Colorado granted permission to talk to Dubas

[herman]

I don't know if taking a strictly analytical approach is a surefire way to win either.  Florida by and large did this, and it seemed to crater their team.  Injuries played a part in their demise for sure, but it seemed that they didn't have the same mojo as the year before.

Ultimately it's probably best to have someone in place who uses these things as a tool to perform the evaluation of the team, but doesn't fall in love with them to the point where it's dictating moves.  I think that for the most part that describes what Dubas did in Sault Ste. Marie.

When they start talking about adding more "intangibles" to the team, that's where I think the problems will start.

Florida's 'approach' was unfortunately exceedingly tone-deaf and they clearly did not have up-and-down buy-in, which would crater any enterprise. The jump they had last year peaks in that nebulous 1st/2nd round exit region and will likely persist until something more drastic happens.

Like you said, I think Dubas has the ability to help both traditionalists and analytics-inclined people see a unified path. He has literally walked both paths in his career in both scouting and GMing and found going down either one to the extreme to be detrimental to success.

At least with all three candidates, Dubas, Lou, Hunter, they're very process and relationship driven. Lou's got some blindspots (presumably from the bling off his rings) and anachronistic preferences; Hunter has an obvious pattern that eventually manifests as a weakness (top-heavy build, no defense); Dubas is super young but seems very open to shoring up advantages at every controllable layer. He did fumble a team sexual assault scandal that I think he was in way over his head on back in 2012 -- probably not something a sports management degree adequately deals with.

 

[Kin]

Just so we're clear, what exactly is the analytical method or statistical model that Florida used that led to Ekblad, Huberdeau, Barkov, Bjugstad and Luongo all getting hurt?

 

[rdowdall]

Just so we're clear, what exactly is the analytical method or statistical model that Florida used that led to Ekblad, Huberdeau, Barkov, Bjugstad and Luongo all getting hurt?

Again, injuries definitely played a part in their season, but the moves to get rid of Gudbranson and Kulikov as well as the decision not to sign Campbell were largely driven from a analytical place, and that played a part in not having the depth on defense to absorb a loss of Ekblad.  Also the decision to fire Gallant was rumored to have been made because his coaching style didn't align with an analytical approach.

I'm not saying that there is a problem with using analytics as a tool.  But when you move three of your top 6 d-men, when you know a 4th isn't coming back based largely on the data that you are getting from advanced stats, that would seem like poor decision making. 

 

[Peter D.]

But I can't help but think the original 'Shanaplan' has been hijacked a little bit.

If this is indeed the case, why would Shanahan allow for this as opposed to putting his foot down?

 

[rdowdall]

But I can't help but think the original 'Shanaplan' has been hijacked a little bit.

If this is indeed the case, why would Shanahan allow for this as opposed to putting his foot down?

Because he doesn't want the Leafs to be known as an organization that will block your personal growth? 

We don't know what is being said behind closed doors.  Shanahan could very well be saying to Dubas "I understand the desire to run your own team, but I think taking that job would be a mistake."

 

[rdowdall]

But for years, going back to his OHL days, Dubas has shown that he could very likely be the right person to lead an analytical approach. Which yes, while I call it that, of course is not a team that makes every decision based solely on numbers from a spreadsheet. Traditional scouting methods and getting to know a player personally will always play a role too. And it's also about a new way of viewing how the game is playing and being the first to spot new trends that can help ultimately win hockey games. That's what Dubas did in the Soo. And prior to Lou putting his stamp on the team that's what he appeared to be doing with the Leafs.

So I'm for keeping Dubas.  I should have made that point.  I think the Leafs should keep him because I think he has shown that he is a person who can grab on to emerging trends quickly and manipulate them to get an edge.  The growth of analytics was one of those trends.  All I was trying to say was that we shouldn't be keeping Dubas simply because he embraces analytics.  We should be keeping him because he appears to be a smart, capable person who will be able to come up with creative soutions to the problems that the Leafs will face while trying to build a contender.

My fear with a Hunter/Lou tandem is that they are too similar, and therefore, might get stuck in a rut without a way to get out of it.

 

[CarltonTheBear]

Again, injuries definitely played a part in their season, but the moves to get rid of Gudbranson and Kulikov as well as the decision not to sign Campbell were largely driven from a analytical place, and that played a part in not having the depth on defense to absorb a loss of Ekblad.  Also the decision to fire Gallant was rumored to have been made because his coaching style didn't align with an analytical approach.

I'm not saying that there is a problem with using analytics as a tool.  But when you move three of your top 6 d-men, when you know a 4th isn't coming back based largely on the data that you are getting from advanced stats, that would seem like poor decision making. 

Kulikov was traded for a better defanceman. Gudbranson was replaced by Demers via free agency, that's a pretty massive upgrade for the Panthers. Campbell was replaced by Yandle via free agency. While I don't think Yandle is an upgrade over what Campbell brought to the Panthers in years past, he's a better player than a 38-year old Brian Campbell is today.

 

[Kin]

Again, injuries definitely played a part in their season, but the moves to get rid of Gudbranson and Kulikov as well as the decision not to sign Campbell were largely driven from a analytical place, and that played a part in not having the depth on defense to absorb a loss of Ekblad. 

But they didn't just "get rid" of those guys. They replaced them with Yandle, Demers and Pysyk. Teams don't keep 8 or 9 NHL quality defense around in case of injuries.

Are you really going to make the case that a bigger reason for their decline is the relative difference between Yandle, Demers and Pysyk vs. Gudbranson, Kulikov and Campbell as opposed to things like their top 2 centres playing a combined 92 games?

 

[Kin]

But I can't help but think the original 'Shanaplan' has been hijacked a little bit.

If this is indeed the case, why would Shanahan allow for this as opposed to putting his foot down?

I think that's only a big question if you thought of Shanahan as some grand mastermind architect as opposed to someone who was always collaborative and somewhat green at the whole team running thing.

 

[rdowdall]

Again, injuries definitely played a part in their season, but the moves to get rid of Gudbranson and Kulikov as well as the decision not to sign Campbell were largely driven from a analytical place, and that played a part in not having the depth on defense to absorb a loss of Ekblad.  Also the decision to fire Gallant was rumored to have been made because his coaching style didn't align with an analytical approach.

I'm not saying that there is a problem with using analytics as a tool.  But when you move three of your top 6 d-men, when you know a 4th isn't coming back based largely on the data that you are getting from advanced stats, that would seem like poor decision making. 

Kulikov was traded for a better defanceman. Gudbranson was replaced by Demers via free agency, that's a pretty massive upgrade for the Panthers. Campbell was replaced by Yandle via free agency. While I don't think Yandle is an upgrade over what Campbell brought to the Panthers in years past, he's a better player than a 38-year old Brian Campbell is today.

I don't think Pysyk is a better dman than Kulikov, not when you factor in experience and familiarity with the team:

http://www.diebytheblade.com/2016/6/25/12029694/buffalo-sabres-trade-mark-pysyk-florida-panthers-dmitry-kulikov-and-picks

"Mark Pysyk was always a fan favorite, thanks to his goofy smile, smooth passes, and good possession numbers. What he failed to bring to the team was offense in terms of point production and any kind of physicality. Pysyk's career high in points came this year, when he scored 11 - not great for a former first round pick. Kulikov, a former first-rounder himself, has had multiple 20 point seasons and a career high of eight goals - basically the Sabres traded good possession numbers and passing for more point production and physicality."

So I think this change on defence in the short term would hurt them.

There is an overall comparison on the trades here:

http://thehockeywriters.com/comparing-florida-panthers-three-new-defensemen/

In each case, analytics was used to determine that the Florida panthers got the better of the deal. 

However, the results didn't pan out on the ice.  Some of it's injuries, but if you have greatly improved you defence based on advanced stats, should that not provide a bit of slack for those injuries?

Also if Ekblad is playing with a better defensive core, should he not have had a better year?

It seems like Rowe went in to the offseason and said I'm going to build the best defence I can based on advanced stats.  It may have been more prudent to show some restraint and slowly pick away at areas that are problems that advanced stats may have highlighted, while maintaining a level on continuity with your defensive core.

 

[Peter D.]

I think that's only a big question if you thought of Shanahan as some grand mastermind architect as opposed to someone who was always collaborative and somewhat green at the whole team running thing.

That pretty much debunks the media and fans notion of a 'Shanaplan'.

 

[Kin]

I think that's only a big question if you thought of Shanahan as some grand mastermind architect as opposed to someone who was always collaborative and somewhat green at the whole team running thing.

That pretty much debunks the media and fans notion of a 'Shanaplan'.

Yes, well, yes.

 

[Kin]

"Mark Pysyk was always a fan favorite, thanks to his goofy smile, smooth passes, and good possession numbers. What he failed to bring to the team was offense in terms of point production and any kind of physicality. Pysyk's career high in points came this year, when he scored 11 - not great for a former first round pick. Kulikov, a former first-rounder himself, has had multiple 20 point seasons and a career high of eight goals - basically the Sabres traded good possession numbers and passing for more point production and physicality."

So I think this change on defence in the short term would hurt them.

Except that didn't turn out to be true. Kulikov had 17 points for the Panthers last year and had 5 in 47 games this year for the Sabres. Meanwhile Pysyk had 17 points for the Panthers.

However, the results didn't pan out on the ice.  Some of it's injuries, but if you have greatly improved you defence based on advanced stats, should that not provide a bit of slack for those injuries?

Yes, a bit. Not enough to compensate for injuries to your #1 defenseman, #1 goalie and #1 and #2 centres.

Also if Ekblad is playing with a better defensive core, should he not have had a better year?

Not if, say, he's playing hurt all year and is also playing without guys like Barkov and Huberdeau in the line-up. Also, you know, individual play does exist in the NHL and guys can have bad years independent of the players around them.

It seems like Rowe went in to the offseason and said I'm going to build the best defence I can based on advanced stats. 

And you haven't really shown any evidence that they didn't do that, or at the very least, didn't build a comparable defence. All you've done is said "well, they had less points as a team so the decisions they made must have been bad".

I mean, Demers had 23 points last year and was a +16. Gudbranson had 9 points and was a +3. Advanced stats or traditional stats, adding him is an upgrade.

 

[CarltonTheBear]

Nik tagged in there and made a number of points I was looking at, including debunking the whole 'Kulikov scores more than Pysyk" claim. So I'll just say that...

Yandle-Ekblad
Matheson-Demers
Petrovic-Pysyk

If healthy, that would have been a top-5 defence corp in the East this season. Maybe even top-3. Washington and Tampa would have been the only teams I'd for sure put ahead of them.

 

[rdowdall]

And you haven't really shown any evidence that they didn't do that, or at the very least, didn't build a comparable defence. All you've done is said "well, they had less points as a team so the decisions they made must have been bad".

I mean, Demers had 23 points last year and was a +16. Gudbranson had 9 points and was a +3. Advanced stats or traditional stats, adding him is an upgrade.

I didn't say the decisions were bad.  I said it may have been better if they had taken a different approach, one that didn't appear to be based on just looking at the overall advanced stats of a player.  They may have been a better team this year if they had kept Gudbranson, despite his bad possession numbers, because he brought something else to the team.  They may very well have been worse.

If team building is all about stats than  anyone with the ability to close a deal would be a great GM if all they did was look at the advanced stats.  Basically look at your team, figure out which players don't have good advanced stats, and trade them for players that do.  I believe there has to be more to team building than that.  Overall that seems to be too simple.  And switching out 4 of 6 d-men on a team in one summer seems like a huge overhaul.

 

[Kin]

I didn't say the decisions were bad.  I said it may have been better if they had taken a different approach, one that didn't appear to be based on just looking at the overall advanced stats of a player.  They may have been a better team this year if they had kept Gudbranson, despite his bad possession numbers, because he brought something else to the team.  They may very well have been worse.

Well, no. What you said was that using analytics the way they did seems to have "cratered the team". What I'm responding to is the approach of looking at the team that was probably most ravaged by injuries in the entire league and saying "Man, look at all those analytical decisions" as the chief reason the team sank.

If team building is all about stats than  anyone with the ability to close a deal would be a great GM if all they did was look at the advanced stats.  Basically look at your team, figure out which players don't have good advanced stats, and trade them for players that do.  I believe there has to be more to team building than that.  Overall that seems to be too simple.  And switching out 4 of 6 d-men on a team in one summer seems like a huge overhaul.

It seems to be too simple because you're presenting it in a very simplistic way. For starters "stats" isn't any one thing. There are lots of analytical approaches to evaluating players, none of them definitive and none of them even claim to have reduced a player down to a single integer or rating. Using analytics as a means of decision making is no more or less likely to lead you to assembling any one kind of team than using scouting is. You can scout for various attributes, various analytical methods measure different things.

Likewise, a lot of analytical measurements tell us things that aren't really surprises. Sure it broadly emphasizes chance creation and possession but if analytics tell us that Sidney Crosby and Connor McDavid are the two best players in the game that doesn't make trading for them easier. Especially when lots of teams are looking at the same analytics.

Remember, the ethos that came out of Moneyball wasn't "use stats to measure players" but rather "look at what people are overvaluing and try to squeeze value of what they're undervaluing". But if what people are valuing are the skills that actually go into winning you can't game the system.

But more to the point not even the most fervent disciple of using analytics would advocate using them to the exclusion of any other consideration. It's a bit of a straw man.

 

[rdowdall]

Well, no. What you said was that using analytics the way they did seems to have "cratered the team". What I'm responding to is the approach of looking at the team that was probably most ravaged by injuries in the entire league and saying "Man, look at all those analytical decisions" as the chief reason the team sank.

Actually what I said was:

"I don't know if taking a strictly analytical approach is a surefire way to win either.  Florida by and large did this, and it seemed to crater their team.  Injuries played a part in their demise for sure, but it seemed that they didn't have the same mojo as the year before."

So the problem is my use of the term  "analytical approach".  I had meant this as strictly looking at possesion numbers.  Analytical approach in that statement meant using advanced stats solely as they relate to possession numbers.  Which is what was rumoured Rowe did in the offseason.  Looking only at possesion numbers, and not the other aspects of what a player can bring, doesn't seem destined to build a winner over any other way. 

Whether that was the main reason they stumbled through the season is up for debate.  I don't think that the decisions Rowe made paid off the way he had hoped.  If it was all just the injuries, then he probably made the right moves last offseason. 

It seems to be too simple because you're presenting it in a very simplistic way. For starters "stats" isn't any one thing. There are lots of analytical approaches to evaluating players, none of them definitive and none of them even claim to have reduced a player down to a single integer or rating. Using analytics as a means of decision making is no more or less likely to lead you to assembling any one kind of team than using scouting is. You can scout for various attributes, various analytical methods measure different things.

Likewise, a lot of analytical measurements tell us things that aren't really surprises. Sure it broadly emphasizes chance creation and possession but if analytics tell us that Sidney Crosby and Connor McDavid are the two best players in the game that doesn't make trading for them easier. Especially when lots of teams are looking at the same analytics.

Remember, the ethos that came out of Moneyball wasn't "use stats to measure players" but rather "look at what people are overvaluing and try to squeeze value of what they're undervaluing". But if what people are valuing are the skills that actually go into winning you can't game the system.

But more to the point not even the most fervent disciple of using analytics would advocate using them to the exclusion of any other consideration. It's a bit of a straw man.

The simplification comes from my use of the term analytics to only refer to possession numbers.  In this case I think they are a useful tool for a GM to use, but shouldn't always be used the be all and the end all when it comes to evaluating their team.

Yes in general analytics should mean the generalized analysis of any given data set. 

 

[herman]

I think that's only a big question if you thought of Shanahan as some grand mastermind architect as opposed to someone who was always collaborative and somewhat green at the whole team running thing.

That pretty much debunks the media and fans notion of a 'Shanaplan'.

Yes, well, yes.

I was always under the impression that the 'Plan' was really more of a 'Process'.

 

[CarltonTheBear]

In this case I think they are a useful tool for a GM to use, but shouldn't always be used the be all and the end all when it comes to evaluating their team.

I can tell you that this phrase basically makes anybody who follows analytics blood boil. Nobody in that community thinks they are the "be all and end all".

 

[rdowdall]

In this case I think they are a useful tool for a GM to use, but shouldn't always be used the be all and the end all when it comes to evaluating their team.

I can tell you that this phrase basically makes anybody who follows analytics blood boil. Nobody in that community thinks they are the "be all and end all".

I hate nomenclature.

Is Rowe in the community or not?  If he uses analytics, but doesn't understand them, or if he uses them and does think they are the be all and the end all, does that revoke his community card?  I have no idea if Tom Rowe is in the analytics community.

Someone can use a hoe wrong in the sense that they think it's the only tool you need to garden.  Gardeners probably get upset if you say that a hoe is the be all and the end all when it comes to gardening.  If you only use a hoe are you in the gardening community? 

I didn't say that all people in analytics say they are the be all and the end all.  If a GM makes a choice and weighs one data set heavily over the complete amount of data available, then that will probably lead to mistakes.