Why is the Norris focused more on offense than defense?

[oXo Cube]

No, it doesn't equate to actual possession. It's used as a close approximation of it.

Zone time isn't an actual measure of possession either, if you want to get technical.

Do you have a point to make with this narrative or are you just being obtuse for the sake of it?

 

[DANTHEMAN1967]

Actually there is.

The traditionalists just don't like them because they also say that Karlsson is the best defender on earth.

I think most people would say that one of a defenseman's primary jobs is to keep the other team from scoring on your own team while you are on the ice.
Erik Karlsson led the League last year with being on the ice for the most goals being scored on his own team (121 goals).
Best defender on the planet...I don't think so!

 

[oXo Cube]

I think most people would say that one of a defenseman's primary jobs is to keep the other team from scoring on your own team while you are on the ice.
Erik Karlsson led the League last year with being on the ice for the most goals being scored on his own team (121 goals).
Best defender on the planet...I don't think so!

Brilliant analysis.

Everyone is now convinced that Erik Karlsson is actually a horrible player. Well played.

 

[Finlandia WOAT]

It's not even "offense", it's power play production. And because few members of the phwa put in the legwork to watch all 31 teams.

 

[StoneHands]

No, it doesn't equate to actual possession. It's used as a close approximation of it.

Zone time isn't an actual measure of possession either, if you want to get technical.

I never said it was an exact measurement but zone time measures possession a lot better than how often a team shoots the puck.

 

[aufheben]

I think most people would say that one of a defenseman's primary jobs is to keep the other team from scoring on your own team while you are on the ice.
Erik Karlsson led the League last year with being on the ice for the most goals being scored on his own team (121 goals).
Best defender on the planet...I don't think so!

Stopping pucks is the goalies job, sans last-ditch efforts by defenders e.g. Faulk (?) the other night. Preventing scoring chances isn't even the sole job for defensemen anymore, except in extreme cases.

Overall impact is IMO the bottomline. If Brent Burns is directly involved in 76 goals, for another player have a greater positive impact, that player would probably have to be elite in either defense/offense and very good in the other.

 

[CantLoseWithMatthews]

The truth is, offense just matters more. You're not going to get scored on while the opponent is running around in their end and fishing the puck out of their net. Even if someone is really good at playing in the defensive end, you're still stuck playing defense. That's not a good thing

 

[CantLoseWithMatthews]

No a good defence is the best defence.

If you have the puck alot and create alot of chances, but just cough up high quality chances the other way, you still won't be good defensively.

Quantity matters more than you're saying here. There was a recent analysis done that shows it's actually more valuable than quality

 

[Kairi Zaide]

No. quite incorrect. It does not "isolate" the player effect from the team effect.

You're a prime example of someone who doesn't understand the numbers you're looking at.

In fact, using CORSI rel, there is MORE of a team effect, not less. You're literally using your own team as a benchmark.

I don't think you understand how statistics work. And you missed "player effect + other effects".

In stats, you can describe a result R as a combination of effect, say the player, a1, the team, a2 and other effects, a3, all weighted with weights w1, w2 and w3.
You have individual effects and combinatory effects. In short, it can be modelized, simply put, as r =w1a1 + w2a2 + w3a3 + w12a1a2 + w13a1a3 + w23a2a3 + w123a1a2a3. Using rel stats "eliminates" (roughly, statistically) w2a2, which has a more or less considerable effect depending on the stat (for Corsi, one of the biggest factor is team/coach). It's not perfect, but it's a way of eliminating variance introduced by factors out of a players control on his play.

 

[DANTHEMAN1967]

Brilliant analysis.

Everyone is now convinced that Erik Karlsson is actually a horrible player. Well played.

I see you have drunk the "Karlsson Kool-Aid" and are firmly in the group which refuse to see any of his flaws, good for you!

I hope that whichever team you cheer for, as long as it isn't the Leafs, sign him for 8 years at $11+ million on July 1st.

 

[ImporterExporter]

Simple answer.

People are lazy.

 

[Kairi Zaide]

Quantity matters more than you're saying here. There was a recent analysis done that shows it's actually more valuable than quality

Yep, the biggest difference between players production will generally be in their ability to create opportunities (for which shots is one of the proxy), since most players fall between a narrow individual shooting % range. That's the predicate of several advanced stats (expected stuff) and explains why those stats do a good job of predicting. Quality does have an effect but the quantity of that quality will never be enough to offset the effect of quantity unless your name is Paul Byron.

 

[TheMule93]

because its hard to quantify defense

 

[joe dirte]

I think most people would say that one of a defenseman's primary jobs is to keep the other team from scoring on your own team while you are on the ice.
Erik Karlsson led the League last year with being on the ice for the most goals being scored on his own team (121 goals).
Best defender on the planet...I don't think so!

He leads the league in goals against at ES nearly every year.

What you're looking at is total goals. And yes, he had more than many who play a lot of PK.

 

[joe dirte]

I never said it was an exact measurement but zone time measures possession a lot better than how often a team shoots the puck.

Sure, I agree. You made the comment that one measure isn't an exact measure of possession. then seemingly offered up another, which doesn't either. Just pointing that out.

 

[joe dirte]

I don't think you understand how statistics work. And you missed "player effect + other effects".

In stats, you can describe a result R as a combination of effect, say the player, a1, the team, a2 and other effects, a3, all weighted with weights w1, w2 and w3.
You have individual effects and combinatory effects. In short, it can be modelized, simply put, as r =w1a1 + w2a2 + w3a3 + w12a1a2 + w13a1a3 + w23a2a3 + w123a1a2a3. Using rel stats "eliminates" (roughly, statistically) w2a2, which has a more or less considerable effect depending on the stat (for Corsi, one of the biggest factor is team/coach). It's not perfect, but it's a way of eliminating variance introduced by factors out of a players control on his play.

Again, it does not isolate his individual effect from his teams. Quite the opposite. Using the players team as a benchmark is actually a source of variance from team to team.

 

[Regal]

And yet, a grinder like Claude Lemieux can win the Conn Smythe.

If anyone can step up and be of ultimate value in the playoffs, it's probably safe to say the same could theoretically happen in the regular season.

Lemieux was a former 40 goal scorer who had 13 goals in 20 games that playoffs. He won mainly because of his scoring. If Ekholm suddenly becomes a 60 point defenseman, he would get consideration for the Norris as well.

 

[MadLuke]

One difference between shutdown defense value and offense/all-around value, is that a shutdown defenseman value is at max has good has is opposition, Vlasic value can jump high if he has McDavid/Crosby to shutdown, it goes down has the opposition goes down.

 

[Bank Shot]

Quantity matters more than you're saying here. There was a recent analysis done that shows it's actually more valuable than quality

If you are talking about corsi, the correlation between success and corsi has actually been trending down the last couple of seasons.

 

[StoneHands]

Sure, I agree. You made the comment that one measure isn't an exact measure of possession. then seemingly offered up another, which doesn't either. Just pointing that out.

You put exact percentages in your post implying that a 55% CORSI means that a players team has the puck for 55% of the time and that's what I said is statistically inaccurate. I brought up zone time in a different paragraph which means it was a different thought and never said that zone time measures possession. I simply said I that I think it should be a commonly used stat. Had I said that a player with 11 minutes of offensive zone time out of 20 minutes of ice time means they had possession for 55% of the game, that would have been inaccurate.

 

[joe dirte]

You put exact percentages in your post implying that a 55% CORSI means that a players team has the puck for 55% of the time and that's what I said is statistically inaccurate. I brought up zone time in a different paragraph which means it was a different thought and never said that zone time measures possession. I simply said I that I think it should be a commonly used stat. Had I said that a player with 11 minutes of offensive zone time out of 20 minutes of ice time means they had possession for 55% of the game, that would have been inaccurate.

okay. Sure, it's another stat. And indicates another thing. That can be interpreted in more different ways, and requires context.

This is why things get difficult. So many stats, and they just tell one small piece of the story.

 

[Kairi Zaide]

Again, it does not isolate his individual effect from his teams. Quite the opposite. Using the players team as a benchmark is actually a source of variance from team to team.

Sorry for the ugly graph, but I don't want to do an ANOVA or anything on that because it's clear that teams do have an effect regarding CF% (probably the same for +/-, GF%, xGF%, etc.).

If you remove that, you partly eliminate one of the many sources of bias on how good a player is and get closer to how beneficial he is to his team. Does it eliminate all the other sources? Nope, but it's a start, and the easiest to remove. Do you completely remove the effect of teams by using rel stats? No, because, as I said, combinatory effects also exist (the effect of a player is different from the effect of a team which both are different from the effect of a player and the team combined)

For Karlsson, I saw you mentionned his terrible GA last season. But, he also happened to have one of the best RelGF% in the league, which means that when he's on the ice, his team was more likely to score and/or the opponent is less likely to score.

 

[joe dirte]

Sorry for the ugly graph, but I don't want to do an ANOVA or anything on that because it's clear that teams do have an effect regarding CF% (probably the same for +/-, GF%, xGF%, etc.).

If you remove that, you partly eliminate one of the many sources of bias on how good a player is and get closer to how beneficial he is to his team. Does it eliminate all the other sources? Nope, but it's a start, and the easiest to remove. Do you completely remove the effect of teams by using rel stats? No, because, as I said, combinatory effects also exist (the effect of a player is different from the effect of a team which both are different from the effect of a player and the team combined)

For Karlsson, I saw you mentionned his terrible GA last season. But, he also happened to have one of the best RelGF% in the league, which means that when he's on the ice, his team was more likely to score and/or the opponent is less likely to score.

Okay this obviously over your head. Yes there is a correlation to CF% to winning.

Corsi rel does not by any stretch of imagination "remove" the effect your team has on your CF.

Do you understand what the "rel" part is?

 

[ole ole]

The best defenceman is the player who plays the position of defenceman and best contributes to his team's success. There is no set formula for what this looks like, but offensive contributions are, and should be, very important to this. You aren't helping your team win if no goals go in while you're on the ice, you're just helping them not lose.

True but if your on the ice for more goals against compared to how many your on the ice for goals scored than your not really helping your team win.

 

[Kairi Zaide]

Okay this obviously over your head. Yes there is a correlation to CF% to winning.

Corsi rel does not by any stretch of imagination "remove" the effect your team has on your CF.

Do you understand what the "rel" part is?

I'm trying to find where I mentionned the word "winning", but I can't.

"Rel" is "Relative to his team".