The Sabres will win the division if...

[bob_sauve28]

I don't agree at all.  I saw a kid in his first NHL season play a tough game in the d-zone while flashing quite a bit of offensive ability.   

This is what I also saw. He was tough, solid and did not commit a huge number of mistakes. I was really impressed. Won his battles, moved the puck smartly and was well positioned. I was very thankful he matured into a regular NHL defenseman. Consider him one of the draft picks we got really lucky on  

[Drunkard]

he still made the moves

So does the GM for Edmonton get credit for McDavid? He made the selection. The fact that they've been in a 10 year long playoff drought and they completely lucked out with the lottery doesn't dsicount it? To me it does.

 

I give Murray credit for O'Reilly, Kane, Bogosian, Lehner, Gionta, and Gorges, plus the blame for Moulson new current contract, but I give him substantially less credit for Eichel and Reinhart. Eichel was the obvious choice, but he does get some credit for retaining Nolan and setting up the roster to finish last. I give him more credit for picking Reinhart over Bennett or somebody else but it's discounted by the fact that Darcy had already set that team up to finish last and he just allowed them to coast to the finish line.

[Norcal]

So does the GM for Edmonton get credit for McDavid? He made the selection. The fact that they've been in a 10 year long playoff drought and they completely lucked out with the lottery doesn't dsicount it? To me it does.

I give Murray credit for O'Reilly, Kane, Bogosian, Lehner, Gionta, and Gorges, plus the blame for Moulson new current contract, but I give him substantially less credit for Eichel and Reinhart. Eichel was the obvious choice, but he does get some credit for retaining Nolan and setting up the roster to finish last. I give him more credit for picking Reinhart over Bennett or somebody else but it's discounted by the fact that Darcy had already set that team up to finish last and he just allowed them to coast to the finish line.

Murray's done well and he'll continue to do well. He's not mistake proof but he's got some balls and he knows how to swing em

The GM from Edmonton is new so McDavid was his 1st move. To be honest idgaf what he does or doesn't do because...Sabres

[Drunkard]

Murray's done well and he'll continue to do well. He's not mistake proof but he's got some balls and he knows how to swing em

The GM from Edmonton is new so McDavid was his 1st move. To be honest idgaf what he does or doesn't do because...Sabres

He still made the move. That was your initial response to my point. To give him credit for choosing Eichel when he was the consensus choice just makes no sense to me. I give him way more credit for trading for an injured Kane to help ensure we finished last and very little for the actual selection is my point. Any GM in his position would have pixked Eichel and any GM that didn't probably would've been fired on the spot.

[dudacek]

Some interesting stuff in there Wildcard, but I don't feel that my questions were answered. In some cases they were reinforced, but I guess that was maybe your point.

Here's another one: doesn't Corsi have exactly the same flaw as +\-?

It's a proven way of measuring the effectiveness of a unit over a period of time, but it also gives individual players credit or blame for things that may have had nothing to do with.

[LGR4GM]

Some interesting stuff in there Wildcard, but I don't feel that my questions were answered. In some cases they were reinforced, but I guess that was maybe your point.

Here's another one: doesn't Corsi have exactly the same flaw as +\-?

It's a proven way of measuring the effectiveness of a unit over a period of time, but it also gives individual players credit or blame for things that may have had nothing to do with.

At this point I think you are asking for Stats to be perfect and they never will be. I could say that player X isn't really a 30g a year guy because 20 of his goals were tap ins and 5 were empty netters so Goals isn't a good stat for measuring how good a player is at scoring.  All stats are taken with a grain of salt and some stats are better than others. Stats need context though. 

[dudacek]

At this point I think you are asking for Stats to be perfect and they never will be. I could say that player X isn't really a 30g a year guy because 20 of his goals were tap ins and 5 were empty netters so Goals isn't a good stat for measuring how good a player is at scoring.  All stats are taken with a grain of salt and some stats are better than others. Stats need context though. 

 

I think analytics is attempting to make stats better, but in the sports message board world they often woefully lack context.

Blue pulled out a lot of numbers that essentially said the Sabres, over the course of the season, were outplayed 5on5 no matter what defencemen they had on the ice, which generally supports his contention that we need better defencemen.

 

But they didn't tell me which defencemen were to blame, or really even establish it was the fault of the defencemen.

What about the forwards, or the system? How much of a factor were they?

[WildCard]

Some interesting stuff in there Wildcard, but I don't feel that my questions were answered. In some cases they were reinforced, but I guess that was maybe your point.

Here's another one: doesn't Corsi have exactly the same flaw as +\-?

It's a proven way of measuring the effectiveness of a unit over a period of time, but it also gives individual players credit or blame for things that may have had nothing to do with.

TBF dudacek, I answered your question about Corsi in my last paragraph.  ;)

I think analytics is attempting to make stats better, but in the sports message board world they often woefully lack context.

Blue pulled out a lot of numbers that essentially said the Sabres, over the course of the season, were outplayed 5on5 no matter what defencemen they had on the ice, which generally supports his contention that we need better defencemen.

 

But they didn't tell me which defencemen were to blame, or really even establish it was the fault of the defencemen.

What about the forwards, or the system? How much of a factor were they?

http://stats.hockeyanalysis.com/showplayerwowycharts.php?pid=1804&season=2015-16&sit=5v5

 

By taking away a player, you can tell their impact. If, when taking away a player, everyone else is negatively affected, then it's pretty obvious of their impact. The most notable of these cases would be Bergeron's. Take him away, and his whole team sucks. With him on the ice however, everyone benefits substantially.

 

http://stats.hockeyanalysis.com/showplayer.php?pid=572&withagainst=true&season=2015-16&sit=5v5

 

 

Oddly enough, for us, the best possession players are Kane (might explain why Murray wanted him so badly), ROR, and Reinhart. 

Edited April 26, 2016 by WildCard

[MattPie]

TBF dudacek, I answered your question about Corsi in my last paragraph.  ;)

http://stats.hockeyanalysis.com/showplayerwowycharts.php?pid=1804&season=2015-16&sit=5v5

 

By taking away a player, you can tell their impact. If, when taking away a player, everyone else is negatively affected, then it's pretty obvious of their impact. The most notable of these cases would be Bergeron's. Take him away, and his whole team sucks. With him on the ice however, everyone benefits substantially.

 

http://stats.hockeyanalysis.com/showplayer.php?pid=572&withagainst=true&season=2015-16&sit=5v5

 

 

Oddly enough, for us, the best possession players are Kane (might explain why Murray wanted him so badly), ROR, and Reinhart. 

 

I wonder if that's an artifact of Kane loving to skate down the wing and shoot, as (currently) possession is based largely on shots right?

[WildCard]

I wonder if that's an artifact of Kane loving to skate down the wing and shoot, as (currently) possession is based largely on shots right?

That's the first thing that came to my mind too. His CA60 is actually impressive though 

[dudacek]

TBF dudacek, I answered your question about Corsi in my last paragraph.  ;)

 

http://stats.hockeyanalysis.com/showplayerwowycharts.php?pid=1804&season=2015-16&sit=5v5

 

By taking away a player, you can tell their impact. If, when taking away a player, everyone else is negatively affected, then it's pretty obvious of their impact. The most notable of these cases would be Bergeron's. Take him away, and his whole team sucks. With him on the ice however, everyone benefits substantially.

 

Sorry if I'm being dense here, but this isn't Corsi, is it? Isn't it Corsi relative to teammates, or something like that?

And can't you do exactly the same exercise with +\-?

 

And you bring up one of my earlier questions again: if those charts are the best stats we've got, why do we say Gorges drags down Risto when the opposite appears to be the case?

 

And how can you say they are wrong when it comes to Jack?

[dudacek]

Here's a common game situation:

Loose puck, Penguins zone in the corner.

Foligno knocks Letang off the puck, spins toward the boards and chips a pass to Bogo at the point.

Bogo whiffs on it and Kessel is away to the races. He crosses the blue line and unleashes a big slapper.

McCabe gets back in time, gets his stick on the puck and it deflects up high into the netting.

 

As a coach I am happy with McCabe and Foligno, pissed at Bogo.

 

Yet if I understand Corsi right, all three get dinged with a negative play, as do Larsson and Gionta, who have done nothing at all.

How is this any better than +\- in measuring how well each individual player played in that sequence?

Edited April 26, 2016 by dudacek

[WildCard]

Sorry if I'm being dense here, but this isn't Corsi, is it? Isn't it Corsi relative to teammates, or something like that?

And can't you do exactly the same exercise with +\-?

 

And you bring up one of my earlier questions again: if those charts are the best stats we've got, why do we say Gorges drags down Risto when the opposite appears to be the case?

 

And how can you say they are wrong when it comes to Jack?

It's all of the stats with/without a player, Corsi being one of them. You could do the same with +/-. I don't think I can really explain why +/- is bad other than the articles I've linked

 

Honestly, I don't know. All of the advanced stats say Gorges pulls his weight, I'm hoping True can jump in on this one

 

Jack's is bad at possession, that's pretty obvious. He's not supposed to be good at it though. His other stats with/without, without looking at them right now, are good I'd wager; such as GF%

[WildCard]

Here's a common game situation:

Loose puck, Penguins zone in the corner.

Foligno knocks Letang off the puck, spins toward the boards and chips a pass to Bogo at the point.

Bogo whiffs on it and Kessel is away to the races. He crosses the blue line and unleashes a big slapper.

McCabe gets back in time, gets his stick on the puck and it deflects up high into the netting.

 

As a coach I am happy with McCabe and Foligno, pissed at Bogo.

 

Yet if I understand Corsi right, all three get dinged with a negative play, as do Larsson and Gionta, who have done nothing at all.

How is this any better than +\- in measuring how well each individual player played in that sequence?

You're exactly right, that is a common game situation. So it's recorded in Corsi, and reflected in Corsi WOWY stats. The key thing about Corsi vs +/-, is the frequency in which they occur. Take your situation with Bogo. In the course of a game and a season, a player will encounter such plays more often than they will goals. Indeed such plays as missed passes, failed pinches, poor outlet passes, etc. occur tens of times a game. Eventually, over a season, that will even out, and a player's actual worth will be reflected in it. For +/-, that play occurs maybe 2-3 times a game, indeed maybe less as +/- isn't recorded on special teams. So, it's a much, much smaller sample size, as well as not indicative of how the player has played that entire game, but, in most cases, how lucky/unlucky he was to be out there. 

[nfreeman]

Here's a common game situation:

Loose puck, Penguins zone in the corner.

Foligno knocks Letang off the puck, spins toward the boards and chips a pass to Bogo at the point.

Bogo whiffs on it and Kessel is away to the races. He crosses the blue line and unleashes a big slapper.

McCabe gets back in time, gets his stick on the puck and it deflects up high into the netting.

 

As a coach I am happy with McCabe and Foligno, pissed at Bogo.

 

Yet if I understand Corsi right, all three get dinged with a negative play, as do Larsson and Gionta, who have done nothing at all.

How is this any better than +\- in measuring how well each individual player played in that sequence?

 

You're exactly right, that is a common game situation. So it's recorded in Corsi, and reflected in Corsi WOWY stats. The key thing about Corsi vs +/-, is the frequency in which they occur. Take your situation with Bogo. In the course of a game and a season, a player will encounter such plays more often than they will goals. Indeed such plays as missed passes, failed pinches, poor outlet passes, etc. occur tens of times a game. Eventually, over a season, that will even out, and a player's actual worth will be reflected in it. For +/-, that play occurs maybe 2-3 times a game, indeed maybe less as +/- isn't recorded on special teams. So, it's a much, much smaller sample size, as well as not indicative of how the player has played that entire game, but, in most cases, how lucky/unlucky he was to be out there. 

 

I think WC has it right here -- i.e. Corsi (and its derivatives) are more accurate metrics than plus/minus because the sample size is much larger.

[woods-racer]

Thanks for the persistence dudacek.

 

I've been lost for a while and glad you asked all the questions.

 

Always seemed to me they where skinning the same dead cat till desired results where achieved. 

 

What all this proves to me the most is, a trained eye from a coach, GM or scout is still far more valuable then the accumulation of  a player stats.

[dudacek]

You're exactly right, that is a common game situation. So it's recorded in Corsi, and reflected in Corsi WOWY stats. The key thing about Corsi vs +/-, is the frequency in which they occur. Take your situation with Bogo. In the course of a game and a season, a player will encounter such plays more often than they will goals. Indeed such plays as missed passes, failed pinches, poor outlet passes, etc. occur tens of times a game. Eventually, over a season, that will even out, and a player's actual worth will be reflected in it. For +/-, that play occurs maybe 2-3 times a game, indeed maybe less as +/- isn't recorded on special teams. So, it's a much, much smaller sample size, as well as not indicative of how the player has played that entire game, but, in most cases, how lucky/unlucky he was to be out there. 

 

Bigger sample size makes it better. Makes sense.

Am I wrong to think that it is still flawed?

 

I think it is a great way to measure team and even unit success.

Not sold on the individual player level.

It measures shots at the net by every player on the ice.

Not saying it's useless, just that there seem to be too many variables to be a dependable measure of a player.

Sorta like plus/minus

Edited April 27, 2016 by dudacek

[WildCard]

Bigger sample size makes it better. Makes sense.

Am I wrong to think that it is still flawed?

 

I think it is a great way to measure team and even unit success.

Not sold on the individual player level.

It measures shots at the net by every player on the ice.

Not saying it's useless, just that there seem to be too many variables to be a dependable measure of a player.

Sorta like plus/minus

It's still flawed, no stat is perfect. That's the reason behind analytics and not just statistics. But, if you take enough of these stats, and combine them on a chart like WAR does, then you get a pretty accurate picture on who a player is. 

Thanks for the persistence dudacek.

 

I've been lost for a while and glad you asked all the questions.

 

Always seemed to me they where skinning the same dead cat till desired results where achieved. 

 

What all this proves to me the most is, a trained eye from a coach, GM or scout is still far more valuable then the accumulation of  a player stats.

I'm not so sure about that last part. Granted I don't have a trained eye, but many of us on here were convinced Gorges is terrible, and his stats don't relay that IMO. Taylor Hall is famously underrated, despite his stats saying otherwise. There are plenty of examples where personal or league wide bias casts an inaccurate light upon a player despite all evidence to the contrary

[TrueBlueGED]

Sorry, I drifted from the main thread of the discussion and didn't express myself very well.

I wasn't arguing your point at all. There are three separate things that I am struggling with.

 

1) based strictly on @fancystats, why do we say Risto needs a better partner? Why aren't we saying Gorges needs a better partner? Or that Risto is not very good?

2) I know the flaws of +\-. I also know the object of the game is to score more than your opponents and that McCabe was on the ice for 14 more even-strength goals by us than the bad guys down the stretch. Why is that irrelevant?

3) Corsi measures the effectiveness of the unit on the ice, why should we conclude the D is responsible for our poor Corsi numbers? Why isn't it Larsson's fault for not shooting enough, or Ennis's for constantly turning the puck over?

 

I'm not "into" fancy stats, but I love me some analysis. Too many internet fanboys will haul out some numbers and say here's proof that so and so sucks without explaining, or even understanding the underlying factors.

 

You aren't one of those, so help me understand.

 

So, I think WildCard ended up doing the vast majority of the heavy lifting here, and for that I'm thankful. I honestly think I might start having papers due in the middle of the semester. But I digress...

 

1) Based strictly on fancystats, it's hard. The tool I would normally point to is WOWY--with or without you statistics (and the superWOWY tool at puckalytics is great if you want to incorporate forwards with the defensemen). Thing is, Risto and Gorges didn't spend too much time apart from one another. They were together for 1196 minutes, Risto was without Gorges for 295 minutes, and Gorges was without Risto for only 140 minutes. Risto's second most common partner was Bogosian at 122 minutes, while Gorges had Pysyk at 53 minutes. Those tiny samples with other defenders are compounded by the fact that the Risto-Gorges pairing when together faced much different competition and usage than the other defensemen. So we really can't glean much from their impact on other skaters relative to their impact on one another. With that said, take the following with a very large grain of salt:

 

              R & G   R w/o G   G w/o R

GF/60         1.71    2.03      1.28

GA/60         2.26    3.05      3.05

GF%           43      40        37.5

CF%           45.5    45.4      47.8

on-ice SV%    92.8    89.8      93.33

on-ice SH%    6.36    7.3       4.55

PDO           99.2    97.1      97.9

 

What it means is open to interpretation. On the one hand, the team is scoring a higher percentage of goals with Risto on the ice without Gorges, but is worse at possession. Also, the team save percentage is dramatically worse without Gorges, though that's offset by a higher shooting percentage. But with PDOs in the 97 range, there's too much noise there to be confident in anything statistically--the low SH% for Gorges and the the low SV% for Risto could just be bad luck. I don't think it's crazy to think the team is going to be more efficient offensively with Risto as opposed to Gorges and better defensively in-zone with Gorges over Risto, but there's just way too much variation in small sample percentages to be even remotely confident from a statistical standpoint. 

 

One thing you could do is look at individual points percentage--the percentage of goals scored while on the ice that the player had a goal or assist on. Risto was at 40.9 while Gorges sat at 32.4. Again, though, I don't think suggesting Risto is better offensively than Gorges is controversial. Ultimately, I think the totality of the stats point to them being reasonable compliments for one another, although collectively overmatched. I will say that going through all of this has made me wonder if perhaps a Brodin-type is a better complement to Risto than a Fowler-type. Something to think about for sure. Kind of makes me wish we had a good offensive LHD option to have tested during the season. C'est la vie. 

 

Anyway, I don't think there's a strong pure statistical case for Gorges being a bigger problem than Risto. Eye test baby. I love the eye test! :angel:

 

2) As others have said, because reasons. +/- is by far the noisiest hockey statistic there is. Not only are nearly half of them recorded erroneously, not only does it rely upon teammates and have a teeny tiny sample size, but it's heavily dependent upon shooting and save percentages, both of which are insanely variable over small stretches. On the sample size, there was an average of 5.42 goals per game this year...while there was an average of 59 shots per game (yes shots, not shot attempts as Corsi would measure...couldn't find the shot attempts per game, but admittedly didn't look all too hard). To the extent that Corsi has similar book keeping problems as +/-, the noise pretty quickly gets drown out (assuming it's random and not systematic, which I believe is the case). If a player has a double digit positive/negative +/- over the course of 6, 7, 8 seasons then yea, I think we can draw some conclusions from it. But one season? I have zero confidence in the measure to provide any valuable meaning. 

 

We could look at individual points percentage, in which Jake McCabe is second lowest (among regular defenders) at 27.5. So sure he's on the ice for some goals, but he's not directly contributing to many of them. Last? Bogosian at 26.2. The real question is how many of the goals against they directly contribute to, which is a question I can't answer statistically. One would hope the Sabres could. If they are contributing to goals against at a higher direct rate than they're contributing to goals for, as I suspect, that's a problem.

 

3) Eye test baby!!!!

 

Really though I'm tired, and don't have a good answer off the top of my head. Isolating effects from the team to the unit to the individual is the hardest thing to do, and that's true whether using stats or eyes or both. For this season's Sabres team I'd posit that both the forwards and defense contributed healthily to the poor possession at even strength. My argument has been, and will remain, I believe in the growth of Jack and Samson and one of our secondary prospect wingers to correct this over the growth of McCabe et al. on the blue line. YMMV.

 

(I do think you can start to get at this with relatively in depth analysis, but again, tired. Also possibly drunkish.)

Thanks for the persistence dudacek.

 

I've been lost for a while and glad you asked all the questions.

 

Always seemed to me they where skinning the same dead cat till desired results where achieved. 

 

What all this proves to me the most is, a trained eye from a coach, GM or scout is still far more valuable then the accumulation of  a player stats.

 

I think you should amend this to say the trained eye of a good coach/GM/scout. I would label the hierarchy as follows:

 

Good GM w/ good analytics team > good GM w/o analytics team > good GM w/ bad analytics team > bad GM w/ good analytics team > spreadsheet > bad GM w/o analytics team >= bad GM w/ bad analytics team. Joe Sakic and the Avs reside somewhere in those last two. You don't want to be Joe Sakic and the Avs.

Edited April 28, 2016 by TrueBlueGED

[qwksndmonster]

That was nice.  I usually am not into the stat talk but I enjoyed it very much.

[TrueBlueGED]

That was nice.  I usually am not into the stat talk but I enjoyed it very much.

 

Your voice on the GoT podcast may be super nerdy, but your posts here are not :p

[TrueBlueGED]

I did much better in the re-record.  I'm still a ###### dork, so I'm glad my personality shined through.  We did our intro and covered the first 19 minutes of the episode.  An hour and a half of talking right there.  It's so fun because this is what we do anyways, now we just try to do it better and say like, uhhhhhh, ANYWAYS, but, uhhhhh less.  Listening to yourself suck at talking: not pleasant.

 

I'm glad my first lectures ~5 years ago were not recorded. Seriously, I'd hate to listen to that . Also, where is your podcast? I need a voice to put to our forthcoming weekly GoT disagreements.

[dudacek]

Good answers.

Appreciate the work Blue.

[WildCard]

True, can you do me a favor and interpret those WAR Hextal charts between Risto and Gorges?

[woods-racer]

 

I think you should amend this to say the trained eye of a good coach/GM/scout. I would label the hierarchy as follows:

 

Good GM w/ good analytics team > good GM w/o analytics team > good GM w/ bad analytics team > bad GM w/ good analytics team > spreadsheet > bad GM w/o analytics team >= bad GM w/ bad analytics team. Joe Sakic and the Avs reside somewhere in those last two. You don't want to be Joe Sakic and the Avs.

 

Very true.