Is Killer Instinct Real?
Some teams, the saying goes, bury you once they have you down. We measured what teams do after a game is decided across 98 seasons, and found no sign that it is a trait.
Hockey has a favorite compliment: "that team has a killer instinct". Once they have you down, they keep going. Contrast this with the team that always seems to ease off the moment a game looks safe.
It's a staple of hockey talk, and until now it has been hard to test. The What If rule changes that, because it marks the exact moment every game was decided. Any goal the winning team scores after that moment is, by definition, one that didn't change the result. So we can finally ask a clean question: do some teams consistently run up the score, and others consistently coast?
What we measured
For every team in every season, we calculated a pile-on rate: how many goals it scored per 60 minutes of game time left after its winning goal, in games it won. Measuring against the clock matters. A game decided in the first period leaves far more time to pile on than one decided in the final minutes, and we didn't want to reward a team for that.
We looked at 1,698 team seasons across 98 NHL seasons, from the mid-1920s through 2025. Shootout goals don't count. The timing measures use only games decided in regulation, where "when was it decided" has a clear answer.
The test is simple. A real trait should carry over: a team with a killer instinct this year should still have it next year. So we compared each team to itself in the following season. On the scale we use below, 1 means perfect carryover and 0 means none. For reference, a team's points percentage carries over at 0.58.
Result one: padding mostly means "good team"
The share of a team's goals that came after the game was decided carries over fairly well, at 0.48. That looks like a trait until you account for how good the team was. Once you remove team quality, the carryover drops to 0.02.
In other words, the share of goals scored after a game is decided tells you how good a team is. It doesn't tell you how that team plays.
Result two: the pile-on rate looks like a trait until you account for offense
The pile-on rate carries over at 0.20, a faint but real-looking signal. But teams that score a lot in general also score a lot in the minutes after a game is decided. The two are correlated at 0.57. After removing team quality and how much a team scores per game, the carryover falls to 0.03, and the uncertainty runs from −0.02 to +0.08. That is statistically indistinguishable from none.
We also tried a "relative" version: how fast a team scores after a game is decided, compared with its own normal pace. It carries over at only 0.07, with a range of 0.01 to 0.12. At most, there's a tiny effect.
| Measure | Carries over to next season | After removing team quality | After also removing goals per game |
|---|---|---|---|
| Points percentage | 0.58 | – | – |
| Goals per game | 0.54 | 0.36 | – |
| Share of goals scored after a game is decided | 0.48 | 0.02 | – |
| Pile-on rate | 0.20 | 0.14 | 0.03 |
| When in the game it was decided | 0.05 | 0.05 | – |
| Goals scored when losing, after it was decided | 0.03 | 0.03 | – |
To put the raw gap in plain terms: a typical team has about 13 hours of already-decided hockey in its regulation wins each season. Between the most and least ruthless quarters of teams, the difference works out to roughly four extra padded goals a season. And the control above suggests most of even that comes from how much a team scores anyway.
Result three: when a game gets decided isn't a team trait either
The average regulation game is decided about 56% of the way through, which is around minute 34 of 60. That holds across the whole history of the league. In every decade, the average falls between roughly minute 31 and minute 36, with the 1980s the latest.
And a team's own average, how early it tends to settle games, carries over at just 0.05. From one season to the next, a team's pattern here looks like noise.
What this does and doesn't show
It doesn't show that no coach ever eases off, or that no team has ever buried an opponent on purpose. Individual games and individual teams will do both. What it shows is narrower: there is no consistent, team-level pattern that survives from one season to the next once you account for how good a team is and how much it scores. If killer instinct is real, it is far smaller than the way it gets talked about.
A few honest limits. Empty-net goals count as padding under the rule, and the data can't separate them out, so some of what we call piling on is a team scoring into an empty net. Overtime and shootout games are left out of the timing numbers. We follow teams by their abbreviation, so a franchise that moved and renamed itself breaks its chain. And a result of "no detectable effect" can't rule out a very small one, which is why we show the ranges.
Why it matters for What If Hockey
The What If numbers measure circumstance more than character. A team's share of goals after a game is decided mostly reflects how good it was, and we'd expect the same to hold for players on those teams. That means a big drop on the Biggest Losers list is more a sign of playing for a team that won a lot and scored a lot than evidence that a player ran up the score. We haven't tested that for individual players here, so treat it as a reasonable bet rather than a result.
The rest is on the site to explore. The FAQ explains how "decided" is defined, and Franchise Leaders and Season Leaders show how the rule treats teams and players you know.