Is the Defense actually better? Does it matter if it was?

[LGR4GM]

First, anyone who has said that recently, I am not trying to throw you under the bus. I was generally curious if this was true and below will be what I find. 

Second, I will not be using goals against or goals against average for this. Goals in the NHL are rare events and shots are a better and more stable indicator IMO of how things are going in the defensive zone. I will use expected goals or xG which makes me a slight hypocrite but xG is more stable than goals (hint Dylan Cozens is going to score more next year). 

With that said, I used Natural Stat Trick for this. 2022 refers to the 22/23 season and 2023 refers to this season 23/24. 

Let's start with a breakdown of different types on ice situations: I cut out special teams but they are reflected in "all situations" where we can see they clearly impact things.

I highlighted the categories we were better in by year. Interestingly to start, our shots for % was higher than our corsi for% this season. While the difference between the 2 is what I would call minimal, it was interesting. Basically we picked up about 2% for our SF% which is nice, means we shoot more relative to shots against and in 23 we reduced shots on net overall. 

Let's talk about expected goals (xG). Our xGA in 22 was 187.83 compared to just 172.81 currently. Now, we are a game short with the data but that does show some improvement in our defensive game. However the issue is that the net gain of 15.02 in xGA was counter balanced by 20.55 difference in xGF so we actually lost -5.53xg over the course of the season. This explains why our xGF% is better in 22 by .9 compared to 23. 

Goals are rare though, so let's look at shots. With 1 game not in the data we have to remember that there are about 30 shots missing from 2023. Not all of them would be high danger shots and that is the focus here. I would guess we are missing somewhere between 5-10 high danger shots for and against for this last game. It doesn't make up the difference we see. Last year we had 823 HDSA in all situations compared to 676 this year or a difference of 147, so we did cut down the total number of high danger shots by about 17%. Again, we can certainly label this improvement on defense. Even if we give 10 hd shots to our last opponent we greatly cut this down. The issue again comes when we look at the inverse. in All situations last year we had 725 HDSF but this year we are down 163 shots at only 562 or 22.5% lower. So the math again fails in terms of the overall, we got better defensively but the dip in offense was so bad that is put us in a net negative situation. 

Finally let's look at the HDSH%. We got absolutely and brutally murdered in 5v5 (-3.86%), All Sit (-3.69%), and ES (-4.12%) HDSH% compared to last year. Simply put, we shot on average 3.89% lower in high danger situations. So we took far less HD shots and then we shot almost 4% lower on the ones we took. Welcome to scoring regression. 

Conclusion: Did the defense get better? Yes, I think we can say that the defense was able to limit shots and compared to last year we saw about a 17.9% drop in HD shots for all situations, a 21.1% drop in even strength, and a 22.5% drop at 5v5. This coupled with better overall goaltending (it actually only ends up about 1% higher in terms of sv% over the season shockingly) is the reason we see such a good drop in our goals against (241 this year compared to 297 last year). We have a major problem though. Don Granato has openly said he changed the system, as far back as December I recall him talking about it. The results show a better defense but the offense cratered. All situations saw a 22.5% drop in HDSF, 20.7% drop at even, and a 21.3% drop at even. 

Put together we have the following

All Sit: 17.9-22.5 = -4.6%

ES: 21.1-20.7 = 0.4%

5v5: 22.5-21.3 = 1.2%

Clearly special teams killed us as the all situations goes negative but the improved defense came at such a sacrifice to the offense that the gains were negligible and well within the variance if I cared to calculate one (I don't, I am lazy). Granato failed and while shutting down other teams on offense and getting solid goaltending helped to make up for some of it, the Sabres offense cratered so hard and so badly that it negates any of the defensive gains. I used HD shots because they are correlated to scoring more strongly than most others. For those of you yelling at your tv to get to the slot or stand in front, that is the hd areas where we need more from. The expected goals for % (xGF%) are the final nail in this coffin for me. All of them are better in 2022 compared to this year meaning, that while yes the defense itself got better, because the offense got so bad it doesn't actually matter. We got worse. I would ask that someone who has a hockeyviz account (mine lapsed) to pull the shot heat maps for the Sabres from the last 2 years so help show some of this. 

*sorry this is long

Edited April 15 by LGR4GM

[Thorny]

18 minutes ago, LGR4GM said:

First, anyone who has said that recently, I am not trying to throw you under the bus. I was generally curious if this was true and below will be what I find. 

Second, I will not be using goals against or goals against average for this. Goals in the NHL are rare events and shots are a better and more stable indicator IMO of how things are going in the defensive zone. I will use expected goals or xG which makes me a slight hypocrite but xG is more stable than goals (hint Dylan Cozens is going to score more next year). 

With that said, I used Natural Stat Trick for this. 2022 refers to the 22/23 season and 2023 refers to this season 23/24. 

Let's start with a breakdown of different types on ice situations: I cut out special teams but they are reflected in "all situations" where we can see they clearly impact things.

I highlighted the categories we were better in by year. Interestingly to start, our shots for % was higher than our corsi for% this season. While the difference between the 2 is what I would call minimal, it was interesting. Basically we picked up about 2% for our SF% which is nice, means we shoot more relative to shots against and in 23 we reduced shots on net overall. 

Let's talk about expected goals (xG). Our xGA in 22 was 187.83 compared to just 172.81 currently. Now, we are a game short with the data but that does show some improvement in our defensive game. However the issue is that the net gain of 15.02 in xGA was counter balanced by 20.55 difference in xGF so we actually lost -5.53xg over the course of the season. This explains why our xGF% is better in 22 by .9 compared to 23. 

Goals are rare though, so let's look at shots. With 1 game not in the data we have to remember that there are about 30 shots missing from 2023. Not all of them would be high danger shots and that is the focus here. I would guess we are missing somewhere between 5-10 high danger shots for and against for this last game. It doesn't make up the difference we see. Last year we had 823 HDSA in all situations compared to 676 this year or a difference of 147, so we did cut down the total number of high danger shots by about 17%. Again, we can certainly label this improvement on defense. Even if we give 10 hd shots to our last opponent we greatly cut this down. The issue again comes when we look at the inverse. in All situations last year we had 725 HDSF but this year we are down 163 shots at only 562 or 22.5% lower. So the math again fails in terms of the overall, we got better defensively but the dip in offense was so bad that is put us in a net negative situation. 

Finally let's look at the HDSH%. We got absolutely and brutally murdered in 5v5 (-3.86%), All Sit (-3.69%), and ES (-4.12%) HDSH% compared to last year. Simply put, we shot on average 3.89% lower in high danger situations. So we took far less HD shots and then we shot almost 4% lower on the ones we took. Welcome to scoring regression. 

Conclusion: Did the defense get better? Yes, I think we can say that the defense was able to limit shots and compared to last year we saw about a 17.9% drop in HD shots for all situations, a 21.1% drop in even strength, and a 22.5% drop at 5v5. This coupled with better overall goaltending (it actually only ends up about 1% higher in terms of sv% over the season shockingly) is the reason we see such a good drop in our goals against (241 this year compared to 297 last year). We have a major problem though. Don Granato has openly said he changed the system, as far back as December I recall him talking about it. The results show a better defense but the offense cratered. All situations saw a 22.5% drop in HDSF, 20.7% drop at even, and a 21.3% drop at even. 

Put together we have the following

All Sit: 17.9-22.5 = -4.6%

ES: 21.1-20.7 = 0.4%

5v5: 22.5-21.3 = 1.2%

Clearly special teams killed us as the all situations goes negative but the improved defense came at such a sacrifice to the offense that the gains were negligible and well within the variance if I cared to calculate one (I don't, I am lazy). Granato failed and while shutting down other teams on offense and getting solid goaltending helped to make up for some of it, the Sabres offense cratered so hard and so badly that it negates any of the defensive gains. I used HD shots because they are correlated to scoring more strongly than most others. For those of you yelling at your tv to get to the slot or stand in front, that is the hd areas where we need more from. The expected goals for % (xGF%) are the final nail in this coffin for me. All of them are better in 2022 compared to this year meaning, that while yes the defense itself got better, because the offense got so bad it doesn't actually matter. We got worse. I would ask that someone who has a hockeyviz account (mine lapsed) to pull the shot heat maps for the Sabres from the last 2 years so help show some of this. 

*sorry this is long

Great stuff.

- - - 

I, too am shocked the sv% is only 1 point higher

…and I knew I was right in predicting last offseason that a run back would result in a shell game. 

Edited April 15 by Thorny

[mjd1001]

I think the defense got better this year, but without going into the stats (I like looking at basic/big picture stats but more rely on what I see in association with those stats), I think the improvement started right about January 1 of this year. 

The bigger thing to me is, the way I think they needed to improve defensively, and the way that I did notice they DID starting in January, should have nothing to do with offensive chances or production.  This teams biggest weakness is/was coverage in the defensive zone...the 'half court' defense.

Last year, and for a good part of this year, the forwards on this team would have ZERO idea what to do in their own zone.  I am hard on Cozens because I think he is the biggest offender, but there are others.  The defensemen seem to play where they 'should' play in D-zone coverage, but the forwards either just float around or do whatever they want to do.  

Now again, this may not may not be reflected in the stats you are bringing up, its just an opinion....but when you do watch the replays of the goals allowed by this team, or just watching the game and you see a big save made by UPL to prevent a goal, more often than not that seems to be (or at least WAS) caused by forwards just being awful.   I have stated this many times, but for sake of discussion I will use Cozens as an example.  The puck is in the Sabres zone, the other team has control of it. The D-men are down low either engaging in a puck battle when the puck comes to their side, or covering a spot on the ice near UPL.  The wingers are covering the side boards or the opposing wingers half way up the boards. Cozens SHOULD be guarding the slot.  But what would happen often, for some reason, Cozens just takes off and runs to the boards to chase the puck, often times when a Sabres player is already there....the puck squirts free, now the entire center of the ice, the slot area in front of UPL where Cozens SHOULD be is wide open...the opposing center or a D-man walk into it, puck on their stick, with time to either pick a corner of the net, pass to someone else who is open, or use the D-men and often times Cozens himself as a screen...and there is your goal against.

Now, as I said, the team is not perfect, but a switch was flipped right around January 1 of this year and they got a lot better.  Cozens is still the biggest offender (why I think and beg for him to be a winger, he can chase the puck all he wants on the boards in the D-zone as a winger without causing damage), but other forwards have done it too.  My main point is...this is an area they have improved on, and they can continue to improve on, and it should have little impact on their offensive zone chances. Play run and gun all you want...your D-zone coverage I mentioned above only matters when the opposing team already has control in your own zone.

Again, I'm not sure if the stats support what I see above, I'd be interested to see if there is a difference in quality chances allowed with Cozens on the ice vs Tage on the ice (Tage actually does NOT make those mistakes that much in the D-zone) or to see if there is a difference in chances from early in the year to post January....but...this is my contribution to the discussion.

[LGR4GM]

16 minutes ago, mjd1001 said:

The bigger thing to me is, the way I think they needed to improve defensively, and the way that I did notice they DID starting in January, should have nothing to do with offensive chances or production.  This teams biggest weakness is/was coverage in the defensive zone...the 'half court' defense.

My main point is...this is an area they have improved on, and they can continue to improve on, and it should have little impact on their offensive zone chances. Play run and gun all you want...your D-zone coverage I mentioned above only matters when the opposing team already has control in your own zone.

Again, I'm not sure if the stats support what I see above, I'd be interested to see if there is a difference in quality chances allowed with Cozens on the ice vs Tage on the ice (Tage actually does NOT make those mistakes that much in the D-zone) or to see if there is a difference in chances from early in the year to post January....but...this is my contribution to the discussion.

2 things. 

First, your main point is not correct IMO. What they do in the defensive zone, how they exit the zone, and their ability to play offense are all interconnected and the stats show that. I don't have access to it but the Sabres zone exit data from last year to this year is like the difference between Earth and Saturn. They worked very hard to shut down chances but they did not do a good job of getting up and creating chances at all. The sacrificed offense for defense and they are 100% connected in hockey, especially modern hockey which is predicated on breakouts and off puck movement. We saw them many times take away passing lanes and then pressure the opposition into giving the puck away or taking the puck away and then... nothing. Maybe a quick rush with a shot from the outside and that was it. There were many times this year as well where the defense did not pinch and keep pucks in but were immediately flying the offensive zone. I watched it live on several occasions where Owen Power in particular would be within a Benson stick length of the puck and he would not challenge it but immediately start his back skating to prepare to defend the rush. Not a knock on Owen here, I think and basically have confirmation that it was what Wilford and Granato wanted. 

Second, this isn't about Dylan Cozens and he wasn't bad enough to have such an impact he alone skews the results. More than welcome to look it up but the point is the overall team and one player is not tanking the teams overall data that hard. Cozens cf% drops 0.81, is sf% goes up 0.99, and his xG% drops 3.37%. So yes he is worse but the real issue is in the offensive zone not the defensive zone. his xGA drops 1.31 but his xGF drop 12.59. It wasn't the defense, it was that the offense was so bad it makes the defense bad. 

Edited April 15 by LGR4GM

[LTS]

All good stuff.  I'd be curious  how those stats break down against teams in the top 50% of the league versus the bottom 50%.  Is it a situation where against lower quality opponents they were able to play better defensively and still score or was it even or worse?

The macro level look certainly indicates what we all thought we saw.

All of that aside.. I still saw too many opponents able to drive our net while not enough Sabres doing the same.  Having an improved defense only matters if it translates into winning and it did not.  Now, if they can continue to improve and find the scoring ability again then fantastic. We'll be able to look back and say this season was where they figured out defense.  If it doesn't.. then it didn't matter at all.

Good stuff by the way.  It should be long. That's research.

[Doohickie]

1 hour ago, LGR4GM said:

*sorry this is long

Thank you for using paragraphs.

What your analysis tells me is that Matt Ellis needs to be shot into the sun.  Or at least replaced as special teams coach.

[mjd1001]

1 hour ago, LGR4GM said:

2 things. 

First, your main point is not correct IMO. What they do in the defensive zone, how they exit the zone, and their ability to play offense are all interconnected and the stats show that. I don't have access to it but the Sabres zone exit data from last year to this year is like the difference between Earth and Saturn. They worked very hard to shut down chances but they did not do a good job of getting up and creating chances at all. The sacrificed offense for defense and they are 100% connected in hockey, especially modern hockey which is predicated on breakouts and off puck movement. We saw them many times take away passing lanes and then pressure the opposition into giving the puck away or taking the puck away and then... nothing. Maybe a quick rush with a shot from the outside and that was it. There were many times this year as well where the defense did not pinch and keep pucks in but were immediately flying the offensive zone. I watched it live on several occasions where Owen Power in particular would be within a Benson stick length of the puck and he would not challenge it but immediately start his back skating to prepare to defend the rush. Not a knock on Owen here, I think and basically have confirmation that it was what Wilford and Granato wanted. 

Second, this isn't about Dylan Cozens and he wasn't bad enough to have such an impact he alone skews the results. More than welcome to look it up but the point is the overall team and one player is not tanking the teams overall data that hard. Cozens cf% drops 0.81, is sf% goes up 0.99, and his xG% drops 3.37%. So yes he is worse but the real issue is in the offensive zone not the defensive zone. his xGA drops 1.31 but his xGF drop 12.59. It wasn't the defense, it was that the offense was so bad it makes the defense bad. 

First, I think you are incorrect about me being incorrect. How they exit the zone and their ability to play offense is impacted minimally by their positioning in the Defensive zone.  In my examples, it may actually support my point.  If ANY forward, including Cozens, is covering the slot (as per my example) he would be available more readily to take a pass out of the corner from a D-man and head up ice...rather than when they just start vacating zones and chasing the puck, being a step behind the other team and the other team having possession.

Next, I think I have been in the past, and was in my post saying Cozens was the worst offender but other forwards were offenders just as well.  With that in mind, not sure why you are pointing out to me this isn't about Dylan Cozens?

If you play a better 'half court' defensive (hockey version of it) not only will you be better in your own zone giving up less chances to the other team, but doing just that might actually HELP your breakouts and your offensive chances.  I wasn't going that far to make that point, I was simply saying the forwards having their heads on in their own zone was/is the issue, and you don't need to 'pull back' on what you do in the neutral zone or offensive zone to fix that.

Edited April 15 by mjd1001

[GASabresIUFAN]

HDCA (775 through 81 games ytd vs 881 in 82 games last season) and SCA (1803 vs 1957 last year) are both down.  This is pretty illustrative that the defense is better.  Unfortunately it also proves that the defense remains poor.  Our HDCA is the 6th worst in the NHL (we were 4th worst last season) and our SCA is 12th worst vs 11th last season.  So while the gross numbers are better so was the entire NHL and our position relative to our peers is about the same. 

The Sabres have needed improved defensive defenseman for years and KA’s response has been to add Stillman, Lyubushkin, Johnson, Clifton and Byrum; a pantheon of defensive stalwarts.  Does this guy have a clue?

[dudacek]

Always appreciate when the numbers support the eye test.

It felt like the Sabres improved in their own zone this year; if I could say it in one sentence, it would be “instead of giving guys a clear lane and a full second to shoot like last year, they were giving him partially obstructed lanes and a half-second.”

Also interested in the offensive data stating they did not create nearly as much off the rush this year. Am I correct in reading that as they were sending one guy to blow the zone instead of 2? Certainly that would affect the numbers of Tuch and Cozens, who I’ve always found to be extremely dangerous off the rush.

This also feeds into the acquisition of Byram, and type of defencemen they gravitate to: one way to compensate for fewer breakout targets is to employ accurate breakout passers.

Another thing I’m curious about is how the numbers at both ends of the ice compared from the first half to the second. I felt by eye test there were noticeable improvements as the season progressed defensively. I did not necessarily feel the same way offensively.

One would expect a break-in period at both ends of the ice, if we were dealing with a new system.

Edited April 15 by dudacek

[Thorny]

27 minutes ago, dudacek said:

Always appreciate when the numbers support the eye test.

It felt like the Sabres improved in their own zone this year; if I could say it in one sentence, it would be “instead of giving guys a clear lane and a full second to shoot like last year, they were giving him partially obstructed lanes and a half-second.”

Also interested in the offensive data stating they did not create nearly as much off the rush this year. Am I correct in reading that as they were sending one guy to blow the zone instead of 2? Certainly that would affect the numbers of Tuch and Cozens, who I’ve always found to be extremely dangerous off the rush.

This also feeds into the acquisition of Byram, and type of defencemen they gravitate to: one way to compensate for fewer breakout targets is to employ accurate breakout passers.

Another thing I’m curious about is how the numbers at both ends of the ice compared from the first half to the second. I felt by eye test there were noticeable improvements as the season progressed defensively. I did not necessarily feel the same way offensively.

One would expect a break-in period at both ends of the ice, if we were dealing with a new system.

Re: your last sentence - 

yes, the plan devotes one full year to the players learning each skill. 26-27 is “how to body check safely”

[GASabresIUFAN]

4 hours ago, Thorny said:

Re: your last sentence - 

yes, the plan devotes one full year to the players learning each skill. 26-27 is “how to body check safely”

What year is learning to back check and learning to drive the net on offense?

[Scottysabres]

There is only 1 sure fire way to credibly substantiate LG's stats, dudacek's eyes, Thorny's gut feeling and GA's skepticism circa 1950's.

Poll the players themselves.

Ranch or Blue Cheese.....