Atlantic Division: the five biggest projected movers for 2026-27
Five players, ranked on projected points per game against last season: the same measure as our Rate Risers list, cut to one division. What is new is the second half of each number. The chart splits every rise into the stages of the projection that produced it, from last season's finishing luck to the age curve, and the stages add up to the change exactly, so the reason on the page is the model's own arithmetic rather than a story fitted to a number.
Everybody here played 40+ games last season and is projected at 0.45 points a game or better (0.35 for a defenceman).
| # | Player | Team | Pos | Age | 2025-26 P/GP | 2026-27 P/GP | Gain | Biggest driver |
|---|---|---|---|---|---|---|---|---|
| 1 | Auston Matthews | TOR | C | 29 | 0.88 (53 in 60) | 1.09 (82 in 75) | +0.21 | Older seasons (+0.11) |
| 2 | Ivan Demidov | MTL | R | 20 | 0.76 (62 in 82) | 0.94 (75 in 80) | +0.18 | Age curve (+0.08) |
| 3 | William Eklund | OTT | L | 23 | 0.68 (53 in 78) | 0.83 (66 in 80) | +0.15 | Calibration and finishing luck (+0.07 each) |
| 4 | Emmitt Finnie | DET | L | 21 | 0.37 (30 in 82) | 0.50 (40 in 80) | +0.14 | Age curve and finishing luck (+0.04 each) |
| 5 | Brayden Point | TBL | C | 30 | 0.79 (50 in 63) | 0.93 (71 in 77) | +0.13 | Older seasons (+0.07) |
Two of the five are rebounds rather than arrivals. Matthews and Point are both carried by the older-seasons term, which is the model saying last season was the outlier; Demidov, Eklund and Finnie are carried by terms that bet on the player. The distinction matters more than the ranking does.
What moves a projection
- Finishing luck: Last season's goals regressed toward his expected goals.
- Older seasons: His recency-weighted multi-season rate, against last season alone.
- Minor-league record: Translated minor-league scoring blended into a thin NHL sample.
- Sample-size pull: Shrinkage toward the position average, weaker the more games he has.
- Age curve: The league's measured change in scoring from this age to the next.
- Young-player growth: Extra growth measured for players 23 and under with fewer than 120 games.
- Elite retention: The correction that stops top scorers being over-taxed.
- League calibration: Rank-band calibration to the realised shape of league scoring.
- Rate floor: The lowest rate the league has shown at his projected workload.
- Operator floor: A floor we set by hand where the model reads a role too low, plus rounding.
Three kinds of term, and it is worth knowing which one a gain rests on. Finishing luck, older seasons and the sample-size pull are corrections to last season: they say last season was not quite the player. The age curve and young-player growth are a bet on the player, measured on how players of his age and experience have actually developed. Elite retention and calibration are league-level: they fix the shape of the whole projection rather than saying anything about one man. The operator floor is the only term that is our opinion, and it is labelled wherever it appears.
One term to watch before reading the sections. League calibration adds between +0.06 and +0.07 to four of these five, and to most of the division's high scorers. It is a statement about the shape of league scoring, not about any of them, so a gain that leans on it is a thinner claim than the same number built from the age curve. Finnie is the exception: his calibration term is +0.02.
Role is not on the list, and that is a finding rather than an omission. The scoring rate is built from a player's own history, age and sample; the depth chart moves his games and his power-play points, not his points per game. A promotion to the first line changes how many games he is projected for, not this chart, which is why the linemate notes below are context for the number rather than the cause of it.
Player by player
1. Auston Matthews, TOR: a rebound, not a step
Centre, 29. 0.88 to 1.09 points a game (+0.21). 53 points in 60 games last season, 82 projected in 75.
More than half of this is the model preferring Matthews's older seasons to his last one: +0.11, the largest single term in the division. Finishing luck adds 0.04 on top, since last season's goals came in under what his chances were worth. Against that, the age curve takes 0.05 away and elite retention hands 0.04 of it back, the model's way of saying a scorer at his level declines more slowly than the 29-year-old average. Nothing in the stack is development. The other underlying factor at play was his injury history, including the significant missed action last season. This is a projection built on the Matthews of two and three seasons ago, along with a fully injury recovered Matthews and McKenna bump.
Around him: Knies stays on his wing after 433 even-strength minutes together, and rookie Gavin McKenna takes the other slot at a projected 0.65. On the first power-play unit with Tavares, Nylander, Raddysh and Knies, his power-play points go from 12 to a projected 21, a swing second only to Point's among the five.
2. Ivan Demidov, MTL: three terms, two of them about him
Right wing, 20. 0.76 to 0.94 points a game (+0.18). 62 points in 82 games last season, 75 projected in 80.
The age curve is worth 0.08 and young-player growth another 0.04: at 20, with 84 NHL games to his name, he sits inside the under-120 band where the model has measured extra development beyond age alone. Calibration supplies 0.06 more, and the sample-size pull takes 0.02 back against his still-thin record. Of the five, his gain leans hardest on terms that are actually about the player.
Around him: Kapanen continues as his centre after 762 even-strength minutes together, projected at 0.50. Chris Kreider arrives from ANA at 0.60 to fill the left wing. The live question is the first power-play unit, where Demidov joins Suzuki, Caufield, Hutson and Slafkovský, and where the model flags him as the challenger for Slafkovský's seat with 8.2 points riding on it. His power-play points go from 20 to a projected 25 if he holds it.
3. William Eklund, OTT: two near-tied drivers, neither one growth
Left wing, 23. 0.68 to 0.83 points a game (+0.15). 53 points in 78 games last season, 66 projected in 80.
Two terms carry almost the whole rise and they are within a rounding error of each other: finishing luck at +0.066 and calibration at +0.067. The first says last season underrated him; the second says nothing about him at all. The age curve contributes only 0.02, which is unusual for a 23-year-old and the reason this is the softest gain in the five. If you doubt the calibration term, roughly 45 per cent of Eklund's rise goes with it.
Around him: Batherson (0.86) and Stützle (1.13) are both significant upgrades on his even-strength environment in San Jose. He joins the first power-play unit with both, plus Cozens and Sanderson, taking his power-play points from 16 to a projected 19. Last season he played 18:32 a night, 2:54 of it on the power play.
4. Emmitt Finnie, DET: half development, half correction
Left wing, 21. 0.37 to 0.50 points a game (+0.14). 30 points in 82 games last season, 40 projected in 80.
The age curve gives 0.04 and young-player growth 0.02, so 0.06 of the rise is the model betting on a 21-year-old developing on schedule. Finishing luck supplies another 0.04, and calibration 0.02. Read honestly, this is not a pure progression story: a third of it is the claim that last season's finishing undersold him, which is a correction and can be wrong in a way that the age curve cannot.
Around him: the most stable situation of the five. Larkin (0.93) and Raymond (1.02) return after 631 and 614 even-strength minutes with him. He moves to the second power-play unit with Faulk, Copp, Compher and Sandin-Pellikka, worth a projected 9 power-play points against 8 last season.
5. Brayden Point, TBL: the older-seasons bet at 30
Centre, 30. 0.79 to 0.93 points a game (+0.13). 50 points in 63 games last season, 71 projected in 77.
Structurally the same projection as Matthews, four seasons of evidence against one down year, but with less of everything: older seasons at +0.07, finishing luck at +0.03, the age curve costing 0.06 and elite retention returning 0.02. The gain survives on the strength of his pre-2025-26 record, so it is only as good as the assumption that last season was injury and finishing rather than decline.
Around him: Guentzel (1.07) returns on his wing after 505 even-strength minutes together, with Goncalves (0.41) at 283. The power play is where the projection is most exposed, 11 power-play points last season against 21 projected, and Cirelli is pushing for the same first-unit seat, a contest the model values at 6.0 points.
Build your own list
Five is one cut of the division. Below is every Atlantic skater with 10+ games last season: filter by position, club, age, games and projected rate, sort by rise or by drop, and press any player for his waterfall. Not yet arrived applies the breakout rule: under 65 points last season and never 0.85 a game in any of the last three seasons.
| Player | Team | Pos | Age | GP | 2025-26 | 2026-27 | Gain |
|---|---|---|---|---|---|---|---|
| Auston Matthews | TOR | C | 29 | 60 | 0.88 | 1.09 | +0.21 |
| Ivan Demidov | MTL | R | 20 | 82 | 0.76 | 0.94 | +0.18 |
| William Eklund | OTT | L | 23 | 78 | 0.68 | 0.83 | +0.15 |
| Emmitt Finnie | DET | L | 21 | 82 | 0.37 | 0.50 | +0.14 |
| Brayden Point | TBL | C | 30 | 63 | 0.79 | 0.93 | +0.13 |
| Lane Hutson | MTL | D | 22 | 82 | 0.95 | 1.07 | +0.12 |
| Tim Stützle | OTT | C | 24 | 80 | 1.04 | 1.13 | +0.09 |
| Easton Cowan | TOR | R | 21 | 66 | 0.44 | 0.53 | +0.09 |
| JJ Peterka | BOS | R | 24 | 82 | 0.57 | 0.65 | +0.07 |
| Owen Power | BUF | D | 23 | 81 | 0.36 | 0.43 | +0.07 |
| Lucas Raymond | DET | L | 24 | 80 | 0.95 | 1.02 | +0.07 |
| Juraj Slafkovský | MTL | L | 22 | 82 | 0.89 | 0.95 | +0.06 |
| Brady Tkachuk | FLA | L | 27 | 60 | 0.98 | 1.05 | +0.06 |
| David Pastrnak | BOS | R | 30 | 77 | 1.30 | 1.36 | +0.06 |
| Brandon Hagel | TBL | L | 28 | 71 | 1.04 | 1.10 | +0.06 |
Where this is most likely to be wrong
Calibration and elite retention are league-level. They are measured on how the whole league's scoring is shaped and they move every good player a little; a rise that rests mostly on them is the projection correcting its own shape, not a claim about the player.
The age and experience terms are averages. They are what players of his age and games have done, measured across a decade, and the one in front of you can be the exception in either direction.
Linemates are our depth chart, not a coach's. The lines in each story are our projection of camp; shared minutes are measured from last season's shift records. A line that changes in October changes the story, not the number.
Projections are for the 2026-27 season, 84 games. Last-season figures are regular season only. The terms are recorded from the projection's own run and sum to each change exactly; ages are on 1 October. Player sections are written from a fact sheet of our own data and checked against it before they run.