Addition by Subtraction

[JoeSchmoe]

I know it's been touched on many times as part of other threads, but I wanted a full discussion on the addition by subtraction for next year's roster. We dropped quite a four boat anchors and by my math we picked up one.

Clifton

Lafferty

Cozens

Gilbert

I feel like Danforth is the only new guy we got that will be drag on the team (I've got no idea what they were thinking with him). 

How much does losing these guys improve our goal differential? 

[PerreaultForever]

Can't agree with the depiction of these players as anchors. 

Lafferty and Gilbert were exactly what they were. Bottom players and extras. Expecting more was stupid.

Clifton lost his game playing in an unstructured system. I said this might happen when we signed him. He always had the tendency to run around and try to over step and then get caught out of position. In this chaotic system he ended up floundering. We broke him. 

Cozens? An anchor? Let's see how Ottawa goes. Quinn's still an anchor until he isn't. Same for Samuelsson. 

[JoeSchmoe]

7 hours ago, PerreaultForever said:

Can't agree with the depiction of these players as anchors. 

Lafferty and Gilbert were exactly what they were. Bottom players and extras. Expecting more was stupid.

Clifton lost his game playing in an unstructured system. I said this might happen when we signed him. He always had the tendency to run around and try to over step and then get caught out of position. In this chaotic system he ended up floundering. We broke him. 

Cozens? An anchor? Let's see how Ottawa goes. Quinn's still an anchor until he isn't. Same for Samuelsson. 

Even in their limited roles against limited competition, Lafferty and Gilbert were two of our worst statistical players. Clifton was less sheltered but was still our worst defensemen next to Bryson.

The guys we got to replace them are pretty decent in similar roles.

[LGR4GM]

10 minutes ago, JoeSchmoe said:

Even in their limited roles against limited competition, Lafferty and Gilbert were two of our worst statistical players. Clifton was less sheltered but was still our worst defensemen next to Bryson.

The guys we got to replace them are pretty decent in similar roles.

The crazy part is they brought back Bryson. 

I think the Sabres looked at some stats and decided to upgrade, I think without better gt, it won't be enough. 

[JP51]

8 hours ago, PerreaultForever said:

Can't agree with the depiction of these players as anchors. 

Lafferty and Gilbert were exactly what they were. Bottom players and extras. Expecting more was stupid.

Clifton lost his game playing in an unstructured system. I said this might happen when we signed him. He always had the tendency to run around and try to over step and then get caught out of position. In this chaotic system he ended up floundering. We broke him. 

Cozens? An anchor? Let's see how Ottawa goes. Quinn's still an anchor until he isn't. Same for Samuelsson. 

Yeah, I think Muel based on his minutes is a major anchor... Gilbert was just an 8th d who was a C minus fighter that was willing... just an Adams failed nod to you need someone to respond.... here lets throw this cannon fodder out there to get beat... cause I know he will jump in when no one else will... 

Cozens I agree.. as unpopular as this may be I think we see Cozens mature over the years and will become a player that people will want... right now he is just too much of a spaz and head case to be an effective player and leader night in and night out...  but in the right situation I think he develops into a guy that will score 25-30 for 50-60 points... and can play two way minutes...  and no, I didnt say now... (especially the 2 way minute part) but if he settles down and develops I believe he can be a taller version of a poor mans Mike Peca or a higher scoring Gaustad 

 

[LGR4GM]

Specifically, the Sabres moved away from a bunch of guys with bad xgf%. Whether that's deliberate or just more dart throwing, idk. The other issue is better coached teams have better xgf% overall. So we'll see what happens. 

Xgf% is impacted by how teams play. Buffalo plays or played a very rush oriented attack with limited ability to cycle, so they don't get a bunch of chances. On the inverse they were mediocre shot getting pucks out so that all comes together. 

Edited 23 hours ago by LGR4GM

[Pimlach]

9 hours ago, PerreaultForever said:

Can't agree with the depiction of these players as anchors. 

Lafferty and Gilbert were exactly what they were. Bottom players and extras. Expecting more was stupid.

Clifton lost his game playing in an unstructured system. I said this might happen when we signed him. He always had the tendency to run around and try to over step and then get caught out of position. In this chaotic system he ended up floundering. We broke him. 

Cozens? An anchor? Let's see how Ottawa goes. Quinn's still an anchor until he isn't. Same for Samuelsson. 

You told us what to expect on Clifton.   In his interview with Jon Scott he talked about the first season in Buffalo, playing on his wrong side, and how he had trouble the adjusting to Granato's system.  He did recover that first season and ended with a +12 after a horrendous start.  He was accountable for his performance though and made no excuses.  

Everything about Clifton's game declined under Lindy which surprised me.  He even had less PIMs and less fights, and he saw very limited action at the end of the season.   It was time for both parties to move on.

I have some big concerns about how the players reacted to Lindy.  The team play declined overall, and the Athletic had that players survey (limited to 111 players) that was not complimentary to Lindy at all.  

Big year coming for both Adams and Lindy.   

Cozens - you have to admit he really struggled again.   I don't understand how he got the "A" and lets face it, he was handed the 2C job years ago.  Cozens  (and others such as Quinn and Peterka) were allowed to play a one-way game in their first few seasons.  

Edited 22 hours ago by Pimlach

[steveoath]

Is complete fantasy, but would have been good to see TP go after Pete DeBoer. 

[DarthEbriate]

54 minutes ago, Pimlach said:

Big year coming for both Adams and Lindy.   

But it's not, really, is it?

It's Lindy's final season as a coach. He could step into the front office, I suppose, as an advisor? But he's not coaching anywhere else in the NHL and Appert is already slotted to take over.

Adams could get fired if they're a trainwreck, but... he listens to the owner. Jarmo is there, but is he really a threat if he doesn't go full ingratiation mode? Would he want to do that? The Bills are still good and building a stadium and therefore command the attention. It's status quo and trades are hard and why goaltend until Levi is ready for Adams.

[Pimlach]

11 minutes ago, DarthEbriate said:

But it's not, really, is it?

It's Lindy's final season as a coach. He could step into the front office, I suppose, as an advisor? But he's not coaching anywhere else in the NHL and Appert is already slotted to take over.

Adams could get fired if they're a trainwreck, but... he listens to the owner. Jarmo is there, but is he really a threat if he doesn't go full ingratiation mode? Would he want to do that? The Bills are still good and building a stadium and therefore command the attention. It's status quo and trades are hard and why goaltend until Levi is ready for Adams.

You are probably correct.  I keep thinking like the 31 other teams.  

Edited 21 hours ago by Pimlach

[dudacek]

1 hour ago, LGR4GM said:

Specifically, the Sabres moved away from a bunch of guys with bad xgf%. Whether that's deliberate or just more dart throwing, idk. The other issue is better coached teams have better xgf% overall. So we'll see what happens. 

Xgf% is impacted by how teams play. Buffalo plays or played a very rush oriented attack with limited ability to cycle, so they don't get a bunch of chances. On the inverse they were mediocre shot getting pucks out so that all comes together. 

Chicken or egg, half the D were a terrible match for the system last year.

People talk a lot about how Byram’s fancy stats got inflated by Dahlin.

They don’t talk about how they got deflated by the other guys.

Clifton, Bryson and Samuelsson were truly awful in Lindy’s system. As in 208, 202 and 223 out of 240 D who played 300 minutes last year.

 

[dudacek]

Dahlin was 50th, 1 slot behind Kesselring, who led Utah.

But Utah overall was very good at that particular stat, with 5 D in the top 79.

It remains to be seen if he (and Timmins, who also is good at this stat) can elevate Byram and Power (165 and 168) at all, or if they get pulled toward the bottom in Buffalo.

[dudacek]

xGF% is an interesting stat.

It gets thrown out in some quarters as the defining characteristic of how good a player is, but it’s really just a measurement of high danger chances allowed versus created that is heavily influenced by team factors such as system, deployment and teammates.

It has all the same inherent weaknesses as plus/minus when it comes to player comparisons.

Or do we really believe Jordan Spence, Shayne Gostisbehere and Nate Schmidt are elite players (ranked 2, 3 and 4), while Brock Faber (169), Brandon Montour (177) and Seth Jones (209) are terrible?

[LGR4GM]

17 minutes ago, dudacek said:

xGF% is an interesting stat.

It gets thrown out in some quarters as the defining characteristic of how good a player is, but it’s really just a measurement of high danger chances allowed versus created that is heavily influenced by team factors such as system, deployment and teammates.

It has all the same inherent weaknesses as plus/minus when it comes to player comparisons.

Or do we really believe Jordan Spence, Shayne Gostisbehere and Nate Schmidt are elite players (ranked 2, 3 and 4), while Brock Faber (169), Brandon Montour (177) and Seth Jones (209) are terrible?

It's why you don't use 1 fancy stat to explain a player. It however has better predictive measures than +/- so what you say there is not correct. 

[dudacek]

3 minutes ago, LGR4GM said:

It's why you don't use 1 fancy stat to explain a player. It however has better predictive measures than +/- so what you say there is not correct. 

Could you clarify?

I understand the Sabres 10-game win streak scenario where fancy stats correctly predicted the team was getting bounces and wasn’t going to sustain.

But does it have any predictive qualities when it comes to individual players?

[Hawerchuk]

2 hours ago, LGR4GM said:

The crazy part is they brought back Bryson. 

 

Agreed! Honestly, Bryson is probably one of the best skaters in the NHL, I'm not kidding but as for anything else, he's USELESS!

[dudacek]

15 minutes ago, LGR4GM said:

It's why you don't use 1 fancy stat to explain a player. It however has better predictive measures than +/- so what you say there is not correct. 

I’m speaking in terms of player to player comparisons on different teams.

Generally speaking, a sheltered offensively deployed defenceman on a good team (Gostisbehere?) is going to have good xG% and a defensively deployed defenceman getting tough minutes on a bad team (Vlasic?) is going to have bad xG%.

I’ve read many posts saying Seth Jones and Bo Byram are supposed to be bad because xG, when each was demonstrably good enough to get significant minutes for Stanley Cup winning teams.

Edited 20 hours ago by dudacek

[LGR4GM]

5 minutes ago, dudacek said:

Could you clarify?

I understand the Sabres 10-game win streak scenario where fancy stats correctly predicted the team was getting bounces and wasn’t going to sustain.

But does it have any predictive qualities when it comes to individual players?

Yea xgf is a better predictor than +/- 

Xgf predicts goals based on shot type location and movement. It better shows what a player contributed to. 

Just now, dudacek said:

I’m speaking in terms of player to player comparisons on different teams.

Generally speaking, a sheltered offensively deployed defenceman on a good team (Gostisbehere?) is going to have good xG% and a defensively deployed defenceman getting tough minutes on a bad team (Vlasic?) is going to have bad xG%.

Yea it's why no one serious about this stuff uses 1 stats alone. 

[LGR4GM]

18 minutes ago, Hawerchuk said:

Agreed! Honestly, Bryson is probably one of the best skaters in the NHL, I'm not kidding but as for anything else, he's USELESS!

He's actually a good puck carrier but he's too small/weak down low

[mjd1001]

Addition by subtraction.  To me its not simply taking someone away, but what you replace them with.

With that said I have always thought that getting rid of Cozens fits the term 'addition by subtraction'...simply because without him at Center, The added minutes that McLeod, Krebs and Kulich got after he left served the team better than the minutes Cozens had.  They 'helped' the team just about as much offensively, and they hurt the team a lot less.

That is not to say Cozens doesn't have talent, but His 16-17 minutes per game going to 0 for the Sabres, and those other guys getting the extra minutes were a bonus.

Basically, would I rather have Cozens getting 18 min per game, McLeod getting 12-14, Krebs getting 10, and Kulich not having a big role at all? 

-OR-

Cozens getting zero (subtraction from the team), Mcleod getting his 16, Krebs getting 12, and Kulich getting 12-14?  <--- I'll take this one.

The productivity/score is just as good (last year maybe better with McLeods and Kulich's game toward the end of the season) and those guys, even Kulich as a rookie, make/made a LOT less costly mistakes than Cozens did in his minutes.

 

I would expect that to accelerate this year with Norris getting any productivity.

As for the rest of the guys, they didn't play enough of a role on the team for me to care about.

Edited 18 hours ago by mjd1001

[mjd1001]

9 hours ago, dudacek said:

xGF% is an interesting stat.

It gets thrown out in some quarters as the defining characteristic of how good a player is, but it’s really just a measurement of high danger chances allowed versus created that is heavily influenced by team factors such as system, deployment and teammates.

It has all the same inherent weaknesses as plus/minus when it comes to player comparisons.

Or do we really believe Jordan Spence, Shayne Gostisbehere and Nate Schmidt are elite players (ranked 2, 3 and 4), while Brock Faber (169), Brandon Montour (177) and Seth Jones (209) are terrible?

Agreed.

I actually prefer ACTUAL goals for/vs against, as it takes into account shooting percentage (from what I can tell, xGF% does not take individual shooting percentage, so it doesn't really take into account that a player like Tage is a 50% more accurate shooter than a guy like Cozens).

Plus-minus is not a stat that can tell you everything about a player, I admit that, but it shouldn't be totally thrown away either.  I think the key with the 'advanced stats' is to use them in combination with each other.  A guy has a good xGF%?  Well, what is the competition he faces, or is he on a line/paring with a guy who is a super accurate shooter or a terrible one?  For me, you have to look at the advanced stats, all of them...if you see anything that stands out (good or bad), think to yourself...why?  Is there something that accounts for this?  Is this something that is a one time/one year thing or a long term thing.

By using all the advanced stats in combination with each other, and asking 'why' when presenting them....you can get a somewhat more accurate guage of how good a player is rather than just using your single favorite one.

Many people on here may know my favorite 'whipping boy' for Sabres problems over the years has been Cozens.  And that isn't because of one or two stats. Its because many/most of his advanced stats are below average (not just one of them), AND they have been for years (even his really good year), and more often than now other players advanced stats are worse when they are playing with him and get better when they are with anyone else, AND simply watching him, the eye test backs all that up.

Personally I usually use the eye test first, form an opinion of a player, and then see if the advanced stats/analytics back up that initial opinion. The only time I really work backwards (analytics first) is when the Sabres trade for someone/acquire someone that I haven't seen play all that much.

Edited 11 hours ago by mjd1001

[LGR4GM]

10 minutes ago, mjd1001 said:

Agreed.

I actually prefer ACTUAL goals for/vs against, as it takes into account shooting percentage (from what I can tell, xGF% does not, so it doesn't really take into account that a player like Tage is a 50% more accurate shooter than a guy like Cozens).

Plus-minus is not a stat that can tell you everything about a player, I admit that, but it shouldn't be totally thrown away either.  I think the key with the 'advanced stats' is to use them in combination with each other.  A guy has a good xGF%?  Well, what is the competition he faces, or is he on a line/paring with a guy who is a super accurate shooter or a terrible one?  For me, you have to look at the advanced stats, all of them...if you see anything that stands out (good or bad), think to yourself...why?  Is there something that accounts for this?  Is this something that is a one time/one year thing or a long term thing.

By using all the advanced stats in combination with each other, and asking 'why' when presenting them....you can get a somewhat more accurate guage of how good a player is rather than just using your single favorite one.

Many people on here may know my favorite 'whipping boy' for Sabres problems over the years has been Cozens.  And that isn't because of one or two stats. Its because many/most of his advanced stats are below average (not just one of them), AND they have been for years (even his really good year), and more often than now other players advanced stats are worse when they are playing with him and get better when they are with anyone else, AND simply watching him, the eye test backs all that up.

Personally I usually use the eye test first, form an opinion of a player, and then see if the advanced stats/analytics back up that initial opinion. The only time I really work backwards (analytics first) is when the Sabres trade for someone/acquire someone that I haven't seen play all that much.

That's not accurate. Xgf takes into account league average sh%. The problem with actual goals for and against is all the extra noise and all the randomness. There's simply a better sample of shots.

We could get into GAR and WAR but that's really complicated and the models vary a ton.

[mjd1001]

13 minutes ago, LGR4GM said:

That's not accurate. Xgf takes into account league average sh%. The problem with actual goals for and against is all the extra noise and all the randomness. There's simply a better sample of shots.

We could get into GAR and WAR but that's really complicated and the models vary a ton.

Actually, that is my point...why do I want to evaluate an individual player based on 'league average shooting percentage', when that player may be quite a bit higher or lower?

Yeah, there may be 'noise' in actual goals, but to me at least it takes into account the difference BETWEEN the league average shooting percentage and that actual player I am evaluating.  That 'noise' will, statistically, likely 'even out' when you look at the actual goal numbers over a long period of time. I get looking at a partial season, or a half season, may not be helpful, but if a guy is below 50 year after year, vs a guy above 50 year after year, the trend is your friend.

Again, If Cozens is, for his career, a 10% or below shooter, and a guy like Tage is 15% or higher....XGF might be the same for both of them but in reality it vastly over-rates Cozens and under-rates Thompson.

Edited 17 hours ago by mjd1001

[LGR4GM]

5 minutes ago, mjd1001 said:

Actually, that is my point...why do I want to evaluate an individual player based on 'league average shooting percentage', when that player may be quite a bit higher or lower?

Yeah, there may be 'noise' in actual goals, but to me at least it takes into account the difference BETWEEN the league average shooting percentage and that actual player I am evaluating.  That 'noise' will, statistically, likely 'even out' when you look at the actual goal numbers over a long period of time. I get looking at a partial season, or a half season, may not be helpful, but if a guy is below 50 year after year, vs a guy above 50 year after year, the trend is your friend.

Again, If Cozens is, for his career, a 10% or below shooter, and a guy like Tage is 15% or higher....XGF might be the same for both of them but in reality it vastly over-rates Cozens and under-rates Thompson.

No xgf wouldn't be the same. You wanna know why? Good shooters get to the good shooting spots. And xgf% is also looking at the defensive side of things. 

You're suggesting actual goals, which are rarer than shots, measure individual players contributing to winning more but stats says they don't. Sure that noise might even out over a career for actual goals but the funny part to me is so would the xgf. The good shooters get to the good spots to shoot, that's the key. 

[LGR4GM]

Let's say Cozens is a 10% shooter, he's that because he doesn't get to the spots he needs to and takes the shot types he needs to. Now Tage is a 15% guy. So they get to point X on the ice and take a wrist shot. You're arguing that Tage will score more there because the league average is 11% and Tage shoots higher, yes. But you're looking at 4 goals on 100 shots. We also know the areas players score more from. Xgf% would not be the same because of Tage scores more from a spot than so does others, so the average goes up. If Tage prevents shots from those spots it prevents xga. Cozens would have to shoot from the same places meaning he's skilled enough to get there and defend the same shots. It's just not happening. 

Versus actual goals for. If I put Tage on a team with a real goalie his actual goals against goes down even if his xga stays the same. He's still giving up what xga says. It's also why on ice sv% is important. 

Idk, I think we just fundamentally disagree about what xgf shows and why it's better statistically than actual goals for and against. Doesn't mean they both can't tell us things.