Fuzzy Analytics

[The CyNick]

There are intangibles like leadership and intimidation that simply cannot be measured statistically except that the team has a better record when player A is in the lineup.

Guys like Komarov and Martin set an example for the kids to follow. Without Polak the leafs D is ridiculously soft.

If Martin takes 100 shots from in front of the net to a stacked pad he will be a hero. You don't need to worry about silly things like players are afraid to go into a corner with him.

 

[Gary Nylund]

advanced

Define "advanced"

 

[Stand Witness]

But how are you measuring the effectiveness of the stats. You made a blanket statement saying advanced stats have been successful. But you can't illustrate how you are measuring success.

Seems like fuzzy analytics.

The R^2 value for having strong possession numbers and winning is greater than anything else we currently have.

 

[Mr Hockey*]

Define "advanced"

me ...

 

[Gary Nylund]

me ...

LOL. I'm glad we had this chat.

 

[Stand Witness]

Why do people always bring up Martin when trying to disprove underlying numbers? He is actually the best 4th liner in the league at suppressing shots. He doesn't generate any offense but he does prevent it and there is a role for that in the NHL when it comes to the PK and even possibly on your 4th line if that is one of the uses of it.

 

[Mr Hockey*]

LOL. I'm glad we had this chat.

 

[Menzinger]

You're missing the point.

PPG is in itself a reliable stat because I know what an assist, what a goal is and what a game played is. And by and large that doesn't change from arena to arena and league to league. Many advanced stats rely on other stats that are very unreliable (ie what is a turnover exactly? ) and each arena and each scorer in each arena has a different definition of what qualifies to count towards those tallies. So the stat itself is at best suspect. PPG is not.

Whether or not YOU feel PPG is a measure you should rely on to determine a players worth is up to you. But the stat itself is a million times more reliable. It's hard to debate that point.

Most of these advanced stats don't need to be 100% precise - they're about identifying longer term trends, so the discrepancies among arena stat recording tends to get evened out.

You can't really get a firm analysis of say a player's possession numbers from just a handful of games.

 

[Menzinger]

So the world came together and decided all new age stats are relevant? I must have missed that vote.

The pro analytics people just scoff at criticism. Most of the people who write about this stuff are just bloggers with little to no credibility.

The Leafs also just hired a bunch of those "bloggers" to help influence team decision making.

As a fan, there's no real need to pay attention to these things, but seems silly to in one post claim you've never really read up on how they're used and in another claim they're deeply flawed....

 

[buttman*]

So I want to start by saying I'm a complete analytics layman. I understand what they mean and do use them to support or challenge my eye test of players. This article is quite good and pretty much how I view and use analytics.

http://blueseatblogs.com/2016/07/24...systems-stats-tools-talent-evaluation-primer/

I would recommend it to anyone who is also an analytics layman.
As I was reading I came across some interesting things like this that got me thinking there is obviously some analytics we are not privy too:



This says to me that the true important analytics are looking for things other than what the market has already identified.

Also when dubas said this:



It made me think that Dubas already has a formula that extends past what we (or I) know as the norm for advanced analytics. He talks about how the total buy in has to be there for it to work, GM, coach, players etc.

So in light of the Martin and Polak signings with the subsequent HF grumblings of signing these players, I submit that there is a fuzzy analytic that says these are the guys we need because the formula we have designed needs X+Y÷Z to be a winning team.

Do any of you that are well versed in analytics know if these analytical formulas exist or have any theories or evidence that will either support or challenge my thoughts?

Good post. Martin and Polak signings are not about analytics. These are Lou signings. Lou understands you need a goalie. Lou also understands you need a certain amount of toughness in the lineup. This toughness needs to take a regular shift. If you don't have this toughness, this is where your good "analytic" players will get run over and intimidated no matter what the numbers say. We saw this for years when the Leafs played the Bruins. You can't have the kids getting run over every time they touch the puck.

 

[Nithoniniel]

But how are you measuring the effectiveness of the stats. You made a blanket statement saying advanced stats have been successful. But you can't illustrate how you are measuring success.

Seems like fuzzy analytics.

Here:

The R^2 value for having strong possession numbers and winning is greater than anything else we currently have.

Now, Cynick, how have you gone from identifying a potential problem (the way data is collected) to draw conclusions on the impact of said problem? Since you have such a strong stance on the subject, I assume you have based that on something solid and substantiated?

Good post. Martin and Polak signings are not about analytics. These are Lou signings. Lou understands you need a goalie. Lou also understands you need a certain amount of toughness in the lineup. This toughness needs to take a regular shift. If you don't have this toughness, this is where your good "analytic" players will get run over and intimidated no matter what the numbers say. We saw this for years when the Leafs played the Bruins. You can't have the kids getting run over every time they touch the puck.

From an analytical standpoint, what we saw was exactly what the numbers indicated. A dominant two-way team that would run over the defensively weak, quick strike team.

But I don't mean to take anything from your point. I think the problem with how analytics is often used is that people use conclusions drawn on the macro level to apply on the micro level. That said, we know that possession is good and it correlates to winning. We know that in general, a player with better possession metrics and/or scoring chance metrics is also the superior player. Applying that on an individual basis is called ergonomical statistical fallacy, because what's true in general doesn't have to be true in each specific case.

That's why an evaluation should never end there. You should look at individual factors. You should scout the player to learn qualitative measures. And you should consider intangibles, example setting and mental aspects. The hard part is measuring these things against each other. How much better does someone have to be to offset not being a leader, an example setter and someone who makes their teammates feel confident in physical situations? There are really no hard answers to that question.

Another part that I find very interesting any time contender building comes up is head-to-head ability. I've mentioned this several times before. For example, Tavares is a great player, however whenever he is on the ice against a top center he generally struggles pretty badly. This is not a problem in the regular season, but in a tough, even 7-game series? That can be a deciding factor. Whereas a guy like Bergeron is someone who generally owns the ice. He'll dominate possession, dominate scoring chances and dominate goal differential. This happens against the Bozaks and Desharnais against the world, just like Tavares dominates those guys. However, he dominates just as much against the best centers of the world. Having him on the ice on a 7-game series is a huge advantage, no matter who is on the other side.

I hope Matthews can be someone like that. Or at least one in the next tier, guys like Getzlaf and Crosby who plays even against anybody (other than Bergeron).

 

[saltming]

Good post. Martin and Polak signings are not about analytics. These are Lou signings. Lou understands you need a goalie. Lou also understands you need a certain amount of toughness in the lineup. This toughness needs to take a regular shift. If you don't have this toughness, this is where your good "analytic" players will get run over and intimidated no matter what the numbers say. We saw this for years when the Leafs played the Bruins. You can't have the kids getting run over every time they touch the puck.

Thanks!
I was wondering if there are analytics or a formula that defines the need for toughness. So you need X amount of skill combined with Y amount of toughness equals a winning team. Basically I am wondering if there are unknown analytics that teams have developed that are not in the main stream.
That said Lou and the gang realized they needed more Y and applied further analytics to identify their preferred type of player Y.
Nithoniniel has said he believes there is a formula for a team but not one for success which I also agree with.

 

[buttman*]

Thanks!
I was wondering if there are analytics or a formula that defines the need for toughness. So you need X amount of skill combined with Y amount of toughness equals a winning team. Basically I am wondering if there are unknown analytics that teams have developed that are not in the main stream.
That said Lou and the gang realized they needed more Y and applied further analytics to identify their preferred type of player Y.
Nithoniniel has said he believes there is a formula for a team but not one for success which I also agree with.

Good question. Not sure there is a formula. Not sure if you could create one. Depends on your personnel. When you have guys like Getzlaf and Perry in the top six -- you don't need much. Guys like Crosby and Malkin also don't seem to get intimidated. Still I would want some for Matthews, Nylander and Marner until they mature.

Edmonton will be an interesting case study since they added Lucic and others in the past year to rectify that problem.

 

[saltming]

Good question. Not sure there is a formula. Not sure if you could create one. Depends on your personnel. When you have guys like Getzlaf and Perry in the top six -- you don't need much. Guys like Crosby and Malkin also don't seem to get intimidated. Still I would want some for Matthews, Nylander and Marner until they mature.

Edmonton will be an interesting case study since they added Lucic and others in the past year to rectify that problem.

I started thinking about it when I was reading that Dubas said that it's important for the entire organization to buy into the analytics and direction. He said that's how the soo turned it around so quickly buy hiring Keefe and having him implement the right system for the players assembled.
Dubas also said that it was important for the management to supply the right type of players to support the coach and system to create a winning team.
All in all I found it fascinating.

 

[4thline]

But I don't mean to take anything from your point. I think the problem with how analytics is often used is that people use conclusions drawn on the macro level to apply on the micro level. That said, we know that possession is good and it correlates to winning. We know that in general, a player with better possession metrics and/or scoring chance metrics is also the superior player. Applying that on an individual basis is called ergonomical statistical fallacy, because what's true in general doesn't have to be true in each specific case.

That's why an evaluation should never end there. You should look at individual factors. You should scout the player to learn qualitative measures. And you should consider intangibles, example setting and mental aspects. The hard part is measuring these things against each other. How much better does someone have to be to offset not being a leader, an example setter and someone who makes their teammates feel confident in physical situations? There are really no hard answers to that question.

Another part that I find very interesting any time contender building comes up is head-to-head ability. I've mentioned this several times before. For example, Tavares is a great player, however whenever he is on the ice against a top center he generally struggles pretty badly. This is not a problem in the regular season, but in a tough, even 7-game series? That can be a deciding factor. Whereas a guy like Bergeron is someone who generally owns the ice. He'll dominate possession, dominate scoring chances and dominate goal differential. This happens against the Bozaks and Desharnais against the world, just like Tavares dominates those guys. However, he dominates just as much against the best centers of the world. Having him on the ice on a 7-game series is a huge advantage, no matter who is on the other side.

I hope Matthews can be someone like that. Or at least one in the next tier, guys like Getzlaf and Crosby who plays even against anybody (other than Bergeron).

I think I love you. Right on the goddamn money, especially the head to head bit. I've been butting my head up against the QoC doesn't impact macro analysis therefore QoC doesn't matter crowd but you can see it clearly in the HtH.

 

[4thline]

Good post. Martin and Polak signings are not about analytics. These are Lou signings. Lou understands you need a goalie. Lou also understands you need a certain amount of toughness in the lineup. This toughness needs to take a regular shift. If you don't have this toughness, this is where your good "analytic" players will get run over and intimidated no matter what the numbers say. We saw this for years when the Leafs played the Bruins. You can't have the kids getting run over every time they touch the puck.

I disagree, I think Polak and Martin were both very much analytics signings, just not as you would think. The toughness is certainly part of it, but we could add anyone for toughness, we got these guys, why?

For example in the case of Polak there's a big analytics driven reason IMO. PK. On the pk Polak's transition struggles and lack of puck skills don't matter as much. His defensive awareness, mobility, and bull strength let him stamp out cycles, clear the crease, grab pucks and ice them. His allSH CA/60 is 38rth out of 155 for dmen with atleast 200 sh minutes over the last three years. Not a bad add to hide on the 3rd pairing, especially with our transition heavy d.



The talent pool is too shallow to have a team full of all around guys. Analytics help you do a better job of finding the right "utility" guys that can play a specific role in driving effective hockey for your team

 

[buttman*]

I disagree, I think Polak and Martin were both very much analytics signings, just not as you would think. The toughness is certainly part of it, but we could add anyone for toughness, we got these guys, why?

For example in the case of Polak there's a big analytics driven reason IMO. PK. On the pk Polak's transition struggles and lack of puck skills don't matter as much. His defensive awareness, mobility, and bull strength let him stamp out cycles, clear the crease, grab pucks and ice them. His allSH CA/60 is 38rth out of 155 for dmen with atleast 200 sh minutes over the last three years. Not a bad add to hide on the 3rd pairing, especially with our transition heavy d.



The talent pool is too shallow to have a team full of all around guys. Analytics help you do a better job of finding the right "utility" guys that can play a specific role in driving effective hockey for your team

Except Polak's analytics are bad overall and 38th out of 155 for a given situation doesn't change that. Lou had tough teams in Jersey -- or at the very least teams that did not get pushed around. Martin and Polak are very much about toughness. Babcock has gone on record saying they will "protect" Matthews.

 

[Mr Hockey*]

Martin and Polak were signed as team "needs", analytically they are overpaid and arguably not NHL worthy.

 

[4thline]

Except Polak's analytics are bad overall and 38th out of 155 for a given situation doesn't change that. Lou had tough teams in Jersey -- or at the very least teams that did not get pushed around. Martin and Polak are very much about toughness. Babcock has gone on record saying they will "protect" Matthews.

You didn't read the second half of the top line,

"The toughness is certainly part of it, but we could add anyone for toughness, we got these guys, why?"

We didn't just go get guys to add toughness. We got guys that added toughness in addition to a high degree of effectiveness at another facet of the game. That' differentiator analytics brings. We could have added any physical 4thliner and 3rd pair d, but we specifically got an upper echelon pk guy (where teams score 16-28% of their goals in a much smaller percentage of time) and an extremely effective forechecker.

And bad overall is a bit of an overstatement considering he's 48/164 at 5v5 CA. He's basically the anti-Gardiner. Good at inzone defense, terrible at transition, and a nice complement to the d-core at his price point.

 

[The CyNick]

Here:



Now, Cynick, how have you gone from identifying a potential problem (the way data is collected) to draw conclusions on the impact of said problem? Since you have such a strong stance on the subject, I assume you have based that on something solid and substantiated?

Im a bigger fan of the team stat of goals for vs against in a game. I find that has a very heavy correlation to success.

 

[The CyNick]

I find it funny that people need analytics to tell them Bergeron is a very good player. Watch a Bruins game or two. You will see it.

 

[Mr Hockey*]

I find it funny that people need analytics to tell them Bergeron is a very good player. Watch a Bruins game or two. You will see it.

This is not a good point to be using in a anti analytics debate Tells me you are 100% uninformed on advanced stats

 

[buttman*]

You didn't read the second half of the top line,

We didn't just go get guys to add toughness. We got guys that added toughness in addition to a high degree of effectiveness at another facet of the game. That' differentiator analytics brings. We could have added any physical 4thliner and 3rd pair d, but we specifically got an upper echelon pk guy (where teams score 16-28% of their goals in a much smaller percentage of time) and an extremely effective forechecker.

And bad overall is a bit of an overstatement considering he's 48/164 at 5v5 CA. He's basically the anti-Gardiner. Good at inzone defense, terrible at transition, and a nice complement to the d-core at his price point.

You're missing my point -- or trying to pretend it was something else. These weren't purely analytic signings -- pretty obvious. They were signed because they brought toughness as well. Certainly todays tough guys aren't Colton Orr's. Rather they need to have the ability to play a regular shift. Which these guys can. But it is quite obvious these guys weren't purely signed because of good or mediocre analytics -- they bring toughness -- which helps others play bigger.

 

[The CyNick]

This is not a good point to be using in a anti analytics debate Tells me you are 100% uninformed on advanced stats

So many shakey head emotions...

 

[The CyNick]

You're missing my point -- or trying to pretend it was something else. These weren't purely analytic signings -- pretty obvious. They were signed because they brought toughness as well. Certainly todays tough guys aren't Colton Orr's. Rather they need to have the ability to play a regular shift. Which these guys can. But it is quite obvious these guys weren't purely signed because of good or mediocre analytics -- they bring toughness -- which helps others play bigger.

How do we measure toughness in this new informed world?

Is it punches taken over chirps per 60?

But then what qualifies as a chirp? I'm sure the stat will work out anyway.