Players Sorted by Defence (The "Minus" Score)

[WordsOfWisdom]

At the risk of boring people to tears with this one, I went ahead and copy/pasted the entire league into Excel and crunched the numbers.  I tried to paste the whole sheet here but it crashed, so maybe the TOP 25 will work?  The stat of interest here is DEFENCE, and higher scores are better. I prettied it up a bit by adding 100 to every player, thereby getting rid of all the negative values. Now it kind of resembles points. The "DEF/GP" is exactly that, the defence stat divided by games played, technically it's the minus stat (not shown) divided by games played, and the defence stat is the minus stat with 100 added to it blah blah blah... your eyes glazed over already... it essentially shows the player's trajectory. On second thought, just ignore that column and focus on the yellow one!  

 

I removed players with fewer than 20 GP.  

 

 

 

Thoughts?   

[WordsOfWisdom]

Responding to quick Q&A:

 

Q: Where are the other 800 players?  

A: Lost in cyberspace. The server blows up if I try to post them.

 

Q: Why so many defencemen on this list?

A: They're the best players defensively... which kind of makes sense.

 

Q: Why yellow?

A: It's my go-to color for highlighting.

 

Q: So Brian Dumoulin of the Penguins is the best defensive player in the league?

A: Yes, according to this stat.

 

Q: Where are the Flyers players?

A: Philippe Myers is on the list at #5.  

 

Q: Why can't any of these players score?

A: Dougie Hamilton can. He's on the list at #19. But in general, these players defend.

 

 

 

 

[WordsOfWisdom]

Range Breakdown:

• 1-100 have >= 95
• 101 - 200 have >= 92
• 201 - 300 have >= 88
• 301 - 400 have >= 84
• 401 - 500 have >= 78
• 501 - bottom have >= 52

The greatest spread is top and bottom of the list, with a lot of players clumped close together in the middle.

[WordsOfWisdom]

Here are the BOTTOM 25 players in the NHL:

 

 

 

Some big names jump out on this list too!

 

Also of note, in 561 records analyzed, the average (mean) defence score is ~88.  Given the nature of the stat, it should also diminish slightly over time as the season progresses.

 

Biggest surprise on this list? Probably Connor McDavid. He has 61 points and he's still a -2 player. That's epic-bad defence. 

[JR Ewing]

17 hours ago, WordsOfWisdom said:

Here are the BOTTOM 25 players in the NHL:

 

 

 

Some big names jump out on this list too!

 

Also of note, in 561 records analyzed, the average (mean) defence score is ~88.  Given the nature of the stat, it should also diminish slightly over time as the season progresses.

 

 

17 hours ago, WordsOfWisdom said:

Biggest surprise on this list? Probably Connor McDavid. He has 61 points and he's still a -2 player. That's epic-bad defence. 

 

I'm not sure why it would be a surprise. This system has never seen Patrice Bergeron (career -0.374/gp) as a good defensive player, nor does it see Bob Gainey that way either (-0.235/gp). What hope does the best point producer have in a system which, by default, sees production as a bad thing? Hell, take a look at poor Nathan MacKinnon who, at +5, is supposedly the 12th worst defensive player in the NHL.

 

Just for the hell of it, I tried running through the top 6 finishers in Selke voting last year, just to see how they looked.

1. Ryan O'Reilly, -33, -0.402/gp

2. Mark Stone, -36, -0.468/gp

3. Patrice Bergeron, -29, -0.446/gp

4. Sidney Crosby, -53, -0.671/gp

5. Aleksander Barkvov, -68, -0.829/gp

6. Sean Couturier, -57, -0.713/gp

 

It sees them all bad, but it see Barkov and Couterier, guys universally seen as an outstanding two-way players, as being particularly bad defensively.

 

Note how Crosby, despite being +18 compared to Mark Stone's +17, is disproportionately punished because he outscores Stone by 27 points on the year. Same goes for Barkov, who was -3 on a team which was a collective -199, and Couturier, who went +2 on a team where only 1 other full-time forward wasn't a negative.

 

And which players does the system see as being the best defensive players? A whole lot of 3rd pairing defensemen playing for good teams. In other words, good opportunity to pick up pluses, and won't ruin it by scoring many points.

 

[WordsOfWisdom]

3 hours ago, JR Ewing said:

I'm not sure why it would be a surprise. This system has never seen Patrice Bergeron (career -0.374/gp) as a good defensive player, nor does it see Bob Gainey that way either (-0.235/gp). What hope does the best point producer have in a system which, by default, sees production as a bad thing? Hell, take a look at poor Nathan MacKinnon who, at +5, is supposedly the 12th worst defensive player in the NHL.

 

The point production isn't a bad thing; however, the idea here is to examine how many times a player boosted their +/- through offensive production and remove that... in order to see how they fair defensively.  It's not a perfect system, but a player like Connor McDavid should be a +30 or +40 player if he's taking care of both ends of the ice. If you're scoring 61 points and you have a negative +/-, then you're not defending at all. It's Phil Kessel syndrome all over again. Phil Kessel is a perfect example of the type of forward who adds no value to his team because he causes more goals allowed than what he scores. 

 

3 hours ago, JR Ewing said:

Note how Crosby, despite being +18 compared to Mark Stone's +17, is disproportionately punished because he outscores Stone by 27 points on the year.

 

Precisely.... because Crosby's +/- is 27 points higher than it should be due to that additional scoring. If Mark Stone had an additional 27 points, he would have a +/- of 44.  Or conversely, if Crosby had 27 fewer points, his +/- would be -9.  That's why it works that way.   

 

Players are only "punished" if you will for offence that raises their +/- score. Anything on the power play doesn't count against them. Even strength and short handed points do however.  

 

3 hours ago, JR Ewing said:

And which players does the system see as being the best defensive players? A whole lot of 3rd pairing defensemen playing for good teams. In other words, good opportunity to pick up pluses, and won't ruin it by scoring many points.

 

 

 

[Hockey-78]

Join the GDT:

 

[JR Ewing]

6 hours ago, WordsOfWisdom said:

 

The point production isn't a bad thing; however, the idea here is to examine how many times a player boosted their +/- through offensive production and remove that... in order to see how they fair defensively.  It's not a perfect system, but a player like Connor McDavid should be a +30 or +40 player if he's taking care of both ends of the ice. It's Phil Kessel syndrome all over again. Phil Kessel is a perfect example of the type of forward who adds no value to his team because he causes more goals allowed than what he scores. 

 

 

6 hours ago, WordsOfWisdom said:

Precisely.... because Crosby's +/- is 27 points higher than it should be due to that additional scoring. If Mark Stone had an additional 27 points, he would have a +/- of 44.  Or conversely, if Crosby had 27 fewer points, his +/- would be -9.  That's why it works that way.   

 

Players are only "punished" if you will for offence that raises their +/- score. Anything on the power play doesn't count against them. Even strength and short handed points do however.  

 

 

 

 

 

Plus/minus was always a terrible proxy for defensive play, and any stat which uses its core components for the same purpose suffers from the same issues. It rewards players on good teams with good goalies and punishes those on bad teams or who have crap goaltending. Plus/minus doesn't account for empty-net situations, so good players on teams who play against opposition empty-nets tend to be reward with +/- opportunities and good players on teams who play with their net empty tend to be punished.

 

These two points can't be ignored. In your case of Connor McDavid (who I don't think will be getting Selke votes any time soon, btw) he has been on the ice with .878 ES SV% when the league average is .916. This has cost him about 7 goals so far this year. Empty-nets? He's been on the ice for 7 of those. All of a sudden that -2 is a whole lot more like +12 on a team which is a collective -113 coming into tonight's game because, yes, that's another short-coming of plus/minus: it doesn't account for a player did relative to his team.

 

The problem with plus/minus and any derivative which comes from it is that comes back to goals: both for and against. A player's GF tells us nothing about how he is defensively, and his GA also tells us a lot about the quality of his goaltending and the choices his coach makes. It flies in the face of what coaches say every night about their players: some guys with horrible plus/minus ratings play upwards of 25 minutes and in all of the most important defensive situations. Players with better plus/minus numbers who play for other teams might play 13-15 minutes and in few high leverage situations. Who is the better player?

 

Anyway... We could probably go round and round with this, and we'll probably have to agree to disagree. Any system which routinely rates guys as being poor defensively when they are universally regarded as being excellent needs to be re-thought. If I came out and told everybody that I had a new and better way of telling who the best offensive players were, and they turned out to be a bunch of 3rd-liners and bottom-pairing guys, I'd be laughed out of the room, and rightfully so.

[WordsOfWisdom]

58 minutes ago, JR Ewing said:

Anyway... We could probably go round and round with this, and we'll probably have to agree to disagree. Any system which routinely rates guys as being poor defensively when they are universally regarded as being excellent needs to be re-thought. If I came out and told everybody that I had a new and better way of telling who the best offensive players were, and they turned out to be a bunch of 3rd-liners and bottom-pairing guys, I'd be laughed out of the room, and rightfully so.

 

Rest assured, I'm working on something new.   

 

59 minutes ago, JR Ewing said:

Plus/minus was always a terrible proxy for defensive play

 

Yes, because +/- is not a defensive stat at all. It's an overall score. The idea behind +/- is that the best overall player in the league will have the highest +/-, meaning if you wanted to build a team of the best possible players, you would fill out your roster with the TOP 12 forwards and TOP 6 defencemen in +/-, and that (in theory) would be the best team you could possibly ice.  (Not sure what that would look like but have a go if you feel up to it.)   

 

1 hour ago, JR Ewing said:

The problem with plus/minus and any derivative which comes from it is that comes back to goals: both for and against. A player's GF tells us nothing about how he is defensively, and his GA also tells us a lot about the quality of his goaltending and the choices his coach makes. It flies in the face of what coaches say every night about their players: some guys with horrible plus/minus ratings play upwards of 25 minutes and in all of the most important defensive situations. Players with better plus/minus numbers who play for other teams might play 13-15 minutes and in few high leverage situations. Who is the better player?

 

True, but then every stat in hockey is related to who you play with. Playing with McDavid gets you a lot of points for example. You can never isolate a player's performance down to just him. There's always an element of team in every player's stats.   

 

That's the quality of opposition thing I imagine, and I racked my brain and found a solution for that tonight.  

 

 

[WordsOfWisdom]

I did an updated version for Toronto:

 

 

This time, the stat also factors in TOI. The more you play, the higher your score. It's basically a "counter balance" to what was there before.

 

Using this new version, Cody Ceci is tops on Toronto. Tavares is at the bottom. Rielly gets a nice bump up to #5. 

 

Thoughts?    

[WordsOfWisdom]

Well, I've discovered the reason why players appear to be getting punished for offence:

 

Whenever a goal is scored, all five players on the ice get a (+). However, only 1, 2, or 3 players receive a point. That means there are always at least two players on the ice who are getting a boost to their +/- (and therefore a boost to their Defence score) which is not being subtracted away. 

 

So again, one could argue that if you're on the ice when your team is scoring goals and you're not getting any points out of it, it's probably because you're one of the players who are back defending rather than at the net trying for points.  It's a bit of a lame argument I know but it's the best we have.  

 

So my advice is, when reading this stat, understand that there is a yin-yang type of nature to it. We will always find top scorers near the bottom, and that's okay, because their role is to score goals, not to prevent them. A top point-getter should focus on getting points and never mind this stat. However, if you're a player who doesn't score many points, then you should care about having a high Defence score otherwise you add no value to your team. Also, If you want to compare players using this stat, you will need to compare them position by position.