ODDPUCK
Between seasonsProjections liveLatest sheet Sat, Apr 11, 2026
all players
Axel Sandin-Pellikka

Axel Sandin-Pellikka

DET·DJersey number #44
Latest captured market
Mar 2 · 2:00 PM ET

Projected · remaining schedule

82 team games · 68 game sample
PTS
25.5
SOG
82.7
G
8.5
A
17.0
PPP
8.4
HIT
21.7
BLK
65.2
PIM
22.0

Each figure is this player's recency-weighted per-game rate over 68 real games × their team's 82 scheduled games. Availability is not projected — injuries, scratches and trades are not modelled anywhere in this project, so the team's schedule is the multiplier, not a forecast of how many games this player will dress for. League board.

What carries into 2026-27

a forecast, not a result

Most of this finishing edge probably won’t repeat

below expectedabove expected
● what happened○ what is projected
2025-26
+2.36
per 100 attempts, vs position
= +3.2 goals over 122
Carries forward
18%
of that rate
measured, 5 seasons
2026-27
+0.43
per 100 attempts, projected
+0.5 goals at the same volume
How this is measured, and what it doesn’t say →

Among defencemen · 508 ranked

Production

Style & role

Percentile against 508 defencemen, 2025-26. Shrunk rate is the per-game rate the rank was computed from, pulled toward the league mean by sample size so a short career does not read as an elite one — so it sits closer to average than the rate actually recorded. This skater’s sample is 68 games. Further right means more, not better: hits, blocks and penalty minutes describe a role, not quality.

Observed rates

per game · recency-weighted, not shrunk · 68 games incl. playoffs
PTS
0.31
G
0.10
A
0.21
SOG
1.00
PPP
0.10
HIT
0.25
BLK
0.79
PIM
0.26

Season totals

regular season
SeasonGPGAPP/GPSOGTOI
2025-2668714210.316816:13

Last 10 · game log

cleared the Mar 2 lineG 1/10PPP 0/10
DateOppSOGGAPPPHITBLKPIMTOI
Apr 15@ FLA000011015:06
Apr 13@ TBL000001017:52
Apr 7vs CBJ30000009:34
Apr 5vs MIN211001215:04
Apr 4@ NYR100000016:51
Mar 6vs FLA200000014:56
Mar 4vs VGK10101009:37
Mar 2@ NSH000000012:16
Feb 28@ CAR000000012:21
Feb 26@ OTT000000011:38

One threshold, applied to all 10: each figure counts how many of the 10 games above would have cleared the Mar 2 posted line. It is not a record against 10 separate lines — this player was priced once, and games after that date are measured against the same number.

Finishing — goals above expected

open play, regular season
SeasonxGGGAxGPAtt
2025-263.87plus 3.2 goals above expected60122

How the 2026-27 projection is made

a forecast, not a result

Finishing persists year over year, but only partly. That share is measured over five seasons of skaters clearing the same 100-attempt floor, and what does not carry regresses toward the rate for the player's position.

Shot volume is not projected, so the goal figure assumes the same number of attempts. A persistent gap means the skater reliably scores above what this model expects — which may be finishing skill or may be shots the model under-rates, and the measurement cannot separate the two. Running hot is a statement about the season just played, not about talent.

This projection is written to a prediction log on the night it is produced, so it can be scored against what actually happens rather than restated later.

Production by strength

G-A-P, regular season
SeasonEven strengthPower playShorthanded
2025-266 goals, 8 assists, 14 points1 goal, 6 assists, 7 points0-0-0

Shot locations

122 attempts · 7 goals · 2025-26
Missed the net (54)Shot on goal (61)Goal (7)

Markets · Mar 2

modeled projections
G
Anytime goal
Line
0.5
Proj
0.0
P(over)
5%
A
Assists
No line posted.
PPP
Power-play points
Line
0.5
Proj
0.0
P(over)
3%
HITS
Hits
No line posted.
BLK
Blocked shots
No line posted.
PIM
Penalty minutes
No line posted.

Per-game rates are recency-weighted over this player's game log. Expected goals come from the xG model over open-play unblocked attempts; totals are not directly comparable across seasons. Every market — anytime goals, assists, power-play points, SOG, hits, blocks and PIMs — is a modeled projection. Every number on this page was worked out overnight and is only read here, never recalculated while you wait.