Plus-Minus: Hockey's Worst Beloved Stat
January 20, 2026 · NHL · 5 min read
Of all the meaningless statistics that sports fans cling to like a security blanket, hockey's plus-minus might be the most aggressively useless. It's been on the back of hockey cards for decades. It has its own column on NHL.com. Broadcasters cite it with the confidence of someone who has never once questioned whether it measures what they think it measures. And what it measures, primarily, is "which team's goalie is better" and "whose coach lets them start in the offensive zone." Truly, the pinnacle of individual evaluation.
How Plus-Minus Lies To You
The concept sounds reasonable enough: you're on the ice when your team scores, you get a +1. You're on the ice when the other team scores, you get a -1. Simple! Also completely contaminated by:
- Your goalie — If your goalie stops 94% of shots instead of 91%, your plus-minus looks dramatically better through zero effort of your own
- Your linemates — Playing with Connor McDavid tends to improve your numbers. Shocking revelation, I know
- Zone starts — Players who start 70% of shifts in the offensive zone will obviously have better numbers
- Score effects — Trailing teams generate more shots, creating noise
- Power plays are excluded but penalty kills aren't — an inexplicable asymmetry that further distorts everything
A Case Study In Absurdity
| Player | Plus-Minus | GAR | What Happened |
|---|---|---|---|
| Player A | +32 | 4.2 | Played with elite linemates, elite goalie |
| Player B | -8 | 12.1 | Played tough minutes on a bad team |
| Player C | +19 | -1.3 | Sheltered deployment, rode a PDO hot streak |
Player B is clearly the best hockey player in this group. Plus-minus would have you believe he's the worst. This isn't an edge case — this happens constantly.
The PDO Problem
Here's the thing that really kills plus-minus: it's heavily driven by PDO (on-ice shooting percentage + on-ice save percentage). PDO is almost entirely random over small samples. A player whose team shoots 9% and saves 93% while he's on ice will have a great plus-minus. Next year, when those numbers regress to the mean, his plus-minus will crater — and he'll have done nothing differently.
It's like evaluating a poker player by whether they won their last five hands. Sure, skill matters in the long run, but you're mostly just measuring the cards they were dealt.
Plus-minus is the stat equivalent of judging a chef by the restaurant's Yelp rating. Maybe the food is great. Maybe the ambiance carries it. Maybe some angry customer gave one star because parking was bad. You have no idea, and neither does plus-minus.
What To Use Instead
- GAR/WAR — Goals/Wins Above Replacement models that isolate individual contributions
- xG impact — How a player's presence affects expected goals for and against
- RAPM — Regularized adjusted plus-minus that controls for teammates and opponents
These aren't perfect, but they at least try to answer the right question instead of confidently answering the wrong one.
Related: NHL Coverage · Advanced Metrics Guide