Rate Fallers: the ten biggest projected points-per-game declines for 2026-27
A week ago we published the ten biggest projected points-per-game gains for 2026-27. This is the same measurement pointed the other way.
These are rate declines, not point declines, and the difference is the whole point. Rank on total points lost and you get a list of players who are hurt or getting older and will simply be on the ice less. Rank on points per game and availability drops out of the question entirely, leaving the one a drafter is actually asking: is he going to be worse when he is out there. Darren Raddysh leads at -0.20 a game.
The causes are a combination of ice time, role change, expected goals against actual goals, puck luck, injury bounce-back, and regression in the model itself. Where a name below is largely a finishing story, the expected-goal line is given so you can judge it rather than take our word for it.
The ten
| Player | Role | 25-26 | 26-27 | Drop | Pts | G over xG |
|---|---|---|---|---|---|---|
| Darren Raddysh | TOR · D1 PP1 | 0.96 | 0.76 | -0.20 | -10 | +8.1 |
| Brad Marchand | FLA · L3 PP2 | 1.04 | 0.85 | -0.19 | +1 | +11.1 |
| Evgeni Malkin | PIT · L2 PP1 | 1.09 | 0.93 | -0.16 | -4 | +2.4 |
| Shayne Gostisbehere | CAR · D2 PP1 | 0.91 | 0.77 | -0.14 | +6 | +3.8 |
| Viktor Arvidsson | DET · L3 PP1 | 0.78 | 0.65 | -0.13 | -8 | +4.7 |
| Anthony Mantha | NJD · L1 PP1 | 0.79 | 0.67 | -0.12 | -13 | +15.8 |
| Morgan Geekie | BOS · L2 PP1 | 0.84 | 0.73 | -0.11 | -10 | +16.6 |
| John Carlson | TBL · D1 PP2 | 0.85 | 0.74 | -0.10 | -8 | +3.5 |
| Pavel Zacha | BOS · L1 PP1 | 0.83 | 0.73 | -0.10 | -6 | +10.2 |
| Mark Stone | VGK · L1 PP1 | 1.22 | 1.13 | -0.09 | -1 | +5.7 |
Drop is points per game. Pts is the projected change in the season total, which moves with games as well as rate, and is why a player can fall on this list and still be worth more points than last season. G over xG is last season's goals minus the expected goals on the same shots: how much of the year was finishing.
Darren Raddysh, TOR
Raddysh will be on a lot of these lists this year, and it will be a case of the believers and the non-believers. The projection barely touches his playmaking, keeping 45 assists against the 48 he just posted. It takes apart the goals: 22 of them on 13.86 expected, a 10.4% shooting rate from a defenceman who shot 4.0%, 5.9% and 5.6% in the three seasons before it. The breakout was real, and the loss of Kucherov will still reduce the chances to pile up points, even with four very talented teammates on his power play and some of his even-strength time. Over 84 games he is on a 64-point pace, which is among the league leaders at the position, so this is a haircut and not a claim that the sky is falling.
Anthony Mantha, NJD
The second-largest finishing overshoot on our board after Geekie: 33 goals on 17.18 expected, a surplus of 15.82. He also played 81 games after a run of 67, 74 and an injury-wrecked 13, so this was his healthy season and his lucky season at once, which is the combination the projection is least willing to extend. In fairness he has shot better than 20% in three of his last four seasons, so this is not a player who has never done it.
The concern is the depth of the Devils, or at least the perceived depth from the way Keefe manages the power play and the top nine. Our depth chart has Mantha on the first unit with Jack Hughes, Bratt, Hischier and Luke Hughes, and Meier, Mercer, Brown, Noesen and Hamilton all queued on the second. If he loses that top unit to one of Meier, Noesen or Brown, that is a lot of lost time with Jack Hughes.
Morgan Geekie, BOS
The largest overshoot on the list, a surplus of 16.57 when you weigh his goals against his expected goals. 39 goals on 22.43 expected is not repeatable, but he has now done it twice running: 72 goals against 41 expected across the last two seasons combined. The underlying number has moved with him rather than staying still. His expected goals have gone 8.87, 14.68, 18.22, 22.43, and his shot volume 81, 130, 150, 181. That is four straight years of real improvement, not a spike.
The model still credits him with a 19.3% shooting rate next season, well clear of the 11.1% league rate, so it is shaving the growth rather than saying he will not repeat. That said, he scored in only three of his last 21 games last season after 34 goals in his first 60, which leaves room to believe some regression follows.
What a decline on this list is not
It is not a prediction of collapse. Every player above is projected to be useful, several of them heavily so, and a fall of a tenth of a point a game is closer to a rounding of last season than a verdict on the player. What the list is for is the draft room: these are the men whose price is set by a season the model does not expect them to repeat, which is a different thing from men it expects to be bad.
Notes
Rates are per game. Ranked from the 164 skaters who played at least 40 games last season at 0.60 points a game or better, which is the pool where a decline is worth a drafter's attention. Expected goals come from our own model on the shot events. Roles are from the projected depth chart. Projections are the set the draft portal and the workbook both read.