Why I rarely use plus/minus

Most of you probably know this by now, but I am a big numbers guy when it comes to hockey and sports in general. The human mind is very selective and deceptive, which is why I pay attention to the statistics instead of relying on soley memory to form my opinion after watching games. Stats are not perfect and they never tell the whole story, but they do help fans get a better understanding of the overall picture and present objective, factual information about the events that occur in a game. This is another issue for another post, though. Today, I am going to talk about one stat that I never use on here, which is plus/minus.

Plus/minus, or +/- as it’s often referred to as, is pretty simple stat to calculate as it takes the difference of the number of goals for and against a player is on the ice for at even strength. It is also one of the most widely used stats in the general hockey community as you’ll see just about every sports site list a hockey player’s +/- rating on his player card. Hockey commentators also often reference a player’s plus/minus in broadcasts and preach it’s importance at times. Hell, there are even fantasy leagues that use it as a scoring method. 

So, why do I never use plus/minus? Because I feel that it tells you absolutely nothing about a player in the grand scheme of things and I’m almost certain that not many people know what exactly this stat tells you about anything. Like I said earlier, it’s an easy stat to calculate since it’s just the differential of goals for and against at even strength but there are plenty of issues that arise with the method of how a player’s plus/minus rating is determined. First, I should mention that empty net goals count towards a player’s plus/minus, so every player that was on the ice for an empty net goal will be penalized or rewarded for it even if they had no role in the play at all.

This brings me to the next issue with plus/minus, which deals with goals in general. Something that is very hard to grasp is that it takes a bit (sometimes a lot) of luck to score in hockey. There are always instances where a goal is scored as a result of a defender turning the puck over or getting his jock handed to him by a forward but there are many other times when random occurrences lead to goals. A good deflection will beat a goalie, a perfect shot might sneak past his glove hand and there are always those goals where the goalie will let one in from long range and be completely at fault for it. With plus/minus, everyone who was on-ice for the team that allowed a goal gets penalized for it even though they may have done nothing to allow a goal being scored against them. The same goes for a positive rating being rewarded to every player on the ice for the scoring team. Plus/minus basically assumes that everyone on the ice is to blame for a goal being scored against their team, which is kind of ridiculous when you consider how much good fortune goes into scoring a goal, especially in today’s NHL.

 

Statistics do not lie, but they are misleading when not taken into context and this always seems to be the case with plus/minus because there are many people who believe this stat tells you how well a player has performed defensively. In theory, that makes sense because a player who is on-ice for a lot of opposing goals at even strength can’t be good defensively, right? Wrong. This fails to take a player’s on-ice save percentage and where he starts most of his shifts into account. If a player is trusted with a heavily defensive role, like Nate Thompson and Adam Hall last season, then there is a good chance that he is going end up with a poor plus/minus based on the fact that he starts most of his shifts in the shadow of his own goaltender. Speaking of which, the performance of a team’s goaltender has a direct impact on a player’s plus/minus because a player getting top-nine minutes is going to be on-ice for fewer goals against if he is playing in front of Jonathan Quick than he will if his team’s goalie is Steve Mason.

This is why most of the “worst” players in the league in plus/minus are guys who play a lot of minutes on bad teams or teams with bad goaltending.

2011

Player Team +/- On-Ice Sv% OZ%
Marco Scandella MIN -22 0.921 43.8
John Moore CBJ -23 0.915 45.6
Nate Thompson TBL -23 0.879 34.2
Shawn Horcoff EDM -23 0.909 43.9
Marty Reasoner NYI -25 0.904 43.4
Brett Clark TBL -26 0.896 45.7
Mark Streit NYI -27 0.891 53.7
Cam Fowler ANA -28 0.888 50.5
Nino Niederreiter NYI -29 0.889 46
Milan Jurcina NYI -34 0.884 51.2

Notice how every player here was either on a bad team (four Islanders players alone) or was given very tough zone starts. It’s also worth mentioning that Marco Scandella is the only player on this list who had an on-ice save percentage that is considered above average, every other player had below average or horrible goaltending whenever they were on the ice at even strength. Is this just a one year thing or has this been the case in all seasons?

2010

Player Team +/- On-ice Sv% OZ%
Andy Greene NJD -23 0.903 54.1
Milan Hejduk COL -23 0.898 53.4
Cam Fowler ANA -25 0.917 50.5
Filip Kuba OTT -26 0.883 54.6
Ilya Kovalchuk NJD -26 0.897 57
Zach Bogosian ATL -27 0.911 47.9
Tyler Bozak TOR -29 0.904 52.4
Erik Karlsson OTT -30 0.905 57.1
David Booth FLA -31 0.889 49.9
Chris Phillips OTT -35 0.901 48

These players weren’t given the burden of difficult zone starts but they were still paying significant minutes on bad teams. With the exception of Cam Fowler, every player on here was on a team that finished in the bottom-10 of the NHL standings. Fowler, Bozak and Bogosian are the only players on this list who didn’t play on a team that finished with a lottery pick. It’s also worth mentioning that every player on this list had an on-ice save percentage that is considered below average, Fowler’s .917 mark being the only one that is close. 

2009

Player Team +/- On-ice Sv% OZ%
Martin Havlat MIN -19 0.896 47
Sheldon Souray EDM -19 0.885 52.6
Michael Del Zotto NYR -20 0.896 58.6
Ryan Potulny EDM -21 0.893 47.1
Taylor Chorney EDM -21 0.912 39.2
Kyle Okposo NYI -22 0.88 48.8
Steve Staios EDM, CGY -27 0.9 50.7
Rod Brind’Amour CAR -29 0.872 53.3
Shawn Horcoff EDM -29 0.891 45.3
Patrick O’Sullivan EDM -35 0.893 49.9

Once again, we have players who may have not been given the toughest minutes (aside from Chorney and Horcoff) but the goaltending they got whenever they were on the ice was very, very poor. You’ll also notice that every player here was on a terrible team as every club these players were on finished in the bottom-ten of the standings. The Edmonton Oilers, the worst team in the NHL by a longshot, had six of the bottom-ten players in plus/minus alone so that should tell you about how much plus/minus depends on team performance. It’s true that some of these players were legitimately terrible (Patrick O’Sullivan and Ryan Potulny being the main culprits) but most of them are just players who logged a lot of minutes on a bad team. Kyle Okposo, Shawn Horcoff and Martin Havlat fit the case there.

2008

Player Team +/- On-Ice Sv% OZ%
Travis Moen ANA, SJS -18 0.926 41.9
Freddy Meyer NYI -19 0.872 36.8
John-Michael Liles COL -19 0.899 52.9
Milan Hejduk COL -19 0.888 46.4
Brad Boyes STL -20 0.887 48.4
Jeff Tambellini NYI -20 0.907 48.1
Vinny Prospal TBL -20 0.907 55.8
Scott Hannan COL -21 0.914 40.2
Rod Brind’Amour CAR -23 0.906 48.8
Brendan Witt NYI -34 0.902 34.4

The same trend applies to 2008-09. With the exception of Moen, every player on here received terrible goaltending when they were on the ice, everyone except Liles had tough zone starts and Moen, Boyes and Brind’Amour were the only players who weren’t on terrible teams at the time. It’s also worth mentioning that the bottom-three teams in the league during this year (NYI, TBL & COL) make up 7/10 of the list.

2007

Player Team +/- On-ice Sv% OZ%
Michal Handzus LAK -21 0.907 43.3
Mike Comrie NYI -21 0.9 51.5
Sam Gagner EDM -21 0.898 52.8
Danny Briere PHI -22 0.909 56.9
Jarret Stoll EDM -23 0.904 36.9
Martin St Louis TBL -23 0.876 61.2
Steve McCarthy ATL -23 0.919 42.5
Brad Richards TBL, DAL -27 0.864 53.8
Dan Boyle TBL -29 0.857 60.7
Radek Bonk NSH -31 0.897 42.4

Tampa Bay, Los Angeles, Atlanta, St. Louis and the Islanders were the bottom-five teams in the league during this season. These five teams made up 3/5 of the bottom-10 teams in +/- during the season. There are some interesting exceptions here as Briere and Bonk were on playoff teams and Briere was used in a very sheltered role on the Flyers during that season. Both were on the bad side of the spectrum in terms of on-ice save percentage, though which is the case for almost every player on the list.

What does this all tell you? That plus/minus tells you more about a player’s team than the performance of a player himself, as most of the time the worst “plus/minus” players in a given year are those who play a lot of minutes on bad teams. This becomes even more obvious when you look at a player’s plus/minus rating over the course of a few years, especially when he changes teams. Michael Ryder’s +/- pattern since the lockout is a good example of this.

Michael Ryder

Team Year +/-
Montreal Canadiens 2005 -5
Montreal Canadiens 2006 -25
Montreal Canadiens 2007 -4
Boston Bruins 2008 28
Boston Bruins 2009 3
Boston Bruins 2010 -1
Dallas Stars 2011 17

Ryder’s plus/minus rating has bounced around a lot over his career but he mostly stayed in the negatives when he was with the Canadiens. He had a very tough year in 2006-07 when he was saddled with a -25 rating and the Habs were not a good team that season either as they finished 10th in the Eastern Conference, 19th overall in the NHL and received below average goaltending. A couple years later when he signed with the Boston Bruins, Ryder saw his +/- rating skyrocket to +28 and this was mostly because he played on a team that received elite goaltending from Tim Thomas AND had the second highest shooting percentage in the NHL that season. Things came crashing back down to Earth for Ryder and the Bruins the next couple of years but he had a bit of a renaissance with Dallas last season, scoring 35 goals and finishing the year with a +17 rating.

Is it possible that Ryder’s “defensive performance” is what caused his plus/minus rating to shift so dramatically on a year-to-year basis or could plus/minus have no predictive power at all be driven mostly by random factors such as team shooting and save percentages? After looking at the performance or Ryder and many other NHLers, I am leaning more towards the latter. Ryder isn’t the only player who this is the case with. Just take a look at the plus/minus of every player since the lockout and you’ll notice a ton of random patterns.

It still might sound crazy but goals are very random occurrences in hockey and shouldn’t be the only benchmark used to judge a player’s performance. Yes, it is true that goals are what ultimately leads to a team winning but it is also important to look at the events that lead to goals, which is why I pay attention to the amount of shots a player is on ice for, where they start most of their shifts, who they play most of their ice-time against and lots of other details to evaluate their performance. In the end, there is no room for plus/minus in my analysis.

Stats courtesy of Behind the Net & NHL.com

Quantcast