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: 2.33 Peer range: 1.68 – 4.20 xGF/60
MeanTyson Kozak
Lower Higher (better)
Player 2.33 · 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
35.0 weighted minutes · 85% prior
On-ice defense · xGA/60
35.0 weighted minutes · 92% prior
Chance generation · ixG/60
35.0 weighted minutes · 85% prior
Shot quality · ixG/attempt
5.7 weighted attempts · 81% prior
Finishing · goals above expected/60
5.7 weighted attempts · 88% prior
History through Jan 9, 2025. 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: 126 five-on-five minutes, 42.5% xG share together
Most often with Beck Malenstyn: 126 five-on-five minutes, 42.5% xG share together
Five-on-five · time together from shift charts (199:50 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
126:28
42.5%
7–8
40.9%
49.8%
46.1%
Bowen Byram
D
71:14
50.7%
4–6
36.9%
42.4%
47.3%
Sam Lafferty
C
71:06
45.9%
2–3
41.8%
44.8%
45.1%
Owen Power
D
65:02
52.5%
3–1
50.8%
42.1%
47.6%
Connor Clifton
D
64:14
38.8%
1–2
40.2%
47.1%
44.9%
Mattias Samuelsson
D
53:36
34.2%
3–2
37.0%
49.7%
46.3%
Rasmus Dahlin
D
50:52
30.2%
1–2
41.7%
50.6%
52.9%
Henri Jokiharju
D
33:11
53.6%
3–3
52.3%
42.7%
44.7%
Alex Tuch
R
29:41
50.4%
2–1
52.9%
43.8%
46.5%
Jacob Bryson
D
29:30
56.7%
1–2
55.2%
41.8%
43.4%
Nicolas Aube-Kubel
R
27:07
52.2%
3–3
44.4%
44.0%
44.0%
Zach Benson
L
21:51
42.4%
1–1
51.2%
45.6%
52.3%
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 / 600.1210/21 GP
Not qualified · 2024–25 forwards
Needs 6 more covered appearances. Needs 166.1 more rated minutes.
vs 113 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%
−0.101
−0.025
Defense
25%
0.351
+0.088
Chance creation
20%
0.045
+0.009
Finishing
10%
0.297
+0.030
Assisting
20%
0.088
+0.018
Performance rate
0.119
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: 16 covered appearances, 268.1 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 102.9 min.
Coverage: partial. 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: 78.9 / 100. The fixed exposure baseline is 50; this is separate from the season peer grade.
0.53 individual expected goals per 60
0.53 individual expected goals per 60
42.6%
42.6% on-ice expected goal share at five-on-five
42.6% on-ice expected goal share at five-on-five
2.23 on-ice expected goals against per 60
2.23 on-ice expected goals against per 60
Defensive metrics
Five-on-five on-ice suppression; individual events in all situations.
Suppression
Player
Percentile
Rank
60.03
—
—
—
—
0.054
—
Puck management
Player
Percentile
Rank
—
—
—
—
—
—
—
—
—
Physical style
Player
Percentile
Rank
—
—
—
—
—
—
Context · not gradedNo rank yet
Player value: 60.03 Peer range: 49.56 – 77.51 / 60
MeanTyson Kozak
Lower (better)Higher
Player 60.03 · mean 59.09. Lower raw values are better. Needs 20 appearances and 300 five-on-five minutes (full season: 20 and 300).
On-ice attempts against per 60, including blocks.
Needs 20 appearances and 300 five-on-five minutes (full season: 20 and 300).
Evidence coverage: high danger 11 of 21 games.
Discipline & puck management
Recorded season event counts · all situations
Penaltiesnet —
Taken —Drawn —
Puck eventsnet —
Giveaways —Takeaways —
Blocks
—
all situations
Hits
—
all situations
Both penalty counts cover 11 of 21 games with every drawn role recorded.
94.74 power-play points per 60
94.74 power-play points per 60
Power play
0:38 recorded ice time · peers need 100 minutes
Share of team PP
—
in games dressed
PP points
1
goals + assists
Unit xGF / 60
15.79
team without: —
Power play rates
Player
Percentile
Rank
94.74
—
—
—
—
94.74
—
12.10
—
15.79
—
—
—
—
No rank yet
Player value: 94.74 Peer range: 0.57 – 94.74 / 60
MeanTyson Kozak
Lower Higher (better)
Player 94.74 · mean 5.58. 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.
Evidence coverage: goals 1 of 5 games · team time 11 of 21 games.
Penalty kill
6:08 recorded ice time · peers need 100 minutes
Share of team PK
—
in games dressed
SH points
0
goals + assists
Unit xGA / 60
8.50
team without: —
Penalty kill rates
Player
Percentile
Rank
8.50
—
—
—
—
127.17
—
—
—
—
0.25
—
No rank yet
Player value: 8.50 Peer range: 8.50 – 9.22 / 60
MeanTyson Kozak
Lower (better)Higher
Player 8.50 · mean 8.85. 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.
Evidence coverage: goals 3 of 9 games · high danger 4 of 9 games · team time 11 of 21 games.
10:29 TOI per game (10:00 at five-on-five · 0:02 PP · 0:18 PK)
10:29 TOI per game (10:00 at five-on-five · 0:02 PP · 0:18 PK)
Results vs expected
Five-on-five · what happened beside the chances created and allowed
Chances
Player
Percentile
Rank
1.66
—
2.23
—
42.6
—
Goals
Player
Percentile
Rank
—
—
—
—
—
—
—
—
—
Luck & variance
Player
Percentile
Rank
—
—
—
—
—
—
—
—
—
Context · not gradedNo rank yet
Player value: 1.66 Peer range: 1.61 – 3.93 / 60
MeanTyson Kozak
Lower Higher (better)
Player 1.66 · mean 2.44. Higher raw values rank higher. Needs 20 appearances and 300 five-on-five minutes (full season: 20 and 300).
Raw on-ice expected goals for per 60; the Season metrics sample.
Needs 20 appearances and 300 five-on-five minutes (full season: 20 and 300).
5v5 goals
—
11 of 21 games
5v5 expected
5.8–7.8
xG for–against
EPV added
+0.43
21 attributed games · all situations
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 42.6% · Off ice unavailable
—
CF%
On ice 43.5% · Off ice unavailable
—
GF%
On ice unavailable · Off ice unavailable
—
On iceOff ice
Scale 40–65% · marker at 50% · right column is the on-minus-off gap in points
Hot 4 points in the last 10 games (0.40 per game vs 0.24 this season)
Hot 4 points in the last 10 games (0.40 per game vs 0.24 this season)
0.36 points per game at home · 0.10 away
0.36 points per game at home · 0.10 away
21 games · most recent at OTT, W 5–2: 0 G, 0 A
21 games · most recent at OTT, W 5–2: 0 G, 0 A
Shot map
All 27 attempts
Tyson Kozak
27 attempts2.00 xG3 G
Goal3Shot on goal8Missed shot16Blocked shot0Larger dot = higher xG
Shooting by zone
Hover a row to light it up · click to pin
Zone
Att
G
xG
Goal rate
Exp.
Goals vs expected
G−xG
Inner slot
12
3
1.1
25.0%
8.9%
+1.9
Point
2
0
0.0
0.0%
1.9%
0.0
Left circle
2
0
0.1
0.0%
3.9%
−0.1
Outer slot
4
0
0.3
0.0%
6.5%
−0.3
High slot
7
0
0.6
0.0%
7.9%
−0.6
All zones
27
3
2.0
11.1%
7.4%
+1.0
Finishing evidence. Scoring 11.1% on chances vs 7.4% expected (+3.7 points), led by inner slot volume and backhand finishing.
Sorted by goals minus expected goals. Goal rate is goals per unblocked attempt against the expected rate.
Shot profile
Context · not graded
On-net rate
40.7%
11 of 27 attempts on goal
Slot share
70.4%
19 from the inner or high slot
Average distance
19.6 ft
from the centre of the net
Rebound attempts / 60
1.36
5 this season · all situations
Shot types
27 attempts
Type
Share
Sh%
G−xG
Backhand
29.6%
50.0%
+1.4
Wrist
29.6%
0.0%
−0.7
Tip-in
22.2%
33.3%
+0.7
Snap
11.1%
0.0%
−0.2
Bat
3.7%
—
0.0
Deflected
3.7%
—
−0.2
Share is context, not a grade. Sh% counts shots on goal only.
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
10:00
—
2
1
3
0.86
Power play
0:02
—
1
0
1
94.74
Penalty kill
0:18
—
0
0
0
0.00
Other
0:10
—
0
1
1
16.51
All situations
10:29
—
3
2
5
1.36
Where the 5 points come from · All situations
Goals3 · 0.82/60
Primary assists— · —/60
Secondary assists— · —/60
Zone starts — offensive of OZ + DZ
Offensive —Neutral —Defensive —
Shifts
— a game · — average
Zone starts: five-on-five · Shifts: all situations
Faceoffs
All situations · zones are read from his side of the draw · values only; no peer ranks yet
Win %
46.2%
67 of 145 draws
Draws per game
6.90
21 games
Offensive zone
50.0%
26 draws
Defensive zone
39.6%
48 draws
Power play
66.7%
3 draws
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: 0.86 Peer range: 0.65 – 2.74 / 60
MeanTyson Kozak
Lower Higher (better)
Player 0.86 · mean 1.40. Higher raw values rank higher. Needs 20 appearances and 300 five-on-five minutes (full season: 20 and 300); shot quality and finishing also need 75 unblocked attempts.
Recorded five-on-five goals plus assists per 60 minutes.
Needs 20 appearances and 300 five-on-five minutes (full season: 20 and 300); shot quality and finishing also need 75 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
Metric2024–25ReferenceDiff
0.14——
0.10——
0.24——
0.52——
First covered season: no earlier personal reference is available.
Each game against his own season
Points in each game, coloured by how good the game wasTrailing 10 games, per gameSeason 0.24
