Columbus Blue Jackets
Washington CapitalsThe first night of Buffalo's division title defence went about as badly as it could. A 6-3 loss in Columbus, Ukko-Pekka Luukkonen beaten five times on 22 shots, a power play that went 0-for-5, and Matthew Knies scoring twice in his first game for the other side. Two nights later Buffalo beat Chicago 4-3 and the panic eased slightly.
Both of those games are noise. The question underneath them is real, and the betting market has already put a number on it: VSIN listed Buffalo's season point total at 96.5 in July. For a team that won the Atlantic with 109 points, that is a prediction of a 12-point drop, with two extra games to play. Is it fair?
Should anyone care about a bad opener?
No. Last season Buffalo was last in the Eastern Conference on 27 November, with nine wins and 22 points from 23 games. It then played like one of the best teams in hockey for four months.
| Buffalo, 2025-26 | Games | Wins | Points | 82-game pace |
|---|---|---|---|---|
| To 27 November | 23 | 9 | 22 | 78 |
| December and January | 30 | 22 | 45 | 123 |
| February to the end | 29 | 19 | 42 | 119 |
We tested how often late-November standings decide the playoff race in a separate piece today, and Buffalo was one of the great exceptions: five points out and 16th in the conference, then division champions. A 6-3 loss on 1 October tells you even less than that.
Was last season's 109 real?
Partly, and the lucky part is narrower than it sounds. We have written versions of this before, in July's regression warning and August's argument that the season was half luck. Here is the short form.
Buffalo's five-on-five expected goals share last season was 50.0 per cent, exactly level and in the middle of the league. Its actual goal share was far better, because it scored 23 more goals than its chances predicted at five-on-five, second in the NHL, and its goaltending saved 14.3 more than expected, sixth. Its five-on-five PDO, shooting percentage plus save percentage, was 102.0, fourth highest in the league.
The goaltending side is less worrying than it looks. Buffalo spread its crease across Alex Lyon, Luukkonen and Colten Ellis, and Lyon and Luukkonen were both comfortably above expected across all situations, at 12.6 and 10.5 goals saved. That is two goalies doing it, which is harder to dismiss than one hot run.
The finishing side is where the market's number comes from.
Who did the finishing, and are they still here?
Two of the three are. Among Buffalo's top scorers, these were the biggest finishers relative to their chances last season, across all situations.
| Buffalo, 2025-26 | Goals | Expected goals | Above expected | Where now |
|---|---|---|---|---|
| Tage Thompson | 40 | 33.9 | +6.1 | Buffalo |
| Alex Tuch | 33 | 23.3 | +9.7 | Washington |
| Rasmus Dahlin | 19 | 14.8 | +4.2 | Buffalo |
Tuch scored almost ten goals more than his chances suggested and then left for Washington, a move we looked at in our Capitals piece. Whatever you think of his finishing, Buffalo no longer has to worry about it regressing, because it already left.
That changes the math in an uncomfortable way. A common analytics rule of thumb is that about six goals is worth a win, so roughly two points. If Buffalo's five-on-five finishing returned to neutral, that is around 23 goals, or about seven or eight points. If its goaltending fell back halfway toward average, another two or three. Take ten points off 109, add a little for two extra games, and you land within a point or two of 96.5.
So why say the total is right for the wrong reason?
Because the market seems to be pricing Buffalo as a lucky team due for a correction, and that undersells how the roster has changed. Last season's luck was concentrated in specific players. Some of it walked out the door with Tuch and will not regress on Buffalo's books at all. Some of it is Thompson, whose finishing has a longer record behind it than most. What is left is a team at 50 per cent of chances that is very young and should be improving: Buffalo had the third-youngest minutes in the league by our ice-time-weighted age measure.
Put those together and 96.5 is a fair middle. The path to it is different from the one the panic after the opener implies. The realistic downside needs no collapse at all, only a team playing at its expected level, which is a bubble team in a strong division.
The case for the over, stated fairly
The over has a real argument. Buffalo played at a 119 to 123 point pace for four straight months once it found its game, which is a bigger sample than the first 23 games. Two goaltenders both finished above expected, Thompson's finishing has history behind it, and Dahlin put up 74 points from the blue line. The roster is young and should improve, and two extra games add about two points on their own.
The under has a simpler one. Expected goals had Buffalo at 50.0 per cent while its PDO was fourth highest in the league, which is the profile that drifts back. Its biggest single source of excess finishing, Tuch, now plays for Washington. A 12-point drop from 109 is a normal regression for a team with Buffalo's underlying numbers, before any injury.
96.5 is about where the numbers put Buffalo. The over needs Thompson and the goaltending to repeat. The under needs only for them to be ordinary.
What it means
Ignore the opener. Buffalo was in a far worse spot at the end of November last year and won the division. The things to watch are the ones that decided last season: Thompson's goals against his chances, whether the crease stays above expected, and whether anyone replaces Tuch's finishing. You can follow all three on Thompson's and Dahlin's player pages and the team stats board.
If you made me pick a side, I would lean under by a hair. Buffalo is a good team, but 109 needed a lot to go right, and one of the people it went right for is now in Washington.