Returning regular · 2025-26
Clayton Keller
Utah Mammoth · RW · L shot · age 28 at the dated snapshot.
- Team
- UTA
- Lane
- RW
- Age
- 28
Production and deployment
How the points were built
Official NHL 2025–26 regular-season reports. A missing split stays “—”; it is never converted to zero.
Organizational context
3 RW regulars in UTA
The all-team median is 3.5. UTA’s farm system ranks #3 in the maintained two-board consensus.
Questions a GM still has to answer
- How much of the scoring comes at five-on-five, on special teams, or into an empty net?
- Is the player creating the play, finishing it, or receiving a secondary point?
- Does deployment reveal trust, scarcity, underuse, or a role the roster cannot offer?
- What contract, acquisition cost, development alternative and Cup-window effect complete the argument?
Tonight
UTA do not play today
League schedule for 2026-09-29; next league date 2026-09-30.
Nothing listed against him.
White noise · as of 2026-09-29 · social 2026-09-29
Clayton Keller is talked about more than most (72th percentile); attention is up on the usual
How much he is being talked about: the league's own stories tagged to him, daily pageviews of his Wikipedia article against his own 90-day baseline, posts naming him on the connected networks (Bluesky), and the roster moves and listings that make noise. A measure of attention. Whether it affects a player is his business, and a question the history will let us test later.
Social: 7 posts named him this week on Bluesky · 6 likes, reposts and replies · no baseline yet, it needs a week of readings.
Contract · UTA
$7.15M cap hit
2020–21 through 2027–28. Certified
Both signals › Cheaper than his position's median cost per point AND drives five-on-five scoring or kills penalties.
- 65% of his five-on-five shifts started in the offensive zone, a sheltered deployment that flatters scoring.
- His team shot 12.0% with him on the ice at five-on-five, well above the league norm near 8–9%. Some of the scoring is likely to regress.
- Falling below 58 points at this cap hit would drop the cost test.
Shot quality · 2025-26
Expected goals
CapOrCup's shot model (distance, angle, shot type, strength, rebounds, rushes), trained on 2023-24 and 2024-25 and tested on unseen games (AUC 0.756). Finishing is goals minus expected goals; it is noisy in one season.
CapOrCup measures · 2025-26
What happens while he plays
Measured from the league's play-by-play and shift charts, not from the scoresheet. Expected goals are counted for every player on the ice, and a goal is credited to whoever was on the ice when the sequence began, including players who had changed off before it was scored.
Advanced scoring · 2025-26
Goals, assists and plus/minus, weighted by what they changed
Every goal counts the win probability it added, scaled so the average goal is 1.0: a late tying goal counts several, a goal at 6-1 or into an empty net late counts little. Plus/minus also credits the players on the ice when the goal's sequence began.
Shot map · 2025-26
Where he shoots from, and how it goes
Every unblocked shot placed by distance, angle and shot type. His rate from each spot is set against the league's rate for the same shot from the same spot, so the edge is against everyone else taking that shot, not against an average.
Other spots he uses: slap shot, 30–40 ft, 30–45° off centre — 0 of 17 (-7.7 vs league).
From the video · model judgement
What the condensed games showed
Read from the league's free condensed games (1 game so far): entries, exits, battles and turnovers the model could see and attribute to his jersey number. A condensed game is a sample of the game, so these are observations, not season counts.
Career arc · era-adjusted points per 82 games
Clayton Keller, season by season
Each dot is one season. The amber bar is what his earlier seasons and his age predicted for that season (80% band), so a dot far above its bar is a season the numbers did not see coming. The shaded band is where most players at his position sit at each age.
Position band: quantile regression (10th/50th/90th) of qualifying seasons on an age spline, weighted for players who drop out.
What the numbers found
No shipped label applies. A label is a measured finding about a run of seasons (for example, a spike that came back to the old level), and it is shown only for label types that passed their test on held-out seasons.
Projection at the draft
At the draft the model expected 137 NHL hours through his age-25 season (80% band 56 to 173). He logged 161 hours, between the median and the 90th percentile.
Every season
| Season | Age | GP | Level | Expected (80%) |
|---|---|---|---|---|
| 2016-17 | 18 | 3 | 57.3 | −8.7 to 95.0 |
| 2017-18 | 19 | 82 | 63.2 | 27.6 to 81.0 |
| 2018-19 | 20 | 82 | 45.0 | 54.5 to 85.0 |
| 2019-20 | 21 | 70 | 49.5 | 41.4 to 71.7 |
| 2020-21 | 22 | 56 | 50.3 | 40.1 to 71.7 |
| 2021-22 | 23 | 67 | 70.5 | 40.1 to 71.1 |
| 2022-23 | 24 | 82 | 77.9 | 52.3 to 81.8 |
| 2023-24 | 25 | 78 | 73.8 | 59.7 to 88.9 |
| 2024-25 | 26 | 81 | 86.3 | 59.4 to 88.6 |
| 2025-26 | 27 | 82 | 81.2 | 66.6 to 95.8 |
Source: CapOrCup careers desk (method and league curves) · as of 2026-09-29
Comparables at the same age
Who looked like this at 24 to 27
Careers are compared as league-adjusted points per game by age, over the same ages. The comparables below are the closest lines, and what matters is the spread of what they did next, not any single name.
This player so far: age 14 Bantam 2.15 → age 15 High-MN 1.51 → age 16 USHL 1.16 → age 17 USHL 1.61 → age 18 NCAA 1.45 → age 19 NHL 0.79 → age 20 NHL 0.57 → age 21 NHL 0.63 → age 22 NHL 0.63 → age 23 NHL 0.94 → age 24 NHL 1.05 → age 25 NHL 0.97 → age 26 NHL 1.11 → age 27 NHL 1.07
| Player | Distance (lower is closer) | What he did afterwards |
|---|---|---|
| Mark ScheifeleC · matched at ages 24, 25, 26, 27 | 0.05 | age 28 NHL 1.13 → age 29 NHL 1.04 → age 30 NHL 0.84 → age 31 NHL 0.97 → age 32 NHL 1.06 |
| Artemi PanarinL · matched at ages 24, 25, 26, 27 | 0.08 | age 28 NHL 1.38 → age 29 NHL 1.38 → age 30 NHL 1.28 → age 31 NHL 1.12 → age 32 NHL 1.46 |
| Jake GuentzelC · matched at ages 24, 25, 26, 27 | 0.10 | age 28 NHL 0.94 → age 29 NHL 1.04 → age 30 NHL 1.00 → age 31 NHL 1.09 |
| Jamie BennL · matched at ages 24, 25, 26, 27 | 0.11 | age 28 NHL 0.96 → age 29 NHL 0.68 → age 30 NHL 0.56 → age 31 NHL 0.67 → age 32 NHL 0.56 |
| Tyler SeguinC · matched at ages 24, 25, 26, 27 | 0.11 | age 28 NHL 0.72 → age 30 NHL 0.60 → age 31 NHL 0.66 → age 32 NHL 0.77 → age 33 NHL 1.05 |
| Claude GirouxR · matched at ages 24, 25, 26, 27 | 0.12 | age 28 NHL 0.86 → age 29 NHL 0.71 → age 30 NHL 1.24 → age 31 NHL 1.04 → age 32 NHL 0.77 |
| Jonathan HuberdeauL · matched at ages 24, 25, 26, 27 | 0.13 | age 28 NHL 1.44 → age 29 NHL 0.70 → age 30 NHL 0.64 → age 31 NHL 0.77 → age 32 NHL 0.50 |
| Patrick KaneR · matched at ages 24, 25, 26, 27 | 0.13 | age 28 NHL 1.08 → age 29 NHL 0.93 → age 30 NHL 1.36 → age 31 NHL 1.20 → age 32 NHL 1.18 |
Reading it
If most comparables became NHL regulars and their later peak sits well above this player's current line, the numbers say there is room to grow. If most did not, the numbers say the odds are against it. Either way it is what the data shows, not a name chosen to flatter.
Goal archive ·
Every goal, with the league's video
Preseason, regular season and playoffs are kept apart. Each row links to the NHL's own highlight page; no video is hosted here. Where a clip has been reviewed, the reading of it is a model judgement of the frames, not a measurement.
Loading…
Club listings · licensed feeds
What the club lists him as
An injury listing and a projected slot are the club's statements relayed by a licensed feed, shown with their dates. CapOrCup infers nothing from them.
Projected slot · line 1 left wing, first power-play unit (UTA depth chart as of 2026-09-28; Sportradar NHL API v7, team depth charts)
Availability
Games missed
The NHL publishes no injury feed. This counts games his club played while he did not dress, for any reason: injury, illness, a healthy scratch, a suspension or a trade gap. No cause is claimed.
Either side of that absence: 2.33 points per 60 in the ten games before, 3.09 in the ten after. A rate change over a small sample, not evidence of a lingering injury.
NHL EDGE tracking · 2025-26
Speed, shot and ice
From the NHL's own player-tracking data. Percentiles are the NHL's: the share of the league below this player.