The analytical Era has started: Are the Canucks ahead or Behind the curve?

[Ernie]

Personally, I think organizations should focus more on hiring graduates from STEM fields who have a good knowledge on sports to analyze data for trends. I would argue that hiring graduated graduate students from a statistics, economics, or physics department would be much more effective than hiring people like Cam Charron who are familiar with the game but don't really know how to use statistics, model, or know how to actually draw valid conclusions and assess their validity.

Indeed. Get some super nerds and pair them with hockey guys. In some ways a lot of hockey knowledge could be a negative - there are way to many learned biases. In addition, find some guys who can write code!

When the league starts tracking ice movements, the Charrons and Dellowses aren't going to be able to keep up unless they double down on the nerd.

 

[Orca Smash]

Noticed this in todays article.

ANALYTICS ARE IN

Linden was asked about the role of analytics in the team’s approach this season, in light of a number of NHL clubs adding formal departments in this area.

“We’re using a company that has specific stats that are related to us. They are more detailed. It’s an emerging philosophy out there. We’re looking forward to using it in both pre- and post-game analysis.”

http://www.theprovince.com/sports/Canucks+coach+plans+keep+Sedins+fresh+with/10189140/story.html

 

[Wilch]

Noticed this in todays article.

ANALYTICS ARE IN

Linden was asked about the role of analytics in the team’s approach this season, in light of a number of NHL clubs adding formal departments in this area.

“We’re using a company that has specific stats that are related to us. They are more detailed. It’s an emerging philosophy out there. We’re looking forward to using it in both pre- and post-game analysis.â€

http://www.theprovince.com/sports/Canucks+coach+plans+keep+Sedins+fresh+with/10189140/story.html

Nice to know this stuff can actually be outsourced.

 

[biturbo19]

Nice to know this stuff can actually be outsourced.

But what if there are dirty filthy Flames spies who infiltrate our subcontractors and learn all of our most precious advanced metric secrets!!!???

 

[Wilch]

But what if there are dirty filthy Flames spies who infiltrate our subcontractors and learn all of our most precious advanced metric secrets!!!???

Meh.

Internal analysts can jump ship to other organizations too.

Then again hockey seems to be a pretty closed off world, so who knows.

 

[biturbo19]

Meh.

Internal analysts can jump ship to other organizations too.

Then again hockey seems to be a pretty closed off world, so who knows.

I was being facetious.

If these "advanced stats" are anywhere near as "objective" as they're portrayed as...the numbers are out there widely available to anyone and everyone who can be bothered to spend a few hours grinding out the data.

And frankly, i couldn't care less if someone got a hold of the Canucks proprietary "advanced stats" data. I don't really know how much of a great insight they impart to the process in the first place. They're a nice side dish, sometimes reveal some idiosyncrasies that warrant further investigation or changes, or back up the things you're seeing on the ice. But the entree is still, and imo always will be good ol' fashioned "eyeball scouting".

 

[Wilch]

I was being facetious.

If these "advanced stats" are anywhere near as "objective" as they're portrayed as...the numbers are out there widely available to anyone and everyone who can be bothered to spend a few hours grinding out the data.

And frankly, i couldn't care less if someone got a hold of the Canucks proprietary "advanced stats" data. I don't really know how much of a great insight they impart to the process in the first place. They're a nice side dish, sometimes reveal some idiosyncrasies that warrant further investigation or changes, or back up the things you're seeing on the ice. But the entree is still, and imo always will be good ol' fashioned "eyeball scouting".

Advanced stats aren't objective; algorithms, metrics, weighting, etc. vary from analyst to analyst.

On the other hand, the raw data is. Those are tough to accurately acquire, but it's the foundation to doing good statistical work.

And yes, seeing the players play is important, that's why it's important to have hard working scouts who have a good eye for the game and actually care, instead of some ******* like Delorme who doesn't even bother showing up to the prospect scrimmages until there's a couple minutes left to go.

Since qualitative assessments will always be disputed due to difference of opinion, advanced stats is a decent alternative in some cases to break a deadlock.

 

[DL44]

I think that advanced stats was one of Gillis' greatest contributions to this club. We are ahead of the curve overall.

Bo Horvat is a prime example of this IMO.

 

[Wilch]

I think that advanced stats was one of Gillis' greatest contributions to this club. We are ahead of the curve overall.

Bo Horvat is a prime example of this IMO.

Did Gillis directly attribute the selection of Horvat on advanced stats?

 

[DL44]

Did Gillis directly attribute the selection of Horvat on advanced stats?

I.e. he was drafted above consensus and isn't a remarable offensive juggernought.

The Canucks jumped on him for other reasons... Not much of a jump to assume it was analytics.

But your skepticism stands.. No proof otherwise.

But it is seems to be common custom that the GM does control the 1st pick... Tradition. scouts with the rest..

They'll even give say the European scout the 3rd pick... Etc..

 

[biturbo19]

Advanced stats aren't objective; algorithms, metrics, weighting, etc. vary from analyst to analyst.

On the other hand, the raw data is. Those are tough to accurately acquire, but it's the foundation to doing good statistical work.

And yes, seeing the players play is important, that's why it's important to have hard working scouts who have a good eye for the game and actually care, instead of some ******* like Delorme who doesn't even bother showing up to the prospect scrimmages until there's a couple minutes left to go.

Since qualitative assessments will always be disputed due to difference of opinion, advanced stats is a decent alternative in some cases to break a deadlock.

Sure, you can come up with all of the algorithms and combinations you want. But it hardly seems there are too many guys coming up with completely revolutionary ideas in that regard...at least, not revolutionary ideas that have any sort of reliable results projection.

By and large, it's teams looking at a lot of the same sort of things. But you're right...even the "raw data" is to a degree "subjective", which is why it's so strange to me when people 'round here tote an "advanced stat" out as an inalienable fact impervious to dispute and essentially, "the gospel".

And you're also right in that the unfortunate truth is, our "eyeball scouting" has been pretty lackluster anyway in recent years. Especially at the Pro Level where Analytics tend to be at least a bit more valuable and reliable. But i'm not sure that's an endorsement of "advanced stats" so much as just an indictment of our "scouting" in general.

 

[Wilch]

Sure, you can come up with all of the algorithms and combinations you want. But it hardly seems there are too many guys coming up with completely revolutionary ideas in that regard...at least, not revolutionary ideas that have any sort of reliable results projection.

By and large, it's teams looking at a lot of the same sort of things. But you're right...even the "raw data" is to a degree "subjective", which is why it's so strange to me when people 'round here tote an "advanced stat" out as an inalienable fact impervious to dispute and essentially, "the gospel".

And you're also right in that the unfortunate truth is, our "eyeball scouting" has been pretty lackluster anyway in recent years. Especially at the Pro Level where Analytics tend to be at least a bit more valuable and reliable. But i'm not sure that's an endorsement of "advanced stats" so much as just an indictment of our "scouting" in general.

We don't know. Teams with success using statistical analysis are not publish these information.

I don't believe teams should rely solely on statistics to assess players, but it should be something that allow scouts to quickly filter out the really crappy players and put red flags on highly rated players with potential problems.

For online discussions, statistics aren't gospel, but it's certainly not far away from truth is employed correctly. When certain arguments are so far away from the truth and posters are unwilling to settle on subjective assessments, it makes sense to use appropriate statistics to support one's argument.

 

[biturbo19]

We don't know. Teams with success using statistical analysis are not publish these information.

I don't believe teams should rely solely on statistics to assess players, but it should be something that allow scouts to quickly filter out the really crappy players and put red flags on highly rated players with potential problems.

For online discussions, statistics aren't gospel, but it's certainly not far away from truth is employed correctly. When certain arguments are so far away from the truth and posters are unwilling to settle on subjective assessments, it makes sense to use appropriate statistics to support one's argument.

We have a pretty solid idea what this new wave of "analytics gurus" are doing with their data though. They haven't been shy in sharing their stuff to get attention. Unless they're inventing radical new algorithms or whatever after being hired, which seems unlikely.

 

[me2]

We have a pretty solid idea what this new wave of "analytics gurus" are doing with their data though. They haven't been shy in sharing their stuff to get attention. Unless they're inventing radical new algorithms or whatever after being hired, which seems unlikely.

We don't have that much of an idea. Fans tend to run of primitive stats like Corsi etc because the data they get is very basic. Almost all of those new wave guys we keep hearing from that got hired were using basic data to derive meaning. I can only imagine what they will be attempting to create when they are given the opportunity to create they own data sets rather than rely on things like shots on goal.

 

[Wilch]

We have a pretty solid idea what this new wave of "analytics gurus" are doing with their data though. They haven't been shy in sharing their stuff to get attention. Unless they're inventing radical new algorithms or whatever after being hired, which seems unlikely.

This is not a direct comparison, but advanced aerospace firms don't file patents on some of their most high end designs. Once they do, it goes public and anyone within its jurisdiction can try and reverse engineer it and skirt patent laws by creating something similar but different. Or most importantly, entities not within its jurisdiction not bound by IP laws such as China can simply make an exact copy and call it their own.

Another example would be Robert Hooke taking Isaac Newton's research and calling it his own.

It's the same with analysts who designed successful and sophisticated algorithms. There's no effective way to enforce a copyright on the usage of an algorithm. As such, there's not a bloody chance anyone with half a brain will disclose that information to the public unless they're legally or contractually required to do so.

These so called gurus disclosing their methodology are mostly amateurs or publish incomplete/flawed algorithms.

 

[biturbo19]

This is not a direct comparison, but advanced aerospace firms don't file patents on some of their most high end designs. Once they do, it goes public and anyone within its jurisdiction can try and reverse engineer it and skirt patent laws by creating something similar but different. Or most importantly, entities not within its jurisdiction not bound by IP laws such as China can simply make an exact copy and call it their own.

Another example would be Robert Hooke taking Isaac Newton's research and calling it his own.

It's the same with analysts who designed successful and sophisticated algorithms. There's no effective way to enforce a copyright on the usage of an algorithm. As such, there's not a bloody chance anyone with half a brain will disclose that information to the public unless they're legally or contractually required to do so.

These so called gurus disclosing their methodology are mostly amateurs or publish incomplete/flawed algorithms.

Sure. But a lot of these guys getting snapped up have talked at some length about the sort of things they're trying to measure with their algorithms, posted results, some even made a living sharing a big chunk of their information via a website dedicated to the stuff. Obviously there are still some "secrets" there, but is it really that hard to reverse engineer what they were doing if someone is so inclined?

As per your analogy, it's hockey analytics, not rocket science.

 

[Wilch]

Sure. But a lot of these guys getting snapped up have talked at some length about the sort of things they're trying to measure with their algorithms, posted results, some even made a living sharing a big chunk of their information via a website dedicated to the stuff. Obviously there are still some "secrets" there, but is it really that hard to reverse engineer what they were doing if someone is so inclined?

As per your analogy, it's hockey analytics, not rocket science.

Like I said, they disclosed incomplete information.

The guy who ran extraskater provided good information, but it wasn't exactly ground breaking.

Whatever comprehensive analysis he'll be doing with his new gig will probably never see the light of day.

 

[biturbo19]

Like I said, they disclosed incomplete information.

The guy who ran extraskater provided good information, but it wasn't exactly ground breaking.

Whatever comprehensive analysis he'll be doing with his new gig will probably never see the light of day.

Possible. Maybe even probable. The equations are bound to become more and more complicated and intricate...

Though my skepticism remains as to what these "super secret hidden algorithms" might illuminate that is of critical importance to NHL teams.

This still isn't baseball.

 

[Kickassguy]

Honestly, a lot of stats analysis companies in hockey do pretty shi*ty work. (PUCKS, anyone?)

I'd feel better actually knowing what their insights are grounded in and that they actually understand the implications, rather than just deferring to "Oh, we outsource this."

 

[me2]

http://fivethirtyeight.com/features/the-people-pushing-the-nhl-into-the-advanced-stats-era/

 

[Wilch]

http://fivethirtyeight.com/features/the-people-pushing-the-nhl-into-the-advanced-stats-era/

Excellent find.