Loading...
Hockey's most famous stat is also its most misleading. Plus-minus has been on the back of every hockey card for decades, but it tells you less about a player than almost any modern metric. Here's why it stuck around anyway.
TL;DR
Plus-minus gives you +1 when your team scores at even strength or shorthanded while you're on the ice, and -1 when the opponent does. Power play goals don't count. It's simple, intuitive, and almost useless for evaluating individual players because it measures team results and calls them yours.
The rules are straightforward. Every goal in an NHL game gets classified, and only certain situations affect plus-minus.
Your team scores at 5v5 (or 4v4, 3v3) while you're on the ice. You get +1 regardless of whether you touched the puck.
The opponent scores at even strength while you're on the ice. You get -1 even if the goal was entirely the goalie's fault.
Shorthanded goals count. If your PK unit scores, everyone on the ice gets +1. If the shorthanded team scores against you, that's a -1.
Power play goals are excluded entirely. The logic: having a man advantage already tips the scales, so counting them would be unfair.
Say a forward plays 18 minutes in a game. While he's on the ice, his team scores 2 even-strength goals and allows 1. His team also scores a power play goal with him on the ice.
EV goals for
+2
EV goals against
-1
PP goal
0
Game +/-
+1
Plus-minus has been a core NHL stat since the 1967-68 season. It's intuitive and easy to explain. So why do analytics-minded fans and front offices treat it with suspicion? Because it has fundamental problems that can't be fixed.
Play on a good team, get a good plus-minus. Play on a bad one, get punished. A responsible shutdown center on a rebuilding team will look terrible in plus-minus while doing his job perfectly.
It doesn't account for who you play with or against. A third-liner sheltered with offensive zone starts against weak competition will look better than a top-line center facing the opponent's best every night.
Play in front of a Vezina-caliber goalie and your plus-minus gets a free boost. Play in front of a struggling backup and it tanks. The skater did nothing different, but the stat swings wildly.
Goals are rare events. A player might be on the ice for only 2-3 goals in a given game. Over short stretches, plus-minus is dominated by randomness rather than skill. It takes a very large sample to even begin to stabilize.
The analytics community didn't just abandon plus-minus and move on. They built better tools that try to solve the same underlying question: is this player helping his team outscore the opposition?
Shot attempt differentials at 5v5. Instead of waiting for goals (rare events), Corsi counts all shot attempts. Much larger sample, much faster signal.
Learn more →
Assigns a scoring probability to every shot based on quality. Tells you whether a team or player is generating and suppressing dangerous chances.
Learn more →
An all-in-one metric that estimates total value over a replacement-level player, isolating individual contributions from team effects.
Learn more →
Regularized adjusted plus-minus. Uses regression to control for teammates, opponents, and deployment. The stat plus-minus wishes it could be.
Learn more →
It would be dishonest to say plus-minus is completely worthless. Over a full season or multi-season stretch, extreme plus-minus values do correlate with player quality. A player who's +40 over a season is almost certainly on a good team AND contributing to it. A player who's -30 likely has real problems, even after you account for team context.
The issue isn't that plus-minus is pure noise. It's that the signal-to-noise ratio is terrible compared to what's available now. If you only had one stat to look at, plus-minus over a multi-year sample would tell you something. But you don't only have one stat. So there's no reason to settle.
When it has value
Here's who sits at the top and bottom of the plus-minus leaderboard this season. Notice how team quality dominates the list.
Best +/-
Playing on the best team in the league will do that for your plus-minus.
Ovechkin's line has benefited from a surprising Caps season.
Worst +/-
Rebuilding teams produce bad plus-minus numbers. That's the stat's flaw in action.
Young players on a struggling roster. Their plus-minus doesn't reflect their development.
Plus-minus tracks whether you're on the ice for more goals scored by your team or against it. You get a +1 when your team scores at even strength or shorthanded while you're on the ice, and a -1 when the other team does. Power play goals don't count. That's it. Simple on the surface, deeply flawed underneath.
Because it measures team results and assigns them to individuals. A great defensive forward on a bad team will have a terrible plus-minus no matter how well he plays. A fourth-liner on a dynasty gets carried to a positive rating. It has no context for deployment, quality of teammates, or quality of competition. Modern metrics like Corsi, xG, and GAR do a much better job isolating individual impact.
Front offices have largely moved past it. You won't find an analytics department making roster decisions based on plus-minus in 2026. But it still shows up on stat sheets, broadcasts, and hockey cards. The NHL still tracks it as an official stat, even if its influence on actual decision-making has faded significantly.
Several metrics do a better job. Corsi and Fenwick measure shot attempt differentials. Expected goals (xG) accounts for shot quality. Goals Above Replacement (GAR) and Evolving Hockey's RAPM models attempt to isolate individual contributions while controlling for teammates and competition. None are perfect, but all provide more signal than plus-minus.