recent scoring against his own season
Recent scoring against his own season
Loading recent games…
Loading splits…
Loading linemates…
Loading playoff games…
Loading game log…
Loading career ledger…
Every metric name opens its definition; the full list lives in Metrics.
Traditional statistics cover all situations. Supporting five-on-five analyses use completed games, raw adjustment and exposure from xG model v4. The two samples can differ.
An em dash means unavailable, not zero. A partial season remains partial. Where a panel reports evidence coverage, its rates use only the covered games. History never substitutes a different model, and a missing appearance interrupts a rolling model window.
Shot maps use the same shot model as Game Center and show which model version scored recent shots beside the rink. Older seasons can use earlier model versions. High danger means a model scoring probability of at least 0.20 on an unblocked attempt; for goalies, medium danger is 0.08 to 0.20 and low is below 0.08.
Special-team buckets include the wider recorded man-advantage and man-down states, read from the player’s own team; for a goalie, penalty kill means facing a power play. Their ranks need 100 recorded minutes and at least 30 qualified peers; below 100 minutes, penalty-kill percentiles are provisional. Playoff values have no ranks or percentiles. Team share is player ice time divided by team ice time in that situation, over the games the player dressed. Zone starts describe faceoff-started five-on-five shifts and exclude on-the-fly starts.
Penalties count events, including one for a double minor; bench and served-by roles are excluded, and victimless infractions have no drawn role. EPV added sums changes in the probability that the possessing team scores next; it is not standings points or an isolated player grade. Rolling windows never cross seasons or season types, and a missed game is not a zero-performance appearance. Career rates use summed counts and exposure, never an average of season rates; a missing contributing season withholds its career metric, and Performance begins in 2023–24.
Most often with Beck Malenstyn: 15 five-on-five minutes, 45.1% xG share together
Most often with Beck Malenstyn: 15 five-on-five minutes, 45.1% xG share together
Five-on-five · time together from shift charts (16:31 of his 5v5 ice time) · xG share together and apart
Teammate
Pos
TOI together
xG share together
GF–GA
CF%
Him without
Teammate without
Beck Malenstyn
L
14:59
45.1%
0–1
56.3%
0.0%
40.5%
Owen Power
D
10:09
14.2%
0–1
42.9%
67.3%
55.6%
Josh Norris
C
7:41
62.7%
0–0
50.0%
8.4%
45.9%
Sam Carrick
C
6:28
11.7%
0–1
50.0%
49.5%
18.4%
Olen Zellweger
D
6:19
23.0%
0–0
45.5%
41.3%
59.3%
Mattias Samuelsson
D
5:41
67.3%
0–0
66.7%
14.2%
51.4%
Zach Metsa
D
4:47
60.1%
0–0
50.0%
34.7%
39.7%
Louis Crevier
D
3:07
—
0–1
0.0%
49.7%
43.6%
Rasmus Dahlin
D
3:02
65.6%
0–0
50.0%
17.1%
51.9%
Tage Thompson
C
1:32
—
0–0
0.0%
43.6%
50.3%
Josh Doan
R
0:57
—
0–0
0.0%
43.6%
49.8%
Peyton Krebs
C
0:39
—
0–0
—
36.4%
48.2%
Without columns are five-on-five xG share apart: him on the ice without that teammate, and the teammate without him (with this club). Bars are centred on an even 50%.
Season grade
Withheld
Small sample
Performance / 60−0.442/2 GP
Not qualified · 2026–27 forwards
Needs 1 more covered appearances. Needs 31.3 more rated minutes.
Provisional comparisons · Small sample.
vs 376 qualified forwards
Percentiles appear once this player qualifies for season standings. The score breakdown is under How this is calculated.
How this is calculated
Score breakdown
Component
Weight
Scaled
Contribution
Offense
25%
−0.750
−0.187
Defense
25%
0.033
+0.008
Chance creation
20%
−0.863
−0.173
Finishing
10%
−0.013
−0.001
Assisting
20%
−0.416
−0.083
Performance rate
−0.436
Per 60 rated minutes, over the scored appearances. Each component is scaled separately, then weighted; the contributions add to the performance rate at full precision, so rounded figures may differ.
Scoring
Season grade is the qualified peer percentile on the same letter bands as Game Grade — A+ ≥88, A ≥78, A− ≥68, B+ ≥58, B ≥50, B− ≥42, C+ ≥35, C ≥28, C− ≥22, D ≥15, F below 15 — and needs qualification plus at least 30 eligible peers. The letter describes this season’s performance, not permanent ability.
Comparisons
Rank and component percentiles compare this season’s qualified forwards on unrounded rates over every checked appearance; Game Grade compares one appearance with frozen historical games.
Component percentiles are never averaged into the score. Recent history is optional, from up to three completed prior seasons; small samples are provisional.
Coverage
Qualification: 3 covered appearances, 48.8 rated minutes, and 90% appearance and workload coverage. Minimums advance with league progress; valid low-sample results stay visible.
Rated minutes exclude play with a goalie pulled: 0.9 min excluded; official TOI 18.3 min.
Coverage: complete. Updated .
Method
Five weighted components add to the performance rate. Five-on-five offense and defense account for recorded teammates, opponents and game state, so shared on-ice results are not isolated individual ability. Finishing is scored against the shot model’s historical-shooter expectation and assisting counts primary assists plus half of secondary assists; both use the strength baseline, chance creation its positional baseline.
Legacy rating
Legacy season score: 8.3 / 100. The fixed exposure baseline is 50; this is separate from the season peer grade.
36.4%
36.4% on-ice expected goal share at five-on-five
36.4% on-ice expected goal share at five-on-five
2.12 on-ice expected goals against per 60
2.12 on-ice expected goals against per 60
Defensive metrics
Five-on-five on-ice suppression; individual events in all situations.
Suppression
Player
Percentile
Rank
39.48
—
0.00
—
0.066
—
Puck management
Player
Percentile
Rank
0.00
—
0.00
—
0.00
—
Physical style
Player
Percentile
Rank
0.00
—
6.56
—
Context · not gradedNo rank yet
Player value: 39.48 Peer range: 34.49 – 78.21 / 60
MeanTyson Kozak
Lower (better)Higher
Player 39.48 · mean 56.08. Lower raw values are better. Needs 7 appearances and 105 five-on-five minutes (full season: 20 and 300).
On-ice attempts against per 60, including blocks.
Needs 7 appearances and 105 five-on-five minutes (full season: 20 and 300).
Discipline & puck management
Recorded season event counts · all situations
Penaltiesnet 0
Taken 0Drawn 0
Puck eventsnet 0
Giveaways 0Takeaways 0
Blocks
0
all situations
Hits
2
all situations
0.00 power-play points per 60
0.00 power-play points per 60
Power play
0:01 recorded ice time · peers need 100 minutes
Share of team PP
0.1%
in games dressed
PP points
0
goals + assists
Unit xGF / 60
0.00
team without: 8.53
Power play rates
Player
Percentile
Rank
0.00
—
0.00
—
0.00
—
0.00
—
0.00
—
0.00
—
No rank yet
Player value: 0.00 Peer distribution not published for this axis.
Tyson Kozak
Lower Higher (better)
Player 0.00 · mean —. Higher raw values rank higher. Needs 100 minutes in this special-teams scope.
Recorded goals plus assists per 60 in this scope; widened power-play and penalty-kill scopes.
Needs 100 minutes in this special-teams scope.
Penalty kill
1:34 recorded ice time · peers need 100 minutes
Share of team PK
12.7%
in games dressed
SH points
0
goals + assists
Unit xGA / 60
11.00
team without: 15.37
Penalty kill rates
Player
Percentile
Rank
11.00
—
0.00
—
229.79
—
0.00
—
17.51
—
No rank yet
Player value: 11.00 Peer distribution not published for this axis.
Tyson Kozak
Lower (better)Higher
Player 11.00 · mean —. Lower raw values are better. Fewer than 30 qualified peers.
Raw on-ice expected goals against per 60; the Season metrics sample.
Fewer than 30 qualified peers.
9:09 TOI per game (8:22 at five-on-five · 0:01 PP · 0:47 PK)
9:09 TOI per game (8:22 at five-on-five · 0:01 PP · 0:47 PK)
0.07 individual expected goals per 60
0.07 individual expected goals per 60
Results vs expected
Five-on-five · what happened beside the chances created and allowed
Chances
Player
Percentile
Rank
1.22
—
2.12
—
36.4
—
Goals
Player
Percentile
Rank
0.00
—
3.59
—
0.0
—
Luck & variance
Player
Percentile
Rank
0.0
—
66.7
—
66.67
—
Context · not gradedNo rank yet
Player value: 1.22 Peer range: 1.22 – 3.84 / 60
MeanTyson Kozak
Lower Higher (better)
Player 1.22 · mean 2.54. Higher raw values rank higher. Needs 7 appearances and 105 five-on-five minutes (full season: 20 and 300).
Raw on-ice expected goals for per 60; the Season metrics sample.
Needs 7 appearances and 105 five-on-five minutes (full season: 20 and 300).
5v5 goals
0–1
for–against
5v5 expected
0.3–0.6
xG for–against
EPV added
—
Attribution unavailable
On ice vs off ice
Buffalo's share with Tyson Kozak on the ice against the same team without him.
Share
On ice and off ice
On-minus-off gap in percentage points
xGF%
On ice 36.4%; off ice 43.6%.
43.636.4
-7.2
CF%
On ice 45.0%; off ice 46.6%.
46.645.0
-1.6
GF%
On ice 0.0%; off ice 60.0%.
60.00.0
-60.0
On iceOff ice
Scale 0–65% · marker at 50% · right column is the on-minus-off gap in points
0 points in the last 2 games (0.00 per game vs 0.00 this season)
0 points in the last 2 games (0.00 per game vs 0.00 this season)
Venue, rest and opponent
Venue, rest and opponent
2 games · most recent vs FLA, W 5–3: 0 G, 0 A
2 games · most recent vs FLA, W 5–3: 0 G, 0 A
Tyson Kozak
1 attempts0.02 xG0 G
Goal0Shot on goal0Missed shot1Blocked shot0Larger dot = higher xG
Shooting by zone
Zone
Att
G
xG
Goal rate / exp.
G−xG
Point
1
0
0.0
0.0% / 1.9%
0.0
All zones
1
0
0.0
0.0% / 1.9%
0.0
Sorted by goals minus expected goals. Goal rate is goals per unblocked attempt against the expected rate. Hover a row to light up those attempts on the rink.
Finishing evidence. Scoring 0.0% on chances vs 1.9% expected (−1.9 points).
Shot types1 attempt
Type
Share
Sh%
G−xG
Snap
100.0%
—
0.0
Share is context, not a grade. Sh% counts shots on goal only.
Shot profilecontext · not graded
On-net rate0 of 1 attempts on goal
0.0%
Slot share0 of 1 from the inner or high slot
0.0%
Average distancefrom the centre of the net
45.4 ft
Rebound attempts / 600 this season · all situations
0.00
Usage by situation
Ice time, share of team ice time, scoring and scoring rate, with where shifts start.
Scope
TOI/GP
Team share
G
A
P
P/60
Five-on-five
8:22
17.9%
0
0
0
0.00
Power play
0:01
0.1%
0
0
0
0.00
Penalty kill
0:47
12.7%
0
0
0
0.00
Other
0:00
—
0
0
0
—
All situations
9:09
15.3%
0
0
0
0.00
Where the 0 points come from · All situations
Goals0 · 0.00/60
Primary assists0 · 0.00/60
Secondary assists0 · 0.00/60
Zone starts 16.7% offensive of OZ + DZ
Offensive 8.3%Neutral 50.0%Defensive 41.7%
Shifts
15.0 a game · 0:37 average
Zone starts: five-on-five · Shifts: all situations
3 NHL seasons: 69 games, 11 points
3 NHL seasons: 69 games, 11 points
Totals cover all situations; rates are five-on-five.
Elite90+Strong70–89Average30–69Weak10–29Poor<10No rank yetBar length is the peer percentile, so longer is always better · tick at the 50th
Player value: 0.00 Peer range: 0.00 – 6.24 / 60
MeanTyson Kozak
Lower Higher (better)
Player 0.00 · mean 1.91. Higher raw values rank higher. Needs 7 appearances and 105 five-on-five minutes (full season: 20 and 300); shot quality and finishing also need 11 unblocked attempts.
Recorded five-on-five goals plus assists per 60 minutes.
Needs 7 appearances and 105 five-on-five minutes (full season: 20 and 300); shot quality and finishing also need 11 unblocked attempts.
How to read these comparisons
Player values are recorded season results. Peer averages are arithmetic means of qualified players in the selected group; each row uses the same peers for its average and rank. Ranks use unrounded values, with tied players sharing a rank and the next rank skipped. Lower xG against ranks better. Bar colours are the five performance tiers; context rows are blue and not graded. Differences from the average describe the season, not isolated player impact.
Goals above expected is the raw goals-minus-xG rate, without the shrinkage used in the older finishing comparison. On-ice measures include teammates, opponents and deployment. These recorded metrics describe different quantities from the adjusted Season score components beside them.
Skaters need 20 appearances and 300 five-on-five minutes; quality and finishing also need 75 unblocked attempts. Early in a season these minimums scale down with games played so far. At least 30 peers must qualify. Playoff ranks are withheld. A reviewed axis whose own evidence fails is withheld and shown with its reason rather than ranked. These workload rules do not express statistical confidence.
Position uses the NHL listed primary position available now; it does not establish historical position or actual deployment. Broad groups are forwards, defence and goalies. This version uses the current model’s covered games only.
Season vs reference
Per game · all situations
Metric2026–27ReferenceDiff
0.000.04-0.04
0.000.09-0.09
0.000.13-0.13
0.000.65-0.65
Reference: 2025–26 · 46 games.
Each game against his own season
Points in each game, coloured by how good the game wasTrailing 10 games, per gameSeason 0
