The NHL's Great Plus/Minus Debate: Useful or Misleading?
When you consult the www.hockeydb.com profile of any player in the NHL, CHL, NCAA, or even European leagues, one column you will find aside from the typical Goals, Assists, Points, PIM & GP is the +/- column.
This is a stat that measures the even strength goals for and against that a player is on the ice for. It excludes power play and penalty kill situations. You do, however, get a plus or a minus for shorthanded goals scored and a plus or a minus regardless of whether your team has pulled the goalie.
From 1983 to 2008, there was an NHL Award for the player with the highest Plus/Minus in the league.
It was clearly meant as a cumulative statistic that was meant to identify long-term trends as you needed to play a minimum of 60 games to qualify for the award.
When I was playing competitive hockey, this was the Emery Edge Award. That ran from 1982-83 (Charlie Huddy) to 1987-88 (late Brad McCrimmon). Oddly enough, the award was named after a painting tool designed to paint straight edges with precision. I just used tape 😊
After a year of being nameless, the award got named after whoever would sponsor it. This included:
1) The Alka-Seltzer Plus/Minus Award (1989 to 1995)
2) The Bud Ice Plus/Minus Award (1996 to 1997)
3) The Bud Light Plus/Minus Award (1998 to 2008)
The last winner of the NHL Plus/Minus award was Pavel Datsyuk. Ever since then, the stats have been tracked with no formal award given. I don’t know whether this is due to the lack of sponsorship, or a determination being made that this award wasn’t significant enough to warrant mention.
More and more, you hear mention of the Plus/Minus stat and how its validity has diminished. This has become more prevalent as analytics such as Corsi For and Against ratings have gathered momentum.
Here are some of the most common criticisms I have heard and unearthed about the Plus/Minus stat:
1) NOT ACCURATELY CALCULATED
In a world where Corsi ratings can be calculated reliably based on time on ice with and without the puck and chances generated in five-on-five situations, I find it very hard to believe that, in a world of instant replay, statisticians can’t count who is on the ice when a puck crosses the line. If they can’t get it right when the play is stopped, what possible hope could they have of getting it right when the play is in motion?
2) DOESN’T FACTOR IN ALL RELEVANT INFORMATION
I would agree on this point in that I don’t think there is a single statistic being collected today that does. It’s impossible. Plus/Minus doesn’t factor in who your linemates are or who your defence partner is or which goalie was playing or even the role you played in any of the goals for and against being scored.
It ignores the strength of your team and how much certain players play versus others. It also doesn’t factor in for good or bad luck associated with outcomes.
Name a statistic that does. Even Corsi for an individual player, For and Against, is the cumulative number of shot attempts by or against a team at net while a player is on the ice. No distinctions are made.
Every statistic being calculated today requires the interpreter to extrapolate from the number to determine what is or is not relevant.
Also, keep in mind that in order to win the former Plus/Minus award for an NHL player, you had to play at least 60 games. That cumulative approach with a minimum benchmark seems to be, at least a cursory attempt, at mitigating one offs, lucky bounces and things beyond a player’s control to paint a truer picture of Emery’s Edger if you’ll pardon the pun.
3) SHOT STATISTICS ARE BETTER PREDICTORS THAN GOAL STATISTICS
This is the debate between the analysts and the purists. It’s a very right or left discussion when a Hybrid approach is likely what is required.
I work in sales. This is a very numbers-oriented position regardless of the industry. In order to evaluate a sales representative’s effectiveness, you would look, not only at what they sell, but also their “activity”. The general belief is that the sales representatives who are most active will have the better long-term success with all other things such as skill and opportunity being equal.
This would be the shot statistics approach to measuring effectiveness. An attempt at making contact with a prospect would be the equivalent of a shot at goal. Eventually, the more of them you generate, the more you stand to score. Again, all other things being equal.
The goal-based approach would be looking at a sales representative’s sales trend quarter over quarter or year over year and predicting success that way. The shot statistic proponent will say this ignores factors such as luck or being in a rich territory that feeds a representative rather than him or her mining it. In the hockey analogy, shot based believers will say this ignores lucky bounces or being on in key scoring situations with the better players to potentially prop you up.
The goal-based believer will say that all other things are not equal and that you need to factor in how good of a shooter a player is and that year over year, a player’s ability to generate offence is not a fluke, but rather is the best indicator of future success.
From a Plus/Minus perspective, the goal-based believer points to those with a trend of generating goals more than allowing them as the primary value to a player.
In reality, both are right. That’s why relying on any one statistic is dangerous. Concluding that you know whether a player is a value add or loss to a team based on their trend of Plus/Minus is presumptuous. If you take that trend and extrapolate relevant factors such as TOI, team place in the standings, shooting % AND Corsi For and Against and you can probably get a fairly good read on a player's overall value and get rid of the outliers. At the very least, it should correspond with what you have seen with your own two eyes.
4) PLUS/MINUS ARBITRARILY EXCLUDES GOALS THAT ARE RELEVANT
To my eyes, this seems very true. We’ve touched on this earlier in the article. This is meant to be a five-on-five or even strength statistic.
Critics mention the fact that, while it excludes power play and penalty killing goals against, it does not exclude shorthanded goals for and against and empty net goals scored when the goalie is pulled from either the plus or minus column.
Again, I don’t debate that. However, because of this, the critics are calling to have the Plus/Minus stat canceled or disregarded. Of course, these are the same people beating the analytics drum. These are the same people who are trying to recreate in the NHL what Billy Beane did with the Oakland A’s. Of course, the sport of baseball is very static and is a much more controlled environment.
Trying to apply this kind of philosophy to a fluid sport like hockey is leading to Paralysis by Analysis and everyone believes their statistic or way is better.
In summary, rather than abolishing the Plus/Minus stat, why not have more Plus/Minus stat? That’s right. The answer isn’t less Plus/Minus stats. The answer is more Plus/Minus stats. The trick is that you have to use them properly.
Why not have:
1) Plus/Minus Global – all goals regardless of category
2) Plus/Minus Even Strength – as it is calculated today
3) Plus/Minus Special Teams – rules out even strength goals
4) Plus/Minus Overtime – excluding power play goals
You could have a playoff equivalent of each of these statistics which is something you don’t see on www.hockeydb.com. Surely those statistics exist and what could be more relevant than comparing a player’s profile in regulation and overtime for both regular season and playoffs? How else would you measure someone’s effectiveness when the chips are down?
Right now, Erik Karlsson is having another Norris Trophy caliber season. Despite having 64 points in 49 games, he is a minus 5. Obviously, when you strip out all of the power play goals for which he is not getting a plus and the penalty kill goals for which he isn’t getting a minus, his even strength game is a minus. Of course, it doesn’t factor in the caliber of his team or the ice time he logs in each game or that his team only has one plus player.
However, if you knew his global Plus/Minus, as well as his even strength Plus/Minus, his special teams Plus/Minus and Overtime Plus/Minus, would this not give you a truer picture of statistical contribution? Perhaps he isn’t the best example as he is an outlier. He leads NHL defensemen in scoring. It doesn’t take a degree in advanced mathematics to know his global goals for and against are going to be favourable.
You don’t need great statistical analysis to know that Connor McDavid, Auston Matthews, Leon Draisaitl, Nathan MacKinnon, Cale Makar and Erik Karlsson are great players. It’s obvious to the naked eye.
Great statistical analysis becomes very helpful when trying to assess player additions to teams during free agency and at the trade deadline or who to give the long-term extension to coming out of entry level. Those aren’t always the easiest decisions to make.
Bringing it back to our beloved Senators, much has been made of the team’s inability to score goals at five on five. They rank second last in the league in goals scored at five on five at 83 yet they have scored 135 goals. Of the players on the active roster right now, 18 are a minus.
No one would dispute that Brady Tkachuk is having a strong season (-13). Many feel that Tim Stutzle should be accompanying him to the all-star game (-12). Claude Giroux has been a homerun free agent signing with 42 pts in 46 games (-4). Mathieu Joseph has been middling since signing his offseason extension for four years (+2).
Looking at that statistic on its own can be misleading. However, Drake Batherson has 39 points in 47 games, yet he is in line to win the “Green Jacket” which is the unofficial award for worst Plus/Minus in the league. That should tell you all you need to know about the Plus/Minus stat. It matters to the players. No one likes to be associated with a large – in the +/- column. As long as it matters to them, that stat is relevant as it motivates performance.
There is validity to any statistic and, if you want to, you can discredit any statistic as well.
The problem I see, from my perspective, is that people are looking for that one stat that does it all or does the most. Teams may be looking for ways to not only validate their scouting, but perhaps circumvent it. That is more dangerous than relying on Plus/Minus stats for player evaluation.
Billy Beane chose a new direction for the Oakland A’s in the static game of baseball. He ignored his scouts and identified the players he wanted based on statistical algorithms. The day the numbers start to replace boots on the ground and genuine player evaluation in the NHL, is the day that organization goes in a different direction….counter clockwise around a toilet bowl.
Yes, ladies and gentlemen, if the answer to Blue Oyster Cult’s “Don’t Fear the Reaper” was more cowbell, then the answer to the NHL’s statistical analysis debate isn’t less Plus/Minus, it’s more Plus/Minus. Take more Plus/Minus and validate it against what is observed on the ice. If the two can be reconciled, fewer bad contracts will be awarded and ultimately bought out, fewer bad trades at the deadline will be made and fewer misses will happen on draft day.
By Pat Maguire | Sens Nation Hockey