[Discussion: where's the proof Benning should rely on analytics?

[Bob Long]

So I'm posting this for 2 reasons:

 

1. The GudBranson trade seems to have highlighted the differences between the advanced stats crowd and the old school character group (which I'm in), and

2. I'm hoping to learn more about why/if Benning should care about analytics.

 

Right now I see no real use for analytics as a decision making tool. I do not see how it brings anything really new to the picture - we already know teams need to have good puck possession and they also need good character. 

 

So.... does anyone have or can they point to proof that analytics should be the thing Benning relies on more than scouting experience?

[thejazz97]

It's like vitamins. They're good to have, but you can't rely on them.

[Crabcakes]

Benning uses analytics.  He has an analytics guy.  Jonathan Wall, Director of Hockey Op's and Analytics.  It's a complimentary tool but there are other tools in his tool box.

 

It is well known that Benning puts a lot of stock in the character of players.  This is one thing that is not measured by analytics. 

[Alflives]

3 minutes ago, Crabcakes said:

Benning uses analytics.  He has an analytics guy.  Jonathan Wall, Director of Hockey Op's and Analytics.  It's a complimentary tool but there are other tools in his tool box.

 

It is well known that Benning puts a lot of stock in the character of players.  This is one thing that is not measured by analytics. 

I wonder if character for JB includes a bunch of intangibles: toughness, standing up for teammate's, willingness to take hits, glue guy - that kind of stuff?  

[Qwags]

3 minutes ago, Alflives said:

I wonder if character for JB includes a bunch of intangibles: toughness, standing up for teammate's, willingness to take hits, glue guy - that kind of stuff?  

Gudbranson fits that bill perfectly, so I would say yes.

[Derp...]

I think the day that Analytics can prove something about what type of team it takes to win a Stanley cup it will be more widely accepted, for now the player by player comparisons are a little pointless given all of the variables that go unaccounted for.

 

One day, but not today.

[MeanSeanBean]

2 minutes ago, Derp... said:

I think the day that Analytics can prove something about what type of team it takes to win a Stanley cup it will be more widely accepted, for now the player by player comparisons are a little pointless given all of the variables that go unaccounted for.

 

One day, but not today.

On this day we fight!

 

Sorry, I had to go Lord of the rings on that one. It was so close. 

[Bob Long]

49 minutes ago, Derp... said:

I think the day that Analytics can prove something about what type of team it takes to win a Stanley cup it will be more widely accepted, for now the player by player comparisons are a little pointless given all of the variables that go unaccounted for.

 

One day, but not today.

 

I actually work in R&D and have a good understanding of basic stats and I just don't see - yet - the value. Sure you get some additional info, but nothing that you already didn't know enough of already to make a decision on a player. Did anyone know from analytics that Dale Weise could be more than Vigneault had him doing? Or was it the Montreal coaching staff that saw something in his game through scouting? The way AV used Weise wouldn't have generated the stats needed, and imo only an experienced eye in Montreal saw a bigger scoring role for him there. 

 

I also don't see anything on the analytics boards that dig deeper into, or value, specific roles - Gary Valk had a good example of guys that stay back and support a puck moving D partner - those guys that stay back will have worse stats but they play a very important role and pay a statistical price for their role. 

 

However, I'm really curious to see if anyone on the boards can 'tell me i'm wrong' as the 1040 guys say. Given that most of the hoopla around GubBranson circles around the stats issue, I'd like to see the evidence for that stance. 

 

This goofy article is a perfect example: "Panthers take advantage of Canucks in lopsided trade" (http://www.tsn.ca/panthers-take-advantage-of-canucks-in-lopsided-trade-1.496247)

 

Nice chart, but no evidence that his analysis has any correlation to player or team success. That's the issue - you can select any set of data to support your preconception, but if he was so bad, why did he get 26 minutes per night in the playoffs and be considered by his team and guys like Pierre Lebrun to be the toughest member of the Pathers to play against?

 

[baumerman77]

Why should any management group including Benning value analytics? Simple, some phenomena are counter-intuitive (for example a lot of hockey minds use to think that hits and blocked shots are correlated to winning when in fact the opposite is true). Also, humans are susceptible to biases and errors in reasoning by using data and explicitly outlining the methodology utilized it invites more objective analysis and thus more objective results. 

 

This topic might as well have been "why should biologists rely on the scientific method?" or "why should psychologists rely on the scientific method?" 

 

The question is not so much what current insights analytics yields now but rather it is more a question of how one goes about making decisions. Benning relies on his own (and his management groups) acquired knowledge over the span of their hockey careers. Now, I know to a lot of people that sounds great. But for the aforementioned reasons it is not the optimal way of making decisions. In the sciences nobody accepts a scientist's theory just because the person is respected or has been a scientist on the topic for decades. The theory stands and falls based on the data and the reasoning. Essentially with Benning we have a guy and management group that makes decisions based on his/their accumulated knowledge without really questioning if his/their beliefs could be wrong, outdated or imprecise - he won't know because he doesn't seem to value the methods that would illuminate the possible errors. 

 

Personally when I hear things like what Benning said about analytics being like vitamins and what Burke said about them being for illumination not support - I immediately think that they are likely to accept analytics when they support their preconceived opinions and reject or rationalize analytics away when they run counter to their preconceived opinions - because they value their opinions much more than analytics. If so this application makes analytics useless. 

[Bob Long]

1 hour ago, Alflives said:

I wonder if character for JB includes a bunch of intangibles: toughness, standing up for teammate's, willingness to take hits, glue guy - that kind of stuff?  

You have to. Where's "leadership" on the stats tables? If this guy really is considered future C material thats something I don't think anyone saw in McCann, but did in Horvat e.g. Boeser also seems to be 'that guy' who brings character as well. Maybe that's what Benning sees in Dubois as well, I don't know. 

[oldnews]

1 hour ago, Bob Long said:

So I'm posting this for 2 reasons:

 

1. The GudBranson trade seems to have highlighted the differences between the advanced stats crowd and the old school character group (which I'm in), and

2. I'm hoping to learn more about why/if Benning should care about analytics.

 

Right now I see no real use for analytics as a decision making tool. I do not see how it brings anything really new to the picture - we already know teams need to have good puck possession and they also need good character. 

 

So.... does anyone have or can they point to proof that analytics should be the thing Benning relies on more than scouting experience?

 

The "advanced analytics crowd" is a flattering description - and doesn't necessarily represent.

 

The competence or credibility of the particular people in the "analytics" crowd that are unfortunately far too outspoken - is questionable at best.

 

People like Drance or Yost simply aren't competent enough in the field to be providing player evaluations.

[Bob Long]

19 minutes ago, baumerman77 said:

Why should any management group including Benning value analytics? Simple, some phenomena are counter-intuitive (for example a lot of hockey minds use to think that hits and blocked shots are correlated to winning when in fact the opposite is true). Also, humans are susceptible to biases and errors in reasoning by using data and explicitly outlining the methodology utilized it invites more objective analysis and thus more objective results. 

 

This topic might as well have been "why should biologists rely on the scientific method?" or "why should psychologists rely on the scientific method?" 

 

The question is not so much what current insights analytics yields now but rather it is more a question of how one goes about making decisions. Benning relies on his own (and his management groups) acquired knowledge over the span of their hockey careers. Now, I know to a lot of people that sounds great. But for the aforementioned reasons it is not the optimal way of making decisions. In the sciences nobody accepts a scientist's theory just because the person is respected or has been a scientist on the topic for decades. The theory stands and falls based on the data and the reasoning. Essentially with Benning we have a guy and management group that makes decisions based on his/their accumulated knowledge without really questioning if his/their beliefs could be wrong, outdated or imprecise - he won't know because he doesn't seem to value the methods that would illuminate the possible errors. 

 

Personally when I hear things like what Benning said about analytics being like vitamins and what Burke said about them being for illumination not support - I immediately think that they are likely to accept analytics when they support their preconceived opinions and reject or rationalize analytics away when they run counter to their preconceived opinions - because they value their opinions much more than analytics. If so this application makes analytics useless. 

 

But where is the evidence that the information is being used scientifically by hockey GMs? The article I posted above is a perfect example of the author picking a limited data set to prove his point, vs. looking at a much broader set and letting the data tell you what the actual correlation is. I don't see the analytics guys doing that yet. Now that may be exactly whats going on behind the scenes and teams might be hanging on to that info and not making it public, but in terms of the "analysis" of the GudBranson trade nothing I saw supports the kind of comments made like "lopsided". 

 

The blocked shots example is an interesting one, and a hand-picked stat view again is potentially misleading. Are the stats skewed because crappier teams block shots more and don't have the puck as much? Does that mean a good possession team should never block shots, or assemble a team of guys unwilling to block shots because they had good possession numbers? Of course not, there has to be a good balance. 

 

 

 

 

 

[oldnews]

4 minutes ago, Bob Long said:

 

But where is the evidence that the information is being used scientifically by hockey GMs? The article I posed above is a perfect example of the author picking a limited data set to prove his point, vs. looking at a much broader set and letting the data tell you what the actual correlation is. I don't see the analytics guys doing that yet. Now that may be exactly whats going on behind the scenes and teams might be hanging on to that info and not making it public, but in terms of the "analysis" of the GudBranson trade nothing I saw supports the kind of comments made like "lopsided". 

 

 

 

The use of 'analytics' with respect to Gudbranson is the perfect example of useless, oversimplified and cherry-picked metrics - that don't begin to represent the player.

 

Few better examples of the maxim:

 

"A little bit of knowledge is a dangerous thing."

[Bob Long]

8 minutes ago, oldnews said:

The use of 'analytics' with respect to Gudbranson is the perfect example of useless, oversimplified and cherry-picked metrics - that don't begin to represent the player.

 

Few better examples of the maxim:

 

"A little bit of knowledge is a dangerous thing."

 

Its also routinely ignored. The Sedin's are producing at rates of guys 10 years younger (the stats prove it http://canucksarmy.com/2015/12/22/how-do-the-sedins-stack-up-against-the-nhl-s-best-dynamic-duos) but they are constantly slammed as being the "sisters" and not tough enough by much of the league and by our own fan base as past their due dates. 

 

What bugs me most I guess about the stats guys is they use their preconceptions just as much as old school guys but try to use the old trick of having "the numbers" on their side as a way of elevating their argument.  Politicians use polling data this way every day to justify one side or another. But I'm ready to convert -show me the objective study that shows analytics top experience and I'll dust off my TI-83 calculator. 

 

[Ghostsof1915]

The four teams in the past two Stanley Cup finals have been: Chicago/Tampa/Pittsburgh/San Jose. 

 

Chicago: Stan Bowman- Scotty Bowman Consultant. Scotty is a legendary/coach executive and a wealth of knowledge. 

Tampa Bay: Steve Yzerman - Star player from a good system like Detroit/Mentor Scotty Bowman.

Pittsburgh: Jim Rutherford- NHL Player/GM Hurricanes Cup win 2006, GM Penguins, make what looked like nutty trades but he's in the Finals, with both Crosby and Malkin not on the top of their game, yet. 

San Jose: Doug Wilson, also ex-NHL'er. 

 

Seems like most of those GM's had NHL or extensive hockey experience. Looks like stats guys are in markets like Florida and Arizona. Teams that were are the right path before ownership went young and advanced stats heavy. Stats have their place, but shouldn't be the only factor you weigh in. 

[Bob Long]

1 hour ago, thejazz97 said:

It's like vitamins. They're good to have, but you can't rely on them.

Vitamins make expensive pee. A traditional diet is all you need 

[oldnews]

9 minutes ago, Bob Long said:

 

Its also routinely ignored. The Sedin's are producing at rates of guys 10 years younger (the stats prove it http://canucksarmy.com/2015/12/22/how-do-the-sedins-stack-up-against-the-nhl-s-best-dynamic-duos) but they are constantly slammed as being the "sisters" and not tough enough by much of the league and by our own fan base as past their due dates. 

 

What bugs me most I guess about the stats guys is they use their preconceptions just as much as old school guys but try to use the old trick of having "the numbers" on their side as a way of elevating their argument. But I'm ready to convert -show me the objective study that shows analytics top experience and I'll dust off my TI-83 calculator. 

 

Absolutely correct.  Attempts to one-up every Benning move are so in-vogue with the (ironically simple and) arrogant analytics kids that it's has lead to some pretenders repeatedly embarrassing themselves with contrived and pedestrian attempts to misrepresent the objective outcomes of players like Gudbranson.

 

It would be one thing if the numbers actually indicated what they're attempting to maintain, but even a cursory look and their positions fall to pieces - from a statistical and analytical perspective - let alone from a more 'traditional' perspective.

 

The irony is that we have some apologists for bad analytics in this thread and on these boards - but they are incapable and unwilling to actually take a position and attempt to provide an analysis of Gudbranson's underlying numbers.  The most outspoken of the pedants prattle on and on about analytics, but never manage to produce any.

 

[Alflives]

1 hour ago, Bob Long said:

You have to. Where's "leadership" on the stats tables? If this guy really is considered future C material thats something I don't think anyone saw in McCann, but did in Horvat e.g. Boeser also seems to be 'that guy' who brings character as well. Maybe that's what Benning sees in Dubois as well, I don't know. 

Guance seems to that "it" factor too.  It's like these guys are the ones who would lead their soldiers into battle, and others will follow. I mean this in the literal sense too.  There is that "it" these guys have.  

[Derp...]

2 hours ago, Bob Long said:

 

I actually work in R&D and have a good understanding of basic stats and I just don't see - yet - the value. Sure you get some additional info, but nothing that you already didn't know enough of already to make a decision on a player. Did anyone know from analytics that Dale Weise could be more than Vigneault had him doing? Or was it the Montreal coaching staff that saw something in his game through scouting? The way AV used Weise wouldn't have generated the stats needed, and imo only an experienced eye in Montreal saw a bigger scoring role for him there. 

 

I also don't see anything on the analytics boards that dig deeper into, or value, specific roles - Gary Valk had a good example of guys that stay back and support a puck moving D partner - those guys that stay back will have worse stats but they play a very important role and pay a statistical price for their role. 

 

However, I'm really curious to see if anyone on the boards can 'tell me i'm wrong' as the 1040 guys say. Given that most of the hoopla around GubBranson circles around the stats issue, I'd like to see the evidence for that stance. 

 

This goofy article is a perfect example: "Panthers take advantage of Canucks in lopsided trade" (http://www.tsn.ca/panthers-take-advantage-of-canucks-in-lopsided-trade-1.496247)

 

Nice chart, but no evidence that his analysis has any correlation to player or team success. That's the issue - you can select any set of data to support your preconception, but if he was so bad, why did he get 26 minutes per night in the playoffs and be considered by his team and guys like Pierre Lebrun to be the toughest member of the Pathers to play against?

 

Made a new thread that digs a little deeper into the Hero charts bs

[Alflives]

2 hours ago, Bob Long said:

 

But where is the evidence that the information is being used scientifically by hockey GMs? The article I posted above is a perfect example of the author picking a limited data set to prove his point, vs. looking at a much broader set and letting the data tell you what the actual correlation is. I don't see the analytics guys doing that yet. Now that may be exactly whats going on behind the scenes and teams might be hanging on to that info and not making it public, but in terms of the "analysis" of the GudBranson trade nothing I saw supports the kind of comments made like "lopsided". 

 

The blocked shots example is an interesting one, and a hand-picked stat view again is potentially misleading. Are the stats skewed because crappier teams block shots more and don't have the puck as much? Does that mean a good possession team should never block shots, or assemble a team of guys unwilling to block shots because they had good possession numbers? Of course not, there has to be a good balance. 

 

 

 

 

 

Look.  They do have their place.  But they are not science.  The data used is flawed - it's collected from the home teams' stats, which are often EXTREMELY biassed. Crap in = crab out.  Yet, as I stated: they have a purpose.  A small one.