Every October the same sentence gets written about a dozen clubs: this team is too old, that team is too young, and the window is either closing or not yet open. It is the most reached-for explanation in hockey after goaltending, and it is almost entirely unsupported by the numbers.
Pucky holds a birth date for almost every rostered player, 1,278 of 1,299, and the ice time each of them played last season. Put those together and you can measure something better than the usual roster average, then check it against how the teams actually did. The answer is that it tells you very little, and the places where it does tell you something are not the places anyone is looking.
A better way to measure how old a team is
The usual method is to average the ages on the roster. That is a poor measurement, because it counts the thirteenth forward and the first-line centre equally, and those two players do not have equal effects on anything.
Weighting each player's age by the minutes he actually played fixes it. A club full of young depth players whose veterans play all the important minutes is an old team in every way that matters, and the plain average will not say so.
Run it across the league and the spread is wide: 4.7 years from top to bottom.
And it predicts essentially nothing
Here is the problem for the standard narrative. Line the 32 teams up by weighted age and by how they finished, and the relationship barely exists. The correlation between age and points is 0.16. Against goal differential, which is the better measure of quality, it is 0.22. Both are close enough to zero that you would not bet on them.
The cleanest way to see it is to compare the good teams with the bad ones.
| Group, 2025-26 | Average ice-time-weighted age |
|---|---|
| 100-point teams | 28.83 |
| Teams under 85 points | 28.79 |
| League average | 29.01 |
The best teams and the worst teams are the same age. Not similar: the same, to within about a fortnight.
The individual cases, checked against the 2025-26 standings, are more embarrassing still. Toronto and Los Angeles have the oldest minutes in hockey, 31.34 and 31.31, and Toronto finished on 78 points with a minus-42 goal differential. Colorado has the third-oldest at 31.00 and won the Presidents' Trophy with 121 points and a plus-105 differential, the best in the league. Those two clubs are about four months apart in age and 43 points apart in the table.
At the other end, Chicago has the youngest minutes at 26.61 and finished on 72 points. Montreal is second youngest at 26.72 and finished on 106. Buffalo is fourth youngest and finished on 109 with a plus-48 differential.
So: one of the two oldest teams was bad, the third-oldest was the best team in the league, the youngest was bad, and the second and fourth youngest were very good. Any theory that survives all four of those is not a theory.
What the measurement is actually good for
The number worth having is not the age. It is the gap between the weighted age and the plain roster average, because that gap tells you how much a club leans on its veterans regardless of how the roster looks on paper.
The Rangers are the example worth sitting with. By plain roster average they are 26.59, third youngest in the league behind only Montreal and Chicago. By minutes they are 28.02. The difference of 1.43 years is the sound of a coaching staff that has young players available and is not using them, and that is a far more interesting fact about a club than its age.
Toronto's 1.53 is the largest in the league, which means one of the two oldest teams by minutes is also the team leaning hardest on them. Nashville, at plain 28.50 and weighted 29.97, looks middling and is not. Seattle, at plain 27.50 and weighted 28.79, has the same problem on a smaller budget.
A negative gap means the opposite: the veterans are on the roster but the minutes are going to younger players. Carolina is at minus-0.37, Minnesota minus-0.32, Dallas minus-0.23. Three of the best teams in the league, all quietly handing minutes down.
The case for age as a factor, stated fairly
Where this lands
Nobody seriously argues that a 34-year-old and a 24-year-old age the same way, and individual decline is real and well documented. A single season is also a small sample: Colorado being old and brilliant in 2025-26 does not mean it will be old and brilliant at 32 and 33, and the teams carrying the oldest minutes are the ones with the least room for a bad injury year. Over a long enough horizon, age is the one variable every club is certain to lose to.
As a predictor of this season, it is close to useless. The correlation with points is 0.16, the correlation with goal differential is 0.22, and the 100-point teams and the sub-85-point teams are the same age to within about a fortnight. The oldest team in hockey finished on 78 points and the third-oldest finished on 121. Any explanation that applies equally to Colorado and Toronto is not explaining anything.
What it means
When somebody tells you a team is too old this season, ask them what they expect it to cause. If the answer is that the club will decline over three years, that is a reasonable claim and the weighted-age table is the right place to start. If the answer is that it explains last season's results or predicts this season's, the numbers say otherwise, and they say it clearly.
The more useful question is the gap. A club whose minutes are materially older than its roster is making a choice every night about who plays, and that choice is visible, reversible and worth arguing about. New York has young players and is not using them. Toronto is at the far end of a bet it has been making for years. Carolina, Minnesota and Dallas are going the other way.
That is a conversation about coaching and development, not about birthdays, and it is the one hiding underneath all the lazy writing about windows.