When a veteran fades, the scouts say he has lost a step. The NHL now measures how fast every player skates, and we used five seasons of it to check. The step is not lost. Something else is.
───────────
It is the oldest line in hockey. A player gets into his thirties, the production dips, and the verdict comes down from the press box and the comment section alike: he has lost a step. It sounds obvious. Bodies age, legs go, the game speeds up around you. Everyone knows it.
Since the 2021-22 season the NHL has tracked the position and speed of every player on the ice through chips in the sweaters, and published it as NHL EDGE. That lets us measure the thing the cliche is about: how fast a player's legs actually move, and how that changes with age. We are not the first to look. Aaron Knodell, at Puck Over the Glass, compared players to their own younger selves and found top skating speed holds basically flat from the early twenties to the late thirties. Our five seasons for 1,069 skaters, lined up every player who appeared in back-to-back seasons, and measured how each player changed against himself, year over year. That gave us 2,385 season-to-season comparisons. The picture that came back is not the one the cliche predicts.
───────────
The top gear does not go
Start with the simplest measure of all: a player's top skating speed, the fastest he goes in a season. If players lose a step, this is where it should show. It barely moves. Comparing each player to himself the next year, top speed declines by just 0.18% per year in the 35-and-over group (95% confidence interval ± 0.45, n = 180 season pairs), and even less before that. In plain terms, a player's fastest burst at 36 is essentially the same as it was at 26.
The average top speed of a 24-year-old NHL skater in our data is 22.3 mph. At 32 it is 22.1. At 36 it is 21.9, and at 37 it is 21.8. That is the whole decline, about half a mile per hour across more than a decade of aging. Whatever happens to players as they get older, losing their top gear is not really it. The redline holds.
This matters because it is measured the honest way, each player against his own younger self, not by comparing today's 36-year-olds to today's 24-year-olds. That within-player approach, the same one Knodell used to measure speed, is how aging research avoids its biggest trap, the fact that only the best players survive into their mid-thirties (Lichtman, 2009; EvolvingWild, 2017). More on that trap below, because it makes the next finding stronger, not weaker.
───────────
The mileage does
Now look at how much ground a player covers in a game. Here the story is completely different. Distance skated per game actually rises slightly in the early twenties (up 1.1% per year in the 23-to-26 group), holds roughly flat from 27 to 30, then falls off a cliff. From 31 to 34 it drops 1.99% per year (CI ± 0.93), and in the 35-and-over group it falls 5.88% per year (CI ± 1.43, n = 180). The same players who keep their top speed almost perfectly are covering dramatically less ice.
Put the two together and the dissociation is stark. After age 31, distance skated declines about seventeen times faster than top speed: 3.07% per year versus 0.18%. The gap is statistically overwhelming (Welch t = -7.36, p < 0.001, with 714 season pairs in each group). Shot speed sits in between, fading 0.76% per year in the late twenties and about 1.01% by the mid-thirties, the upper-body power going gently where the leg's top-end speed does not. The flat speed line on its own is not news. Setting it next to the collapse in distance is what changes the story: the engine still hits its redline, it just spends far less time there, and covers far fewer miles doing it.
───────────
Why this is the opposite of what we say
The phrase "lost a step" describes a loss of speed. But speed, the maximal kind, is the one thing these players keep. What they lose is volume and repetition, the capacity to cover ground shift after shift, period after period, night after night. Exercise physiology has a clean distinction for this. Maximal, single-effort power, the one explosive burst, is relatively well preserved with age, while the ability to sustain and repeat high-intensity work declines earlier and faster (Korhonen et al., 2006; Tønnessen et al., 2015). A 2025 study of professional soccer players using wearable trackers found the same split, top sprinting speed holding while high-intensity running volume dropped after the early thirties (Older et al., 2025).
Our data shows that same physiological signature, for the first time, inside actual NHL games rather than a lab or a fitness test. The veteran has not gotten slower in the way we mean when we say it. He has gotten less able to do it over and over. That is a real decline, but it is a decline of endurance and workload, not of the top gear, and the difference is not pedantic. It changes what you would do about it: a player who has lost repeatable volume can be managed with usage, rest, and role. A player who had truly lost his top speed could not be.
───────────
About the survivors
One honest caution. Even measuring players against themselves, the skaters who are still in the league at 36 are the ones who aged well, the ones who declined hardest are already out, so our curves understate how badly the average career ages. But that bias works against the distance finding and still it survives, which makes it more convincing, not less. And it cannot manufacture the dissociation, because top speed and distance are measured on the very same surviving players. If survivorship were inflating preservation, it would inflate both equally. Instead, on identical players, one holds and the other collapses.
There is a second limit worth stating plainly. We have five seasons of tracking data, which is enough to measure year-over-year change across the age range but not to follow a single career from start to finish. As the EDGE record grows, the curves will sharpen. The dissociation, though, is already clear and large.
───────────
What is really fading
If a thirty-five-year-old keeps his speed and his shot, and still slips, then the thing fading is not in his legs. It is in everything around the raw tools: the miles he can give you, the shifts he can repeat, the role the team can still hand him. The body's headline number, top speed, is intact. The decline is happening in the parts of the game that the tools alone do not capture.
That is a quietly important idea for how we judge players, young and old. We reach for the visible, physical explanation, he is slower, because it is easy to picture. The data says the easy explanation is mostly wrong. What separates a veteran from his younger self is not the top speed they share. It is the volume, the durability, and the decisions that fill the space the radar gun never measures. The step was never the story.
───────────
Data and method: NHL EDGE public tracking data, regular seasons 2021-22 through 2025-26, retrieved from the league's public data feed, for 1,069 skaters with at least 20 games played in a season (3,526 player-seasons, 2,385 consecutive-season pairs). Aging effects were estimated with the delta (within-player) method, comparing each player to his own prior season and averaging the changes within age bands, the standard guard against the survivorship bias in cross-sectional aging curves. The per-band figures, confidence intervals, and significance test are reported in the text above. Interactive figures built with Plotly.js. Full data and code available on request.
───────────
References
Brander, J. A., Egan, E. J., & Yeung, L. (2014). Estimating the effects of age on NHL player performance. Journal of Quantitative Analysis in Sports, 10(2), 241-259.
EvolvingWild (2017). A new look at aging curves for NHL skaters (Parts 1-2). Hockey Graphs. hockey-graphs.com.
Knodell, A. (2024). NHL EDGE: Skating Speed, Part 4 (aging: max speed and burst rates by age, within-player). Puck Over the Glass. puckovertheglass.substack.com.
Korhonen, M. T., Cristea, A., Alen, M., Hakkinen, K., Sipila, S., Mero, A., Viitasalo, J. T., Larsson, L., & Suominen, H. (2006). Aging, muscle fiber type, and contractile function in sprint-trained athletes. Journal of Applied Physiology, 101(3), 906-917.
Lichtman, M. (2009). How to properly calculate aging curves (the delta method). The Book Blog / MGL on baseball.
National Hockey League (2023). NHL EDGE: Puck and Player Tracking data. nhl.com/edge.
Older, J. and colleagues (2025). Variations in physical capabilities of professional football players by chronological age, measured with wearable tracking (Catapult). Journal of Functional Morphology and Kinesiology, 10(4), 385.
Schuckers, M., and colleagues. NHL aging curves using functional principal component analysis. Working paper, Simon Fraser University / St. Lawrence University.
Tonnessen, E., Svendsen, I. S., Olsen, I. C., Guttormsen, A., & Haugen, T. (2015). Performance development in adolescent track and field athletes. PLoS ONE, 10(6), e0129014.
Vollman, R. (2016). Stat Shot: The Ultimate Guide to Hockey Analytics. ECW Press.