Players Sorted by Defence (v3.0) AKA: Done right this time!

[WordsOfWisdom]

Here we go, I posted the top players in the league, all players on Toronto, all players on Philly, and the worst players in the league. See below:

 

Only the BLUE column matters. Ignore the rest.

 

BEST IN THE NHL:

 

 

 

WORST IN THE NHL:

 

 

FLYERS ONLY:

 

 

LEAFS ONLY:

 

 

What the BLUE column (which I will now officially call "Defence" or "Def%") means: This is the percentage of the player's minutes played at even strength where they did not allow a goal. The closest analogy I can provide is that of Fielding Percentage in baseball. Examine how that gets calculated and this will make perfect sense. This is the hockey equivalent to fielding percentage but for goals allowed. 

 

It takes into account time on ice automatically, and offensive stats are not considered AT ALL. You'll find this addresses all of the problems debated about in previous threads.

 

Take a good look at these figures.

 

Thoughts?    

 

 

 

 

 

 

[WordsOfWisdom]

NOTE: Higher is better.

 

NOTE: A perfect score is 1.000.

 

NOTE: Minimum 15 GP to appear on any of these lists. Anyone else was regarded as a SCRATCH player. 

 

NOTE: Like any percentage-based stat (batting average, fielding percentage, etc.... baseball has loads of them), it begins FUZZY and then becomes CLEAR with playing time. The more ice time someone has, the more accurate their defensive statistic is.

 

NOTE: Average seems to be around .833.  This is not a weighted average and this is not a complete average (because it excluded the scratch players). It's an approximate average. 

 

Edited January 26, 2020 by WordsOfWisdom

[Podein25]

Where are the heat maps? Spreadsheets are so passe, I want pretty pictures

[WordsOfWisdom]

8 minutes ago, Podein25 said:

Where are the heat maps? Spreadsheets are so passe, I want pretty pictures

 

I did make the columns pretty colors though. That counts a little bit!   

[Podein25]

3 minutes ago, WordsOfWisdom said:

 

I did make the columns pretty colors though. That counts a little bit!   

 

I appreciate the effort. Actually, I love a spreadsheet. And charts. Pie charts are the best of the charts.

 

Edited January 26, 2020 by Podein25

[WordsOfWisdom]

I can't speak for Flyers fans, but does that spreadsheet seem like a accurate portrayal of where your guys stand defensively? Does anyone shock/surprise you? (JVR I'm guessing.) 

 

On the Leafs it's still Cody Ceci #1 and by now people need to get used to it, because he's not fading away. Every metric I use has him appearing #1. He's Toronto's best defenceman if you only count defence. Auston Matthews now gets his due as Toronto's best forward without question. We can see now why Engvall is playing so much. Spezza on the other hand, is a weak link defensively and Marincin is not an NHL defenceman. 

 

The .900 mark seems to be the realistic "gold standard" number to strive for. It's like "100" in points. Only 17 players are .900 or higher, and some of them may drop out as the season progresses, making it a very exclusive group. 

 

[WordsOfWisdom]

If you look at the Flyers data compared to Toronto, the real difference is that the Flyers have six players (Myers, Couterier, Hagg, Ghost, Provorov, Vorecek) in the TOP 100 defensively while Toronto only has one (Ceci).  

 

The rest of the rosters are pretty comparable and Toronto's bottom guys aren't quite as bad as Philly's. 

 

What do you think JR?  @JR Ewing

 

 

[JR Ewing]

1 hour ago, WordsOfWisdom said:

What do you think JR?  @JR Ewing

 

 

 

-I think that a defensive stat which begins with offense isn't really a stat which measures a player's defensive ability.

-I think that a defensive stat which is heavily dependent on the quality of the goaltender behind the player isn't really a stat which measures that player's defensive ability.

-I think that a defensive stat which doesn't take into account the different roles that players have is going to leave room for statistical illusions. A player who is buried with a huge number of defensive-zone faceoffs may appear to be worse defensively than a guy who is sheltered with mostly offensive-zone faceoffs.

 

Which players tend to give up fewer shots, fewer scoring chances, and fewer dangerous scoring chances? As an example, Auston Matthews is seen as being TOR's best defensive forward, yet the opposition gets a lot more scoring chances (350) against the Leafs when he's on the ice: more than any other TOR forward (Nylander is #2 at 319), despite being given the lightest amount of zone starts among Toronto forwards. To put this in perspective, Patrice Bergeron's line takes more defensive-zone faceoffs than any other forward group in BOS, but he's given up only 199 scoring chances against.

 

 

[WordsOfWisdom]

39 minutes ago, JR Ewing said:

 

-I think that a defensive stat which begins with offense isn't really a stat which measures a player's defensive ability.

-I think that a defensive stat which is heavily dependent on the quality of the goaltender behind the player isn't really a stat which measures that player's defensive ability.

-I think that a defensive stat which doesn't take into account the different roles that players have is going to leave room for statistical illusions. A player who is buried with a huge number of defensive-zone faceoffs may appear to be worse defensively than a guy who is sheltered with mostly offensive-zone faceoffs.

 

Which players tend to give up fewer shots, fewer scoring chances, and fewer dangerous scoring chances? As an example, Auston Matthews is seen as being TOR's best defensive forward, yet the opposition gets a lot more scoring chances (350) against the Leafs when he's on the ice: more than any other TOR forward (Nylander is #2 at 319), despite being given the lightest amount of zone starts among Toronto forwards. To put this in perspective, Patrice Bergeron's line takes more defensive-zone faceoffs than any other forward group in BOS, but he's given up only 199 scoring chances against.

 

 

 

 

Playing the role of Don Music... WordsOfWisdom:

 

 

 

 

 

In any event, responding to your comments:

 

1. It doesn't consider offence at all now. The source data is the yellow columns in the sheet: GP, ATOI-ES, GA/60.  No goals. No assists. No points. No +/-.  

 

2. I get your point, but it's a team game. We can't completely isolate the player from the team.

 

3. Faceoffs are one part of the game, but defensive play includes the entire duration a player is on the ice, of which the faceoff is one small part of that time... and (depending on when teams change lines) there may not even be a faceoff. The intention of this stat is to be simple. It doesn't get down to that level of granularity although someone could track it that way if they wanted to. 

 

4. Shots are already measured by Corsi/Fenwick, and while shots do indicate scoring chances, they don't account for the quality of those chances. Ultimately, goals are what matters. But I think there is a place for both stats and they should be used together to evaluate defence. 

 

5. Keep in mind the intention here: This is ONE stat. It is one tool used to evaluate a player's defensive play. But there is more than one tool available. No one tool can build you a house. You need several. 

 

 

 

[WordsOfWisdom]

22 minutes ago, WordsOfWisdom said:

Which players tend to give up fewer shots, fewer scoring chances, and fewer dangerous scoring chances? As an example, Auston Matthews is seen as being TOR's best defensive forward, yet the opposition gets a lot more scoring chances (350) against the Leafs when he's on the ice: more than any other TOR forward (Nylander is #2 at 319), despite being given the lightest amount of zone starts among Toronto forwards. To put this in perspective, Patrice Bergeron's line takes more defensive-zone faceoffs than any other forward group in BOS, but he's given up only 199 scoring chances against.

 

By the def% stat yes. 

 

Are they considering time on ice? Matthews plays a lot of minutes.

 

Are those figures totals for the season? If so, then GP becomes a factor. 

 

Bergeron: .838

Matthews: .873

 

87.3% of the time Auston Matthews is on the ice, the opposition is NOT scoring against Toronto.

83.8% of the time Patrice Bergeron is on the ice, the opposition is NOT scoring against Boston. 

 

By this statistical measure, Matthews is doing better. Now if that's because Matthews is dodging defensive zone faceoffs, then we can split up the stat like Corsi/Fenwick do and have a def% score for defensive zone starts and a separate score for offensive zone starts. We can keep going to more and more detail if necessary. 

 

 

Edited January 28, 2020 by WordsOfWisdom

[JR Ewing]

https://www.naturalstattrick.com/playerteams.php?fromseason=20192020&thruseason=20192020&stype=2&sit=5v5&score=all&stdoi=oi&rate=n&team=ALL&pos=S&loc=B&toi=150&gpfilt=none&fd=&td=&tgp=410&lines=single&draftteam=ALL

 

Scoring chances by quality.

 

(sorry I don't have time to say more: am racing online atm)

 

Edited January 28, 2020 by JR Ewing

[jammer2]

On 1/26/2020 at 10:36 AM, WordsOfWisdom said:

 

I did make the columns pretty colors though. That counts a little bit!   

Any stat that has Ceci as TOs best defensemen is totally flawed. The eye test says he is a below average d man and I believe they will have trouble literally giving him away at the deadline.

[WordsOfWisdom]

2 hours ago, jammer2 said:

Any stat that has Ceci as TOs best defensemen is totally flawed. The eye test says he is a below average d man and I believe they will have trouble literally giving him away at the deadline.

 

Not the best OVERALL defenceman on Toronto, but the best DEFENSIVE defenceman on Toronto.  That's a CRITICAL distinction.  This is not a case of "high def% = Norris trophy". This is a case of "high points + high def% = Norris trophy". 

 

Looking at Toronto's defenders, this is what we get for Defensive%:

1. Ceci (.900)
2. Rielly (.865)
3. Dermott (.863)
4. Barrie (.858)
5. Muzzin (.854)
6. Holl (.845)
7. Marincin (.779)

 

Points:

1. Rielly (27)
2. Barrie (27)
3. Muzzin (13)
4. Holl (13)
5. Dermott (8)
6. Ceci (7)
7. Marincin (0)

 

So Toronto's best overall defenceman by total ranking is... Rielly (2+1=3),

followed by Barrie (4+2=6),

Ceci (1+6=7),

Dermott (3+5=8) and  Muzzin (5+3=8),

Holl (6+4=10),

Marincin (7+7=14)

 

 

[jammer2]

28 minutes ago, WordsOfWisdom said:

 

Not the best OVERALL defenceman on Toronto, but the best DEFENSIVE defenceman on Toronto.  That's a CRITICAL distinction.  This is not a case of "high def% = Norris trophy". This is a case of "high points + high def% = Norris trophy". 

 

Looking at Toronto's defenders, this is what we get for Defensive%:

1. Ceci (.900)
2. Rielly (.865)
3. Dermott (.863)
4. Barrie (.858)
5. Muzzin (.854)
6. Holl (.845)
7. Marincin (.779)

 

Points:

1. Rielly (27)
2. Barrie (27)
3. Muzzin (13)
4. Holl (13)
5. Dermott (8)
6. Ceci (7)
7. Marincin (0)

 

So Toronto's best overall defenceman by total ranking is... Rielly (2+1=3),

followed by Barrie (4+2=6),

Ceci (1+6=7),

Dermott (3+5=8) and  Muzzin (5+3=8),

Holl (6+4=10),

Marincin (7+7=14)

 

 

I dont buy it...he might be the worst d man on that team..offense, defense or anything else.

[flyercanuck]

Yaaa...when I look at a list for best defensive players on the Flyers and see Ghost at #4, ahead of Provorov, Voracek at #6,  JVR in the top 10 I gotta laugh. I don't want any one of those three out there with a one goal lead and a minute left.

 

 Ceci, Mathews Reilly as the Leafs top 3? Tyson Barrie top 6? Any eye test will tell you this is completely flawed. 

[JR Ewing]

10 hours ago, WordsOfWisdom said:

 

1. It doesn't consider offence at all now. The source data is the yellow columns in the sheet: GP, ATOI-ES, GA/60.  No goals. No assists. No points. No +/-.  

 

My mistake.

 

10 hours ago, WordsOfWisdom said:

2. I get your point, but it's a team game. We can't completely isolate the player from the team.

 

We can never completely isolate, but we do it where it's completely obvious that it makes a difference. Scenario: two players of equal defensive ability, both playing 1,500 minutes and surrender 500 shots of equal quality.

 

Player A has Vezina-calibre goalie behind him, with a .930 SV%: 35 goals against.

Player B has Kelly Hrudey playing with his bandana over his eyes, and has a .900 SV%: 50 goals against.

 

 

10 hours ago, WordsOfWisdom said:

3. Faceoffs are one part of the game, but defensive play includes the entire duration a player is on the ice, of which the faceoff is one small part of that time... and (depending on when teams change lines) there may not even be a faceoff. The intention of this stat is to be simple. It doesn't get down to that level of granularity although someone could track it that way if they wanted to. 

 

Zone starts give us an idea of the role a player has on the team: it's as old as the game itself. When a coach finds himself in high leverage situations, he puts the players out that he trusts the most. Your system sees Matt Benning as one of the top defensive defensemen in the NHL, yet when the game has been on the line Dave Tippett hasn't once started him in his own. If we stretch that to the last 3 seasons, we see:

 

 

 

Late and close, with the game on the line, Matt Benning has consistently been his coach's last choice. Deployment can tell us a lot about what players bring to the table.

 

 

10 hours ago, WordsOfWisdom said:

5. Keep in mind the intention here: This is ONE stat. It is one tool used to evaluate a player's defensive play. But there is more than one tool available. No one tool can build you a house. You need several. 

 

 

 

 

I completely agree with the bolded portion.

 

[WordsOfWisdom]

8 hours ago, jammer2 said:

I dont buy it...he might be the worst d man on that team..offense, defense or anything else.

 

So why can't opposing teams score goals when he's on the ice?  It's the same Frederick Andersen in goal behind him. 

 

Look, nobody is ever going to say Cody Ceci generates offence.  In fact, he has a lower Corsi/Fenwick score because of the fact that the Leafs aren't generating offence when he's on the ice. Corsi/Fenwick are as much a measure of offensive play as they are defensive play. Corsi/Fenwick is the "+/-" of shots on goal. Thus, a player with a high Corsi/Fenwick score is not necessarily good defensively.

 

 

[WordsOfWisdom]

3 hours ago, flyercanuck said:

Yaaa...when I look at a list for best defensive players on the Flyers and see Ghost at #4, ahead of Provorov, Voracek at #6,  JVR in the top 10 I gotta laugh. I don't want any one of those three out there with a one goal lead and a minute left.

 

 Ceci, Mathews Reilly as the Leafs top 3? Tyson Barrie top 6? Any eye test will tell you this is completely flawed. 

 

Okay, who do you think are the three best defensive players on the Leafs?  

 

[WordsOfWisdom]

1 hour ago, JR Ewing said:

We can never completely isolate, but we do it where it's completely obvious that it makes a difference. Scenario: two players of equal defensive ability, both playing 1,500 minutes and surrender 500 shots of equal quality.

 

Player A has Vezina-calibre goalie behind him, with a .930 SV%: 35 goals against.

Player B has Kelly Hrudey playing with his bandana over his eyes, and has a .900 SV%: 50 goals against.

 

The way to tackle that issue is to either compare def% against fellow team members only or to adjust for the strength of the goalie in the same way that MLB adjusts for ballpark factors and things of that nature. 

 

But overall, unless you're talking about the extremes (best goalie vs worst goalie), I don't think it's that much of a factor. Most goalies are somewhere in the murky middle, and their stats are also affected by the players they have in front of them. It's hard to be a great goalie on a weak defensive team because they're going to give up more chances (thus, higher GAA will result) and they'll give up the most dangerous scoring chances (thus, lower sv% will result). 

 

(lol at Hrudey comment.)

 

1 hour ago, JR Ewing said:

Zone starts give us an idea of the role a player has on the team: it's as old as the game itself. When a coach finds himself in high leverage situations, he puts the players out that he trusts the most. Your system sees Matt Benning as one of the top defensive defensemen in the NHL, yet when the game has been on the line Dave Tippett hasn't once started him in his own.

 

1 hour ago, JR Ewing said:

Late and close, with the game on the line, Matt Benning has consistently been his coach's last choice. Deployment can tell us a lot about what players bring to the table.

 

I think that tells a lot about a coach's personal bias.  You are using a stat based on how a coach uses his players and we just saw how well that worked in Toronto under Mike Babcock. A lot of that has to do with reputation, and who the coach "likes" versus who is seen to be "unproven" in the eyes of the coach. 

 

Who knows? Maybe Edmonton would be doing a lot better if they gave more defensive zone starts to Benning. Maybe they keep trotting out the same guys who aren't getting the job done because the coach refuses to give up on them. (Or conversely, maybe they're doing "good enough" in that role that the coach isn't inclined to change it and see if it can even better than it currently is.)

 

Zone starts is entirely subject to the coach's bias. Look at TOI for example. Matthews was the best scorer on the Leafs and Babcock wouldn't give him top line minutes. If you only went by TOI and said that statistic was an indicator of how great the player was, you would say Matthews wasn't that great, but it was a coaching bias or coaching philosophy that was holding him back and now he's getting more ice time... nearly 2 more minutes on average per game.

 

 

Edited January 28, 2020 by WordsOfWisdom

[flyercanuck]

1 hour ago, WordsOfWisdom said:

 

Okay, who do you think are the three best defensive players on the Leafs?  

 

 

I can't say i follow the team, but I see enough of them due to geography. When I watch Mathews I see a super talented guy who doesn't have a lot of interest in backchecking. Reilly is an offensive defenceman (Barrie is worse defensively) and Ceci is just a bad player.

 

 Maybe the team is built so heavy on offence it's just offensive defensively.

[Podein25]

22 minutes ago, flyercanuck said:

I can't say i follow the team, but I see enough of them due to geography.

 

Condolences

 

24 minutes ago, flyercanuck said:

 Maybe the team is built so heavy on offence it's just offensive defensively.

 

Clever dingo

[WordsOfWisdom]

15 minutes ago, flyercanuck said:

I can't say i follow the team, but I see enough of them due to geography. When I watch Mathews I see a super talented guy who doesn't have a lot of interest in backchecking. Reilly is an offensive defenceman (Barrie is worse defensively) and Ceci is just a bad player.

 

When you look at the Leafs roster, only one Leafs player cracks the TOP 100 in the NHL in def%.  The Flyers have six players. Right away that figure seems to "jive" with our expectations given how much better the Flyers are defensively than Toronto. The Flyers core group defends better than Toronto's. 

 

If you look at Rielly, he's primarily offence, but on this team, he's #2 defensively..... simply because none of Barrie, Holl, Dermott, or Marincin are better. Rielly may not be great (the Flyers have a half dozen players on their roster better in def%), but he's better than the alternatives on this team.

 

Ditto with Matthews. Who would defend better? Nylander? Marner? Tavares? It's just not there. Matthews is the best of a group that in general, does not defend well.   

 

25 minutes ago, flyercanuck said:

Maybe the team is built so heavy on offence it's just offensive defensively.

 

That's pretty much it.   

[WordsOfWisdom]

Here's a good litmus test. Let's compare Toronto to Boston:

 

Boston has EIGHT players in the TOP 100 in def%:

1. Carlo (.898)
2. Krejci (.896)
3. Lindholm (.896)
4. Chara (.896)
5. McAvoy (.890)
6. Grzelcyk (.885)
7. Heinen (.881)
8. Clifton (.880)

With the exception of Lindholm and Clifton, those guys log huge minutes for Boston. 

[WordsOfWisdom]

UPDATE:  I made a huge blunder! 

 

Upon further examination, the GA/60 stat is already giving me what I need. For some reason I wasn't "seeing" that stat for what it is. (Probably doesn't help that I'm getting the information in three completely different formats either but anyways.)

 

I kept thinking GA/60 was how many goals the player was allowing per game at even strength, and then I thought I needed to factor in the time on ice (because players who play more will allow more goals) but GA/60 is already the rate at which the player is allowing goals, so it already takes into account TOI. 

 

In other words, GA = "GA/60" / 60 * TTOI

 

So long story short, chalk this up as another deleted stat.   

 

I think at this point I will "retire" from trying to do "new" stats. They're already all there. There's nothing new to be made. Every aspect of the game is now analyzed, and the data is there for people who want to hunt around long enough to find it.

 

 

/end thread

[radoran]

27 minutes ago, WordsOfWisdom said:

UPDATE:  I made a huge blunder! 

 

It's OK. It's one of the classics:

 

1) don't get involved in a land war in Asia

2) never go in against a Sicilian when DEATH is on the line

3) the GA/60 stat is probably already giving you what you need