Prop LabSpeed, faceoffs and lines
Do Faster NHL Teams Win More? What NHL EDGE Speed Data Says
We built a pregame speed rating for every lineup from NHL EDGE tracking data and tested it against expected goals, wins and closing odds. Speed matters a little, the market mostly knows, and most “style matchup” stories are noise.
By Edgehalla ResearchPublished 7 min read
How we measured team speed
NHL EDGE tracks every skater. The most stable speed number it offers is bursts over 20 mph per 60 minutes. A player’s rate repeats year to year at r 0.92. Team top speed, by contrast, repeats at only about 0.1, because one extreme skate sets it.
For each game, we averaged last season’s burst rates of the skaters who dressed, weighted by ice time. Using last season keeps it pregame-only: the number cannot “know” the game it is predicting.
The rating works. It correlates 0.80–0.91 with the team’s actual bursts that season, but only 0.2–0.47 with team strength. So speed and quality are different things, and we can test them separately.
In 2025-26, Colorado led the league at 32.2 bursts per game, then Edmonton (29.6) and Buffalo (27.0). Calgary was slowest at 14.4, behind Boston (16.4) and Nashville (16.6). You can browse each club on our team hubs.
Does speed win? A little, beyond strength
We controlled for Elo, an expected-goals rating, home ice, rest and the starting goalie. Then we asked whether the faster lineup still did better. It did.
Per one standard deviation of speed gap, a team gained +0.37 points of 5-on-5 expected-goals share in 2022-23 to 2024-25 (7,872 team-games). In the 2025-26 holdout it gained +0.73 points (2,624 team-games). Both were tested against rules we wrote in advance.
Wins are noisier. The plain “speed gap predicts wins” test failed in the discovery seasons (p 0.13) and then passed in the holdout. Grouping teams into tiers shows the same story.
Speed tier matchups: wins beyond team strength
Win rate minus the baseline model’s expectation, in points
Fast teams beat Slow teams by 1.1 points more than expected in 2022-25 and 5.8 points more in 2025-26.
- 2022-25 (discovery)
- 2025-26 (holdout)
The effect has grown every season. Measured in log-odds of winning per standard deviation of speed gap, it went from about zero in 2022-23 to +0.11 in 2025-26. That could be a faster league rewarding speed more, or it could be noise. Either way, do not assume the 2025-26 size will hold.
Speed’s effect on winning, by season
Log-odds of a win per standard deviation of lineup speed gap, beyond the baseline
The speed effect on winning grew from 0.01 in 2022-23 to 0.11 in 2025-26.
Is speed priced into NHL odds?
Mostly. The closing moneyline already correlates 0.40 with the speed gap. But speed still added a little: about +0.054 log-odds per standard deviation beyond the no-vig closing price, roughly 1.3 points of win probability. It had the same sign in both seasons with odds (2024-25 and 2025-26).
Back the Fast team vs a Slow team (moneyline)
Fast teams won 59.6 percent against a fair 56.1 percent, a 3.2 percent return whose range includes zero.
Why only “promising”? The range includes zero. The plain speed-to-wins test failed in the discovery seasons. And both odds seasons overlap seasons we had already studied. That is why we cap it here.
A real verdict needs a live record. With about 280 Fast-vs-Slow games a season, one season would catch a 3.5-point edge only about half the time. Plan on two seasons.
Totals: real effect, already priced
Two fast teams do produce more total expected goals: +0.056 xG per standard deviation of combined speed in 2022-25 and +0.130 in 2025-26. But the closing total already moves with combined speed (r 0.25). Betting the over when both teams were fast won 49.8% (ROI −5.7%), and the under won 49.4% (ROI −5.4%).
Style matchups are mostly stories
Hockey talk is full of style matchups: “they cannot handle fast teams,” “speed beats a heavy forecheck.” We tested both kinds, and neither held up.
What repeats from one season to the next?
Year-over-year correlation of team traits (126 team-season pairs)
Team speed and physical style repeat strongly; a team’s supposed weakness against fast opponents does not.
- “Team X gets beaten by speed” is noise. Team-specific slopes against opponent speed repeat at split-half r −0.02 and year-to-year r −0.07. After shrinking, every team’s slope is about zero.
- Scheme does not change how speed plays. Speed × forecheck, × pinching defensemen, × hitting, × back-to-back and × third period were all null in both stages.
- The best-looking cell was luck. Of 180 speed-by-scheme cells, the strongest scored z 3.68. Shuffled labels produce a best cell of 3.44 at the 95th percentile, and that cell failed the holdout.
- No fantasy angle. Fast forwards did not score more DraftKings points or shots against slow teams. The sign flipped between seasons.
Styles themselves are real. Forecheck, blocking and defensemen joining the rush all repeat at r 0.62–0.65, and partly travel with the coach. They just are not matchup levers.
How to use team speed
- Use it as context, not a system. The faster lineup typically controls about +0.4 to +0.7 points more 5-on-5 xG share per standard deviation of gap than strength alone predicts.
- Lineups matter. The rating is built from who dresses. A scratch or call-up can move a team a tier. Check line combinations today.
- Watch the live record. The Fast-vs-Slow moneyline lean began a public watch list on October 11, 2026, flagging Buffalo against Florida and Ottawa at New Jersey. It is 0–0. Treat those as tracked leans, not picks.
For the other half of our team-play research, see faceoff matchups and line chemistry vs stacking.
Frequently asked questions
Who are the fastest teams in the NHL?
By NHL EDGE bursts over 20 mph per game in 2025-26: Colorado (32.2), Edmonton (29.6) and Buffalo (27.0) led. Calgary (14.4), Boston (16.4) and Nashville (16.6) were the slowest. Lineup changes can move a team between tiers.
Does skating speed matter in the NHL?
A little, beyond team strength. Per standard deviation of lineup speed gap, the faster team gained +0.37 points of 5-on-5 expected-goals share in 2022-25 and +0.73 points in 2025-26, after controlling for Elo, xG rating, home ice, rest and goalie.
Can you bet fast NHL teams against slow teams?
It is a lean we are tracking, not an edge. In 2024-25 and 2025-26, Fast teams beat Slow teams 59.6% of the time against a fair 56.1% (n 562), a +3.2% ROI with a 95% range of −3.8% to +10.5%. It has no live record yet.
Do fast NHL teams go over the total?
They create more chances, but the closing total already accounts for it. Betting the over when both teams were fast won 49.8% with an ROI of −5.7% in our test.
Are some NHL teams bad against fast opponents?
Not in a way that repeats. Team-specific “weak against speed” slopes had a split-half correlation of −0.02 and a year-to-year correlation of −0.07, which is noise.
More in Speed, faceoffs and lines
Sources and method
- NHL EDGE skating and tracking data
- NHL stats (NHL.com)
- The Odds API (historical NHL moneylines, puck lines and totals)
- False discovery rate and the Benjamini–Hochberg procedure
- Wilson score interval
- Empirical Bayes shrinkage
- Responsible Gambling Council (Canada)
- ConnexOntario (1-866-531-2600)
- National Council on Problem Gambling (1-800-GAMBLER)
Research and analysis, not betting advice. Past results do not guarantee future returns; bet only what you can afford to lose.