Speculation: Oilers hire Tyler Dellow

[Everest]

It's certainly a pretty young discipline (and even if it wasn't, that wouldn't stop people from trying to improve it), and there's always room for improvement. The argument, which is sound, is that Corsi and Fenwick are the best available predictors of playoff success in the long term, which I have yet to see effectively refuted.

The argument is not that either measure is anywhere near perfect, however; merely that we don't have anything that is a better predictor of success right now.

But...that's pointing out the obvious as far as an NHL coach is concerned.

Any brainiac already knows Corsi & Fenwick are revealing evidence of strong play.

Coaches need to know how to GET to the point where they've got a team producing good results under those columns.

And that's when advanced stats become useless.

The numbers are a (growing) collection of whats already happened. For sure.

The prediction those numbers give us...is... well...predictable.

I remember an Eakins quote in regards to Tyler Dellow; where he talked about Dellows' "passion for the team to improve."

This, folks, was a lie.

Tyler Dellow wants to show his readership how smart he is. And, really, he already achieved this...within his field...a long time ago.

Thing is...by actually agreeing to work for an NHL team...he effectively reversed this confirmation.

If Tyler Dellow was truly smart...he would have stayed doing what he was good at. Eventually...a proper use for his work would have been ascertained, Im sure.

As it is...he's affirmed just how flawed his philosophy must be. He doesn't know where his numbers belong. He doesn't understand his own plot!~

 

[Replacement*]

But...that's pointing out the obvious as far as an NHL coach is concerned.

Any brainiac already knows Corsi & Fenwick are revealing evidence of strong play.

Coaches need to know how to GET to the point where they've got a team producing good results under those columns.

And that's when advanced stats become useless.

The numbers are a (growing) collection of whats already happened. For sure.

The prediction those numbers give us...is... well...predictable.

I remember an Eakins quote in regards to Tyler Dellow; where he talked about Dellows' "passion for the team to improve."

This, folks, was a lie.

Tyler Dellow wants to show his readership how smart he is. And, really, he already achieved this...within his field...a long time ago.

Thing is...by actually agreeing to work for an NHL team...he effectively reversed this confirmation.

If Tyler Dellow was truly smart...he would have stayed doing what he was good at. Eventually...a proper use for his work would have been ascertained, Im sure.

As it is...he's affirmed just how flawed his philosophy must be. He doesn't know where his numbers belong. He doesn't understand his own plot!~

He jumped at a chance to take a ride on the titanic with bravado feeling he would alter fate..

I question his smarts for that as well. Plus what does this guy actually do besides work numbers all day and play a weak hire position for a hockey club..They probably pay him less that Equipment staff get for a part time gig.

Him and Eakins are bookends of pumped up ego feeling they have the secret keys to the kingdom of hockey. Both perennially felt they were more astute students of the game than the top hockey minds. Both felt they would rewrite how pro hockey is played. Such avarice. Always on display too. Eakins and Dellow both loved their own thoughts. Always thought their versions were infallible. Well here we are in the land of worst ever NHL team.

Anyway wasn't Dellow supposed to be a lawyer or something? He can't get any better job then pretend hockey advisor?

Something always struck me a little strange about all that.

Perhaps the bolded is even more telling.

Rainbow chaser?

 

[NikF]

You need a certain kind of person for this to work. A lot of the "champions" of the advanced stats community are very combative and extremely fundamentalist in their approach, while that certainly aided in shaping the stats community, I'm not sure it serves anyone in an NHL team context.

You aren't there to prove you're right and they are wrong (an approach that was needed for the advanced stats community to shape itself in the first place and get their place under the hockey sun), you're there to improve the team. Different mindsets. One that goes in with the presumption that he will change how the game is thought or prove that his way is better than the other way won't accomplish much. A lot of advanced stats figureheads have that kind of approach, which is good in a blogosphere setting and to gain readership and even lends itself to further statistical analysis, because if progress is to be made you do need a bit of that fundamentalism, but it also really tends to show some pretty big holes in their thinking within the context of hockey as a whole because they rigidly stick to the stats dogma even in situations where doing so is laughable at best. And in those situations they bend the reality to fit their beliefs instead of bending the beliefs to fit the reality.

 

[doulos]

For a guy who doesn't know what he's talking about or what he's doing (according to many here), Dellows got himself employed by an NHL team (even if it is a poor excuse for one). That's better than the rest of the crew here who thinks they have it all figured out.

The advanced stats folks who think it's 'the' solution for cracking the hockey code (and I can't say I know any of them, but that's sort of the straw man being argued against) are wrong.

There are a whole lot of 'saw him good' folks who look silly every time they discuss this topic as well.

A little bit of balance on this issue is needed but it seems many people are too emotionally invested in being right about the issue and the actual chance for dialogue suffers because of it.

 

[LeafsFIO*]

really glad for the positivity of this move from an Oilers standpoint, good to know they'll be evaluating properly moving forward.

If I'm an old-guard guy in the NHL I do exactly what the Oilers did. Hire a top-notch analydiot, pick his brain for a few months then dump him by the wayside. Happy for the future in Edmonton

 

[doulos]

really glad for the positivity of this move from an Oilers standpoint, good to know they'll be evaluating properly moving forward.

If I'm an old-guard guy in the NHL I do exactly what the Oilers did. Hire a top-notch analydiot, pick his brain for a few months then dump him by the wayside. Happy for the future in Edmonton

If you don't believe in the value of his work then why employ him at all?

 

[Slats432]

I find it funny how too many people are either pro advanced stats or against them.

I also find it funny how people can easily discount Dellow's contribution without knowing what his contribution was. (Basically people are saying he is a failure based on what people on HF have said about what they think the contributions were?)

Personally, I don't like how condescending Dellow can be. That doesn't mean his data doesn't have value.

I was at a golf tournament with a guy that owns a company whose product is advanced hockey stats. Bright guy.

He stated that advanced hockey stats digs way deeper than what the Corsis and Fenwicks do. There are 45 staff that go over each second of film in a game charting everything from what we hear about to things like skating speed, pass speed, etc etc.

Chicago is a big proponent of advanced stats. http://www.cbssports.com/nhl/eye-on...m-stan-bowman-says-analytics-give-team-leg-up

So, you get tons of different data. You sift through and try to determine what it means and try to implement things. Just because it doesn't work, doesn't mean the data is bad. It could mean the interpretation and implementation of data based strategy is bad.

If I have a stat that says Ales Hemsky loses the puck within 3 feet of the opposing blueline 48% of the time, I might coach him to do several different things. (Dump in, pass over the blueline, carry if he is not the first player in the zone)

What if none of what I coach him to do works any better? Is the stat terrible? Nope.

Let's remember to take stats for what they are. Information. Or let's at least do it 65.3% of the time.

 

[Arpeggio]

I find it funny how too many people are either pro advanced stats or against them.

I also find it funny how people can easily discount Dellow's contribution without knowing what his contribution was. (Basically people are saying he is a failure based on what people on HF have said about what they think the contributions were?)

Personally, I don't like how condescending Dellow can be. That doesn't mean his data doesn't have value.

I was at a golf tournament with a guy that owns a company whose product is advanced hockey stats. Bright guy.

He stated that advanced hockey stats digs way deeper than what the Corsis and Fenwicks do. There are 45 staff that go over each second of film in a game charting everything from what we hear about to things like skating speed, pass speed, etc etc.

Chicago is a big proponent of advanced stats. http://www.cbssports.com/nhl/eye-on...m-stan-bowman-says-analytics-give-team-leg-up

So, you get tons of different data. You sift through and try to determine what it means and try to implement things. Just because it doesn't work, doesn't mean the data is bad. It could mean the interpretation and implementation of data based strategy is bad.

If I have a stat that says Ales Hemsky loses the puck within 3 feet of the opposing blueline 48% of the time, I might coach him to do several different things. (Dump in, pass over the blueline, carry if he is not the first player in the zone)

What if none of what I coach him to do works any better? Is the stat terrible? Nope.

Let's remember to take stats for what they are. Information. Or let's at least do it 65.3% of the time.

This. As far as I'm aware the real "advanced" stats involve player tracking. Corsi and Fenwick are really just an improved measurement of shots. Which is still valuable, as is obvious by the fact that the best teams still tend to have better Corsi and Fenwick numbers than bad teams.

Just because the Oilers suck balls doesn't mean that advanced stats aren't valuable, that's stupid.

 

[Replacement*]

For a guy who doesn't know what he's talking about or what he's doing (according to many here), Dellows got himself employed by an NHL team (even if it is a poor excuse for one). That's better than the rest of the crew here who thinks they have it all figured out.

The advanced stats folks who think it's 'the' solution for cracking the hockey code (and I can't say I know any of them, but that's sort of the straw man being argued against) are wrong.

There are a whole lot of 'saw him good' folks who look silly every time they discuss this topic as well.

A little bit of balance on this issue is needed but it seems many people are too emotionally invested in being right about the issue and the actual chance for dialogue suffers because of it.

In fairness you missed the years of avarice, arrogance and what got communicated on this board. Suffice to say that some of these individuals would come on board and basically heckle the forum here. This was their behavior. Dellow often included in that albeit he was tame compared to the others. heres something else. A very sarcastic advanced stats member would have as his M.O to flame this board and go just ballistic in a thread he created and would then delete the thread after 50 or so abusive posts thereby removing the trace. Same poster would come back the next day and refuse to acknowledge any wrong doing. This poster later admitted last year that he used several online aliases and including here which is a clear contravention of site rules. He would come in here and people used to challenge why it seemed like 5 henchmen of the advanced apocalypse were always coming in here at once. Well it was all the same guy and in retrospect contemptible board behavior.

just for some background on why many of the posters that have been here for a longwhile have their backs up at some of this stuff.

Its long since been time to get over it, sure, but sometimes the background is perhaps needed when people wonder why their is apparent polarization noted.

What went on here also went on on any blogger site. The raids would happen there too. It was almost an abusive evangelical approach.

 

[guymez]

I find it funny how too many people are either pro advanced stats or against them.

I also find it funny how people can easily discount Dellow's contribution without knowing what his contribution was. (Basically people are saying he is a failure based on what people on HF have said about what they think the contributions were?)

Personally, I don't like how condescending Dellow can be. That doesn't mean his data doesn't have value.

I was at a golf tournament with a guy that owns a company whose product is advanced hockey stats. Bright guy.

He stated that advanced hockey stats digs way deeper than what the Corsis and Fenwicks do. There are 45 staff that go over each second of film in a game charting everything from what we hear about to things like skating speed, pass speed, etc etc.

Chicago is a big proponent of advanced stats. http://www.cbssports.com/nhl/eye-on...m-stan-bowman-says-analytics-give-team-leg-up

So, you get tons of different data. You sift through and try to determine what it means and try to implement things. Just because it doesn't work, doesn't mean the data is bad. It could mean the interpretation and implementation of data based strategy is bad.

If I have a stat that says Ales Hemsky loses the puck within 3 feet of the opposing blueline 48% of the time, I might coach him to do several different things. (Dump in, pass over the blueline, carry if he is not the first player in the zone)

What if none of what I coach him to do works any better? Is the stat terrible? Nope.

Let's remember to take stats for what they are. Information. Or let's at least do it 65.3% of the time.

Completely agree.

Since when did prejudice against information work to a teams advantage? According to many posters in this thread omission of more detailed information is superior to inclusion.

More information is never the problem...incorrect application and/or interpretation is the issue.

 

[Replacement*]

I find it funny how too many people are either pro advanced stats or against them.

I also find it funny how people can easily discount Dellow's contribution without knowing what his contribution was. (Basically people are saying he is a failure based on what people on HF have said about what they think the contributions were?)

Personally, I don't like how condescending Dellow can be. That doesn't mean his data doesn't have value.

I was at a golf tournament with a guy that owns a company whose product is advanced hockey stats. Bright guy.

He stated that advanced hockey stats digs way deeper than what the Corsis and Fenwicks do. There are 45 staff that go over each second of film in a game charting everything from what we hear about to things like skating speed, pass speed, etc etc.

Chicago is a big proponent of advanced stats. http://www.cbssports.com/nhl/eye-on...m-stan-bowman-says-analytics-give-team-leg-up

So, you get tons of different data. You sift through and try to determine what it means and try to implement things. Just because it doesn't work, doesn't mean the data is bad. It could mean the interpretation and implementation of data based strategy is bad.

If I have a stat that says Ales Hemsky loses the puck within 3 feet of the opposing blueline 48% of the time, I might coach him to do several different things. (Dump in, pass over the blueline, carry if he is not the first player in the zone)

What if none of what I coach him to do works any better? Is the stat terrible? Nope.

Let's remember to take stats for what they are. Information. Or let's at least do it 65.3% of the time.

Just replying but I think you would know I've followed Dellows contributions as long as he's been around posting them. I've had countless discussions with Dellow. Dellow was the easiest to talk to of the advanced stats bloggers imo. He would at least enter into point counterpoint discussion. I don't discount for a second that Dellow had interesting ideas and that he spent a lot of time on exploring hockey minutiae.

But heres where it fell apart for him and that Everest alluded to. He did think he had a better way in hockey and did express countless contempt for what he saw NHL coaches, managers, players do. He stridently emphasized certain players over others. mocked deals, trades etc, championed certain players that weren't always that great, advocated particular schemes, break out schemes, face off sets etc.

Reading his stuff one always got the impression that he was doing a lot of armchair criticism. As if he knew better. A lot of us do similar, I do, but I would never pretend, or say that I should be in any related employ, or accept any and I would of course find it laughable if anybody ever offered me a hockey related job (except maybe writing about it haha) but one gets the feeling with Dellows and others in the advanced stats community that they really are convinced they know better than veteran and accomplished hockey minds that are actually in that employ and have been for decades. Indeed the pair bond between MacT/Eakins, and Eakins/Dellow involves a similar self conceit that is quite apparent and often voiced. With Eakins a lot of his voiced approaches being basically unheard of in the NHL but which he stubbornly continued with to no success. Indeed even at the end of the sorry Eakins coaching Saga MacT is up there stating that Eakins is still a great coach as if abject failure had never occurred.

Also its interesting (perhaps lies) but people that knew Dellows and communicate with Dellows made statements to the effect that Dellows was working with Eakins on such things as breakout schemes, face off sets, even pk sets. The most interesting thing about the disclosure is that these were things that Dellow discussed a lot. So I almost wonder whether some of Eakins hackneyed ideas were in consulation with Dellows. Plus that Eakins was citing his work quite often and that they were breaking it down and using it.

Anyway it didn't work out so well.

Anyway just my take.

 

[Everest]

To be fair...depending on what were filing under the definition of 'advanced' there are already some advanced stats in use. Thing is...these aren't items which require a specialist to detail or explain.

Its the more complicated attempts to numerate & itemize every "event" in the course of a game that will never be truly useful for the purpose of coaching.

Interesting & useful in some other dimension? Sure. Just not coaching.

And like I said before...its really ironic some advanced stats proponents are finding their hands tied these days...unable to provide hard evidence of advanced stats altering how a team was coached. Either for better or worse.

I don't have the time to sit here...day after day...hour after hour...explaining it...but...ultimately...any NHL coaching staff needs to be able to accomplish what advanced stats are essentially in pursuit of. But in a fraction of the time.

U gotta coach by feel. U gotta know the truths of the game without seeing it in a number.

 

[Slats432]

To be fair...depending on what were filing under the definition of 'advanced' there are already some advanced stats in use. Thing is...these aren't items which require a specialist to detail or explain.

Its the more complicated attempts to numerate & itemize every "event" in the course of a game that will never be truly useful for the purpose of coaching.

Interesting & useful in some other dimension? Sure. Just not coaching.

And like I said before...its really ironic some advanced stats proponents are finding their hands tied these days...unable to provide hard evidence of advanced stats altering how a team was coached. Either for better or worse.

I don't have the time to sit here...day after day...hour after hour...explaining it...but...ultimately...any NHL coaching staff needs to be able to accomplish what advanced stats are essentially in pursuit of. But in a fraction of the time.

U gotta coach by feel. U gotta know the truths of the game without seeing it in a number.

I would have to say that it can be implemented in coaching. For example....if I have a stats guy saying that when Ference is playing left side on the road against Anaheim he gets beaten wide on average 3 times a game, that he is coached to try to prevent this from happening.

What if the RNH line gets more shots on goal when they dump and chase vs. carry? What if Roy at home wins 65% of his draws in his own zone against lefties on the right dot? And Gordon is 62% on right dot and 67% on left dot?

Put Roy on right dot for a left face off? You bet. Gordon on the other side? Guaranteed.

But if Roy loses the draw you are an idiot because Gordon has the best FO % overall.

They can be used, but which ones, and when? That's for the stats guys and coaches to decide.

 

[Replacement*]

I would have to say that it can be implemented in coaching. For example....if I have a stats guy saying that when Ference is playing left side on the road against Anaheim he gets beaten wide on average 3 times a game, that he is coached to try to prevent this from happening.

What if the RNH line gets more shots on goal when they dump and chase vs. carry? What if Roy at home wins 65% of his draws in his own zone against lefties on the right dot? And Gordon is 62% on right dot and 67% on left dot?

Put Roy on right dot for a left face off? You bet. Gordon on the other side? Guaranteed.

But if Roy loses the draw you are an idiot because Gordon has the best FO % overall.

They can be used, but which ones, and when? That's for the stats guys and coaches to decide.

The bolded is a good example of where things go sideways and I followed the the dialog on this closely. The stats in this were being used to indicate that player X Y Z SHOULD carry the puck in more and that its silly that the player is shooting the puck in when they do. This is the inference I got when reading that. That player X Y Z is much more effective carrying the puck in.

But this also ignores several other fundamentals;

1)That players and lines have vastly different skillsets in carrying puck in and that are NOT a fixed quantity but vary through time as player, and line confidence varies. This being the human element.

2)The stats failed to cite the latter consequence of what occurs when players always try to carry the puck in. i.e. turnovers and GA. Success isn't a zero sum game of whether team got a shot on goal because they carried the puck in. If the other team is getting goals off the same tendency it may not be an optimal approach

3)It ignores situational specifics, schemes NZ, blueline schemes and who you are playing against and who is checking you on any given night. To this end telling player X Y Z to increase amount of time they carry puck in is at best white noise which is differentially effective related to myriad opponent variables. Which are not broken down in the analysis. For instance no numbers are ever run citing that player X should definitely carry in if they are playing with players Y Z but not carry on if they are playing with players A B, and not carry in if they are playing against the Detroit Red Wings..
In sum the advice is never broken down like that and if it was the dataset would be too small anyway to be of clear value. But as stated it is of nebulous value that only creates confusion in player and team concept. Which leads to my next point.

4)LA KINGS/NYI/Sutter tm hockey. ALWAYS get the puck deep. Get the puck deep if it costs you 10 stitches. GEt the puck deep like your life, and the team depends on it (which they do) NEVER turn the puck over at the blueline trying to carry it in when its not there. When in doubt dump, work a chip on the boards and chase puck in, or pass back and get rush set over.

Sutter will basically **** down the throat of any player that doesn't get the puck deep with consistency. Because the whole machine breaks down there.

 

[Slats432]

The bolded is a good example of where things go sideways and I followed the the dialog on this closely. The stats in this were being used to indicate that player X Y Z SHOULD carry the puck in more and that its silly that the player is shooting the puck in when they do. This is the inference I got when reading that. That player X Y Z is much more effective carrying the puck in.

But this also ignores several other fundamentals;

1)That players and lines have vastly different skillsets in carrying puck in

2)The stats failed to cite the latter consequence of what occurs when players always try to carry the puck in. i.e. turnovers and GA. Success isn't a zero sum game of whether team got a shot on goal because they carried the puck in. If the other team is getting goals off the same tendency it may not be an optimal approach

3)It ignores situational specifics, schemes NZ, blueline schemes and who you are playing against and who is checking you on any given night. To this end telling player X Y Z to increase amount of time they carry puck in is at best white noise which is differentially effective related to myriad opponent variables. Which are not broken down in the analysis. For instance no numbers are ever run citing that player X should definitely carry in if they are playing with players Y Z but not carry on if they are playing with players A B, and not carry in if they are playing against the Detroit Red Wings..
In sum the advice is never broken down like that and if it was the dataset would be too small anyway to be of clear value. But as stated it is of nebulous value that only creates confusion in player and team concept. Which leads to my next point.

4)LA KINGS/NYI/Sutter tm hockey. ALWAYS get the puck deep. Get the puck deep if it costs you 10 stitches. GEt the puck deep like your life, and the team depends on it (which they do) NEVER turn the puck over at the blueline trying to carry it in when its not there. When in doubt dump, work a chip on the boards and chase puck in, or pass back and get rush set over.

Sutter will basically **** down the throat of any player that doesn't get the puck deep with consistency. Because the whole machine breaks down there.

I am using examples in a general sense. I don't expect to debate my examples. I am saying that advanced stats can direct coaching. (And the stat could be as detailed as (When RNH line is on ice second half of shift and RNH has the puck on the left side of the ice, the play is generally to carry the puck in.).

IMO advanced stats can provide coaching moments.

 

[Replacement*]

Next, some disclosure here. I've been a Chess player since I was 4yrs old. Played competitive chess, speed chess, and against Computer chess programs since they were first mass produced. Played many chess engines on line etc.

I'll mention that until recently chess computers, which have existed in some form since the earliest computational systems existed, have often been limited. Lets understand as well that the top minds in the world program, make submissions to, and help to write computer chess programs. It took decades before Chess computer programs were even competitive against reasonably good chess players. Until say the 80's I was beating Chess programs at any selected level of play. You would note several algorithms that were wrong and exploit them as much as you wanted to. Getting bored you would find and exploit other errors in the programming.

I only mention this because one of the basic problems in chess programming is how many moves in advance a program could detect and play. As such programs were susceptible to such things as disguised pins, traps, sacrifices for positional gain, and pitfalls. With all of these having to be programmed in to programs to make them somewhat better. But with instances of real play approaching infinity. Until recent decades programmers had to work with computing limitations that involved processing capacity, speed, memory etc. So even though Grandmasters were involved in programming the intelligence of the programs was limited due to computer limitation.

Next hockey. Infinite possibilities, infinite variables, fastest game on earth. Infinite sequences and possible results on ANY shift which can't be predicted.

Yet I'm inundated with advanced stats that every day of the week cite such things as player X should be carrying the puck in more instead of dumping and that's it. That's the end all conclusion.

Well friends that a ONE MOVE analysis that ignores ANY subsequent result on shifts that can last over a minute and contiguous play and sequence that can be several minutes at a time.

Hockey is not chess, but is certainly not in anyway a one move, one analysis at a time game where it makes sense to make simplistic statistical observations such as player X should carry puck more, shoot puck more, whatever, because it results in one variable (More shots on net) while ignoring every possible resultant that follows.

A chess algorithym that explored only one move recommendation would be a waste of anybodies time.

 

[Replacement*]

I am using examples in a general sense. I don't expect to debate my examples. I am saying that advanced stats can direct coaching. (And the stat could be as detailed as (When RNH line is on ice second half of shift and RNH has the puck on the left side of the ice, the play is generally to carry the puck in.).

IMO advanced stats can provide coaching moments.

I wasn't requiring you to debate it. I trust that you took the time to read it.

Clearly what I'm indicating, and took the time to carefully lay out is that in one of the most often cited "sensible uses of advanced Stats" that what is being advocated isn't really worth all that much to professional veteran hockey coaches coaching at the highest level.

This should certainly not provide "direction" for such skilled coaches. At best it might provide some real time information for use only in real time one specific instance in one specific game. But that info is areadly forthcoming in the form of spotters that might indicate player X is biting at the blueline and is susceptible to being beat with a rush along the boards. for example.

I'm just a peon here breaking down the information entirely and citing the many drawbacks of such simplistic one move analysis and recommendation. Which really doesn't give advanced coaching much that it doesn't already have at its disposal.

 

[oilerbear]

He jumped at a chance to take a ride on the titanic with bravado feeling he would alter fate..

I question his smarts for that as well. Plus what does this guy actually do besides work numbers all day and play a weak hire position for a hockey club..They probably pay him less that Equipment staff get for a part time gig.

Him and Eakins are bookends of pumped up ego feeling they have the secret keys to the kingdom of hockey. Both perennially felt they were more astute students of the game than the top hockey minds. Both felt they would rewrite how pro hockey is played. Such avarice. Always on display too. Eakins and Dellow both loved their own thoughts. Always thought their versions were infallible. Well here we are in the land of worst ever NHL team.

Anyway wasn't Dellow supposed to be a lawyer or something? He can't get any better job then pretend hockey advisor?

Something always struck me a little strange about all that.

Perhaps the bolded is even more telling.

Rainbow chaser?

I have no doubt you would!

 

[oil slick]

Next, some disclosure here. I've been a Chess player since I was 4yrs old. Played competitive chess, speed chess, and against Computer chess programs since they were first mass produced. Played many chess engines on line etc.

I'll mention that until recently chess computers, which have existed in some form since the earliest computational systems existed, have often been limited. Lets understand as well that the top minds in the world program, make submissions to, and help to write computer chess programs. It took decades before Chess computer programs were even competitive against reasonably good chess players. Until say the 80's I was beating Chess programs at any selected level of play. You would note several algorithms that were wrong and exploit them as much as you wanted to. Getting bored you would find and exploit other errors in the programming.

I only mention this because one of the basic problems in chess programming is how many moves in advance a program could detect and play. As such programs were susceptible to such things as disguised pins, traps, sacrifices for positional gain, and pitfalls. With all of these having to be programmed in to programs to make them somewhat better. But with instances of real play approaching infinity. Until recent decades programmers had to work with computing limitations that involved processing capacity, speed, memory etc. So even though Grandmasters were involved in programming the intelligence of the programs was limited due to computer limitation.

Next hockey. Infinite possibilities, infinite variables, fastest game on earth. Infinite sequences and possible results on ANY shift which can't be predicted.

Yet I'm inundated with advanced stats that every day of the week cite such things as player X should be carrying the puck in more instead of dumping and that's it. That's the end all conclusion.

Well friends that a ONE MOVE analysis that ignores ANY subsequent result on shifts that can last over a minute and contiguous play and sequence that can be several minutes at a time.

Hockey is not chess, but is certainly not in anyway a one move, one analysis at a time game where it makes sense to make simplistic statistical observations such as player X should carry puck more, shoot puck more, whatever, because it results in one variable (More shots on net) while ignoring every possible resultant that follows.

A chess algorithym that explored only one move recommendation would be a waste of anybodies time.

I like the chess analogy. But to follow through on the analogy, you now have mobile phones that are grandmasters. And they are able to play at this level simply because with smart programming, a computer can take into account a lot more variables than a human able to think about. Data Analysis has taken off recently because there are techniques that see patterns and find correlations in way more data than humans can contemplate.

I agree with you that you can't just point to corsi or fenwick and get any massive benefit. But I think there is potential for data analysis.

Personally, if I were an NHL team that wanted to make use of analytics, I wouldn't be hiring a lawyer to do it -- I'd be going to MIT and getting a stable of the brightest nerds around. It seems like most of the hockey analytics I see on the web are not very advanced, and seems like it's a bit of a weekend warrior game. But I'd be really interested to see what a competent group of data scientists could do.

 

[Oilslick941611]

I find it funny how too many people are either pro advanced stats or against them.

I also find it funny how people can easily discount Dellow's contribution without knowing what his contribution was. (Basically people are saying he is a failure based on what people on HF have said about what they think the contributions were?)

Personally, I don't like how condescending Dellow can be. That doesn't mean his data doesn't have value.

I was at a golf tournament with a guy that owns a company whose product is advanced hockey stats. Bright guy.

He stated that advanced hockey stats digs way deeper than what the Corsis and Fenwicks do. There are 45 staff that go over each second of film in a game charting everything from what we hear about to things like skating speed, pass speed, etc etc.

Chicago is a big proponent of advanced stats. http://www.cbssports.com/nhl/eye-on...m-stan-bowman-says-analytics-give-team-leg-up

So, you get tons of different data. You sift through and try to determine what it means and try to implement things. Just because it doesn't work, doesn't mean the data is bad. It could mean the interpretation and implementation of data based strategy is bad.

If I have a stat that says Ales Hemsky loses the puck within 3 feet of the opposing blueline 48% of the time, I might coach him to do several different things. (Dump in, pass over the blueline, carry if he is not the first player in the zone)

What if none of what I coach him to do works any better? Is the stat terrible? Nope.

Let's remember to take stats for what they are. Information. Or let's at least do it 65.3% of the time.

this the problem with advanced stats. you don't need to crunch numbers to realize something like that and its the same way for 99% of the "advanced stats"

advanced stats for are for people to argue with over the internet. the hockey people in managment positions don't need to crunch numbers to learn what they already now by just watching.

Advanced stats lend themselves better to static games like baseball or football where a play lasts 4-30 seconds. Hockey is too dyamic and there are too many variables too many other factors (like other players on your team and the other team) affecting any play at any given time.

 

[InjuredChoker]

I wonder what Scotty Bowman thinks of this analytics ****. Spend the money elsewhere and just go get some honest hockey players. This team is the PERFECT exaple of why alanytics arent worth jack ****.

It should be noted, the Blackhawks hockey analytics’ position reports directly to Scotty Bowman, Senior Advisor, not Stan Bowman (GM), Norm Maciver (Asst. GM) or anyone on the Hawks’ coaching staff. The same goes for the limited manual advanced statistics coming out of Rockford, those are emailed to Scotty after every game as well.

Scotty Bowman has been a proponent of using hockey analytics in various forms for many years, dating back when he was still with Detroit.

The Blackhawks and Carolina Hurricanes each went in on new advanced software applications three years ago and Scotty has overseen those numbers ever since.

http://thethirdmanin.com/2014/09/20...cs-front-office-changes-info-major-name-gone/

 

[Master Lok]

Advanced stats lend themselves better to static games like baseball or football where a play lasts 4-30 seconds. Hockey is too dyamic and there are too many variables too many other factors (like other players on your team and the other team) affecting any play at any given time.

I guess you know better than other professional hockey people and organizations who use it like the Blackhawks...

 

[Oilslick941611]

I guess you know better than other professional hockey people and organizations who use it like the Blackhawks...

exceptions aren't the rule.

also if you read my post you'd know I'm not saying they are useless, I'm saying they don't offer anything you can't learn from watching. I'm saying they are redundant.

 

[Halibut]

exceptions aren't the rule.

also if you read my post you'd know I'm not saying they are useless, I'm saying they don't offer anything you can't learn from watching. I'm saying they are redundant.

They are a useful tool to prevent user bias. If you are seeing a player as great but his stats show up as him being horrible it's a reason to double check what it is you think you are seeing. They're not always true but neither is what you think you are seeing in games.

 

[TKB]

Next, some disclosure here. I've been a Chess player since I was 4yrs old. Played competitive chess, speed chess, and against Computer chess programs since they were first mass produced. Played many chess engines on line etc.

I'll mention that until recently chess computers, which have existed in some form since the earliest computational systems existed, have often been limited. Lets understand as well that the top minds in the world program, make submissions to, and help to write computer chess programs. It took decades before Chess computer programs were even competitive against reasonably good chess players. Until say the 80's I was beating Chess programs at any selected level of play. You would note several algorithms that were wrong and exploit them as much as you wanted to. Getting bored you would find and exploit other errors in the programming.

I only mention this because one of the basic problems in chess programming is how many moves in advance a program could detect and play. As such programs were susceptible to such things as disguised pins, traps, sacrifices for positional gain, and pitfalls. With all of these having to be programmed in to programs to make them somewhat better. But with instances of real play approaching infinity. Until recent decades programmers had to work with computing limitations that involved processing capacity, speed, memory etc. So even though Grandmasters were involved in programming the intelligence of the programs was limited due to computer limitation.

Next hockey. Infinite possibilities, infinite variables, fastest game on earth. Infinite sequences and possible results on ANY shift which can't be predicted.

Yet I'm inundated with advanced stats that every day of the week cite such things as player X should be carrying the puck in more instead of dumping and that's it. That's the end all conclusion.

Well friends that a ONE MOVE analysis that ignores ANY subsequent result on shifts that can last over a minute and contiguous play and sequence that can be several minutes at a time.

Hockey is not chess, but is certainly not in anyway a one move, one analysis at a time game where it makes sense to make simplistic statistical observations such as player X should carry puck more, shoot puck more, whatever, because it results in one variable (More shots on net) while ignoring every possible resultant that follows.

A chess algorithym that explored only one move recommendation would be a waste of anybodies time.

That is a great analogy and observation. A similar analogy I like regarding the carry v dump decision is comparing zone entries to tennis shots. A tennis player isn't trying to hit a winner with every shot, rather he/she is working to (hopefully) set up a favorable position to deliver the "winner".

Likewise not ever zone entry has to produce a goal or significant shots, an important aspect if no immediate opportunity is available is to maintain a good structure, and prevent the "winners" or good set ups going the other way.