HFBoards

Go Back   HFBoards > NHL Western Conference > Pacific Division > Edmonton Oilers
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
Notices

The applicability of baseball-style "moneypuck" and "sabremetrics" in hockey

Reply
 
Thread Tools
Old
01-12-2015, 07:23 PM
  #1
Cloned
Dial M for Michelle
 
Cloned's Avatar
 
Join Date: Aug 2003
Posts: 25,013
vCash: 500
The applicability of baseball-style "moneypuck" and "sabremetrics" in hockey

So I've been reading the Dellow thread and the argument is inevitably come back to the applicability of using advanced stats in hockey. Rather than have that thread turn into a back and forth about that issue, I figured we could discuss it in a separate thread.

Here's my take on it:

The problem with applying advanced statistics to hockey (and any other major sport really) is that baseball is unique in its disposition.

It's a game with a deliberate and repeatable mechanic - i.e. the pitcher throws a ball to the batter and the outcome can be classified in a limited number of discrete possibilities - and as such the value of individual, isolated statistics can tell you a lot about how well a player can contribute to wins. There isn't a lot of variation in the set of circumstances that may affect said outcome (at least as compared to other sports). And this deliberate mechanic occurs literally thousands of times a season. Statistically speaking, it's a dream: you have what essentially amounts to a repeated set of circumstances giving rise to a LARGE sample size with a limited set of outcomes. It's so dreamy that they've even been able to MEASURE the effect of "luck" (i.e. variability in predictable outcomes) with certain stats that tell you how lucky or unlucky a batter or pitcher has been compared to the "average" player in baseball.

There is not really any other major North American sport like that. Certainly not hockey, which is a free-flowing game and probably the farthest you can get (aside from soccer) from a game with a deliberate and repeatable mechanic.

Is there value in advanced stats in hockey? Yes.

Will it ever be as useful or as reliably predictive of player/team performance as in baseball? No, not ever. I feel 100% confident in saying that. Put it this way: in baseball you can actually measure WAR (wins above replacement level) for every position on the field. And it actually correlates pretty well with team performance. In hockey? Good luck.

__________________

Sig AND X-mas avatar courtesy of The Nemesis

"Pull yourself together!" - Solid Snake to Otacon, multiple times in the series
Cloned is offline   Reply With Quote
Old
01-12-2015, 07:40 PM
  #2
iCanada
Registered User
 
Join Date: Feb 2010
Location: Edmonton
Country: Canada
Posts: 4,326
vCash: 400
I think the other big reason that sabremetrics in hockey have failed is because all the math guys in hockey are taking the stats in the wrong direction.

Baseball took a bunch of sexy numbers and turned them from rarara flashy statistics (batting percentage? Who gives a damn if you miss the ball two times and get hit with it once before knocking it out of the park every single time you hit the plate, that's way better than completing on over 90% of your attempts and never making it to base.) and turned them into numbers that actually compute who scores the most points for a team. They went from a flashy meaningless number that didn't correlate into points, to a one that directly did.

We took a number that literally decides how many points a team has in the game, and are trying to replace it with a number that says how many times you attempt to score. What the **** is that, I dont care how many shots you take or how often you have the puck, I care if you score more goals than the other team. We went from a number that says how many points you get, to a flashy number that says how many times you attempt to get points... backwards from baseball.

Ideally, if there was a way to quantify how many chances you would get to score, a conversion percentage stat would be great, but with the random nature of the game its kind of impossible to use that as a predictor.

iCanada is offline   Reply With Quote
Old
01-12-2015, 07:47 PM
  #3
joestevens29
Registered User
 
joestevens29's Avatar
 
Join Date: Apr 2009
Posts: 30,397
vCash: 2627
Baseball never used special stats. They took common stats like on base %, runs, etc.. And applied common sense

joestevens29 is offline   Reply With Quote
Old
01-12-2015, 07:51 PM
  #4
Cloned
Dial M for Michelle
 
Cloned's Avatar
 
Join Date: Aug 2003
Posts: 25,013
vCash: 500
Quote:
Originally Posted by joestevens29 View Post
Baseball never used special stats. They took common stats like on base %, runs, etc.. And applied common sense
Well, there are some pretty advanced stats out there in baseball. Most of them are rooted in common sense but some of the manipulations are not obvious.

Cloned is offline   Reply With Quote
Old
01-12-2015, 07:53 PM
  #5
I am the Liquor
Wrong Way Eberle
 
I am the Liquor's Avatar
 
Join Date: Oct 2008
Location: Sunnyvale
Country: Canada
Posts: 36,876
vCash: 1400
They keep moving the goalposts = a distinct lack of credibility./thread

I am the Liquor is offline   Reply With Quote
Old
01-12-2015, 07:55 PM
  #6
iCanada
Registered User
 
Join Date: Feb 2010
Location: Edmonton
Country: Canada
Posts: 4,326
vCash: 400
Quote:
Originally Posted by joestevens29 View Post
Baseball never used special stats. They took common stats like on base %, runs, etc.. And applied common sense
Base % was a special stat at one time.

lol.

We just havn't found the right stat(s) yet. You get me a stat that can quantify how many chances a player can get with strong predictive qualities, and a stat that can qualify how many chances of those a particular player will convert into a goal, and you can predict the records of teams in the same manner as baseball, using a league-wide average and you can easily predict how a team will fare.

Now, you might say we already have that, but we dont. We have stats that measure the chances generated, and we have stats that measure the likelihood someone will score on a chance, but the problem is the chances stats we have are not predictive of future chances. Thing is, I'd be shocked if such a thing isn't developed within the next 30 years.

Now, in hockey you'll never get near the accuracy of Baseball, but we could get into striking range of what we see in Basketball easily. Its quite obvious some players generate more chances than others, we just need to find a way to quantify this, and then find a pattern. No one would call me crazy if I said Crosby made more scoring chances than Gazdic, we just have no way to measure this or predict this at this time.

iCanada is offline   Reply With Quote
Old
01-12-2015, 08:12 PM
  #7
joestevens29
Registered User
 
joestevens29's Avatar
 
Join Date: Apr 2009
Posts: 30,397
vCash: 2627
You'll never find a pattern or be able to predict it, if you could figure that out you would t be wasting time on sports stats.

There is no point in looking at any sport of the major sports and trying to find a way to look at stats like baseball. There is just too many variables in team sports that rely on teamwork to make a single play.

joestevens29 is offline   Reply With Quote
Old
01-12-2015, 08:13 PM
  #8
jumptheshark
McDavid Headquarters
 
jumptheshark's Avatar
 
Join Date: Oct 2003
Location: Lord of HFBOARDS
Country: United Nations
Posts: 68,634
vCash: 131
Will keep it short--I could go on for several pages on the subject

breakdown of flaws with "fancy stats"

1) Hockey is more of a team sport then baseball
2) While baseball team play 3 and 4 game series in the same city--hockey teams play 3 games in 4 nights in 3 different time zones--a few years ago I did a hockey trip where I went from Ft Lauderdale to NYC and then to St Louis in 6 days and it was a hassle--imagine if I had to play a 3 hour hockey game as well?
3) Wear and tear on players
4) When teams start "showcasing players" ie they will move a player to the top line to help improve his numbers because they want to move him and if you look at the ten games he played on the top line it will give a fake idea of how said player is (two players I will use as an example in 84/85 a guy names Warren Young scored 40 goals and came out of nowhere--who was he on a line with? A guy names Mario--based partly upon fancy stats--the Redwings gave Warren a huge deal to mentor Stevie Y--it became painfully obvious that 99% of Young stats were based upon his linemates. Second example " My dead granny could score 20 on that line" THis was coined after Dave Semenko score back to back 12 goal years--if you saw Sammy Skate you understand--Sammy ran shotgun on wayne's wing when team misbehaved. Now it is only 12 goals or 24 over 2 years--but if you are old enough to have seen the Samenk-Gretz-Kurri line play together--you are still doubled over in watching Sammy trying to catch up to the other two. Dave Lumley also spent time on that line as well--
5) Not only who are you playing with but who are you playing against. There used to be site--where it broke down which players did well against what teams. There were a few players who on fancy stats looked really good--but what off set these stats was the fact they lit their numbers up against the bad teams and did nothing against the good teams. People may argue the point a goal is a goal is a goal-- but player A scores 20 goals in a year--but scored 18 goals against non-playoff teams--mean he scored only 2 against what we would call competitive teams. We know all top lines are not the same and not all team are the same--if the NHL suddenly went back to a 21 team league that means 180 players are out of the NHL--If Edmonton and Carolina suddenly distbanded--how many players on each team still be in the nhl? Single digits each team is my guess. How many Hawks players would still in a 21 team league? 18--yes I know number 5 will confuse some


my post was longer then I wanted and might confuse some people

jumptheshark is offline   Reply With Quote
Old
01-12-2015, 08:47 PM
  #9
iCanada
Registered User
 
Join Date: Feb 2010
Location: Edmonton
Country: Canada
Posts: 4,326
vCash: 400
Quote:
Originally Posted by joestevens29 View Post
You'll never find a pattern or be able to predict it, if you could figure that out you would t be wasting time on sports stats.

There is no point in looking at any sport of the major sports and trying to find a way to look at stats like baseball. There is just too many variables in team sports that rely on teamwork to make a single play.
Ah but I disagree, everybody needs a passion, everybody needs a hobby. A Geophysicist needs to unwind somehow, as does an Engineer, a Mathematician, etc.

For you collecting, analyzing, creating regressions, and suggesting trends / formulas based upon tonnes of data is a crap load of work and boring. For another it might be how they de-stress and unwind on the coach with a beer on a Saturday morning. I dont think its a matter of if it will happen, its a matter of when, imo.

Here is the thing, you dont even necessarily need to find a perfect variable, all you need is something that correlates strongly enough.

Quote:
Originally Posted by jumptheshark View Post
Will keep it short--I could go on for several pages on the subject

breakdown of flaws with "fancy stats"

1) Hockey is more of a team sport then baseball
2) While baseball team play 3 and 4 game series in the same city--hockey teams play 3 games in 4 nights in 3 different time zones--a few years ago I did a hockey trip where I went from Ft Lauderdale to NYC and then to St Louis in 6 days and it was a hassle--imagine if I had to play a 3 hour hockey game as well?
3) Wear and tear on players
4) When teams start "showcasing players" ie they will move a player to the top line to help improve his numbers because they want to move him and if you look at the ten games he played on the top line it will give a fake idea of how said player is (two players I will use as an example in 84/85 a guy names Warren Young scored 40 goals and came out of nowhere--who was he on a line with? A guy names Mario--based partly upon fancy stats--the Redwings gave Warren a huge deal to mentor Stevie Y--it became painfully obvious that 99% of Young stats were based upon his linemates. Second example " My dead granny could score 20 on that line" THis was coined after Dave Semenko score back to back 12 goal years--if you saw Sammy Skate you understand--Sammy ran shotgun on wayne's wing when team misbehaved. Now it is only 12 goals or 24 over 2 years--but if you are old enough to have seen the Samenk-Gretz-Kurri line play together--you are still doubled over in watching Sammy trying to catch up to the other two. Dave Lumley also spent time on that line as well--
5) Not only who are you playing with but who are you playing against. There used to be site--where it broke down which players did well against what teams. There were a few players who on fancy stats looked really good--but what off set these stats was the fact they lit their numbers up against the bad teams and did nothing against the good teams. People may argue the point a goal is a goal is a goal-- but player A scores 20 goals in a year--but scored 18 goals against non-playoff teams--mean he scored only 2 against what we would call competitive teams. We know all top lines are not the same and not all team are the same--if the NHL suddenly went back to a 21 team league that means 180 players are out of the NHL--If Edmonton and Carolina suddenly distbanded--how many players on each team still be in the nhl? Single digits each team is my guess. How many Hawks players would still in a 21 team league? 18--yes I know number 5 will confuse some


my post was longer then I wanted and might confuse some people
1) Tell that to Basketball / football.

2) Learning curve and fatigue could be accounted for, and I'd argue to some extent are fairly negligible, seeing how the entire league goes through this process, and its rather easy to quantify which teams have the most fatiguing schedule regardless.

3) This is pretty well identical to #2.

4) I wont argue with you here. Complex as all hell. Doesn't mean you cannot quantify it. I'll put it this way: pick up a textbook on something like Thermodynamics or Stress Analysis; if we can quantify rates of rates of change with respect to 20 to 50 different variables we can model how many chances to score a player will generate with respect to time with a moderate to good amount of success quite easily.

5) This is kind of negligible as well. Its also a factor in baseball. Measure a bunch of guys on-base percentages against a bunch of 15 yearolds and against the club that wins the world series. The biggest thing here i think is that in Baseball you have a larger sample size so the statistical noise is easier to weed out; theoretically everything will be distributed on a normally distributed curve with regards to opposition, however when you only play an opposing team twice instead of 20 times it can be mistaken as the norm rather than the remarkable top 5% or something.

iCanada is offline   Reply With Quote
Old
01-12-2015, 08:51 PM
  #10
jumptheshark
McDavid Headquarters
 
jumptheshark's Avatar
 
Join Date: Oct 2003
Location: Lord of HFBOARDS
Country: United Nations
Posts: 68,634
vCash: 131
Quote:
Originally Posted by iCanada View Post
Ah but I disagree, everybody needs a passion, everybody needs a hobby. A Geophysicist needs to unwind somehow, as does an Engineer, a Mathematician, etc.

For you collecting, analyzing, creating regressions, and suggesting trends / formulas based upon tonnes of data is a crap load of work and boring. For another it might be how they de-stress and unwind on the coach with a beer on a Saturday morning. I dont think its a matter of if it will happen, its a matter of when, imo.

Here is the thing, you dont even necessarily need to find a perfect variable, all you need is something that correlates strongly enough.



1) Tell that to Basketball / football.

2) Learning curve and fatigue could be accounted for, and I'd argue to some extent are fairly negligible, seeing how the entire league goes through this process, and its rather easy to quantify which teams have the most fatiguing schedule regardless.

3) This is pretty well identical to #2.

4) I wont argue with you here. Complex as all hell. Doesn't mean you cannot quantify it. I'll put it this way: pick up a textbook on something like Thermodynamics or Stress Analysis; if we can quantify rates of rates of change with respect to 20 to 50 different variables we can model how many chances to score a player will generate with respect to time with a moderate to good amount of success quite easily.

5) This is kind of negligible as well. Its also a factor in baseball. Measure a bunch of guys on-base percentages against a bunch of 15 yearolds and against the club that wins the world series. The biggest thing here i think is that in Baseball you have a larger sample size so the statistical noise is easier to weed out; theoretically everything will be distributed on a normally distributed curve with regards to opposition, however when you only play an opposing team twice instead of 20 times it can be mistaken as the norm rather than the remarkable top 5% or something.
Fancy stats has not been perfected in either Basketball or football yet--they are working at it and that is one reason why some team gut their team year and after year

jumptheshark is offline   Reply With Quote
Old
01-12-2015, 08:54 PM
  #11
McOkMcgoMcoil
Registered User
 
McOkMcgoMcoil's Avatar
 
Join Date: Aug 2005
Location: Edmonton, AB
Posts: 14,087
vCash: 500
You are right. Variables in hockey are to correlated with each other. I think Soccer is the extreme, that is a sport where they can't even seem to come up with a good statistic for who is the best offensive player, let alone all around best player.

Bottom line some sports lend them selves to stats others don't. Take something like darts, I mean clearly you can break every element down. The basic score keeping does just that.

Any way, doesnt' mean advanced stats are meaningless in hockey, but you have to be realistic. Will take a lot more work to get there from here than it did in baseball.

McOkMcgoMcoil is offline   Reply With Quote
Old
01-12-2015, 09:28 PM
  #12
jumptheshark
McDavid Headquarters
 
jumptheshark's Avatar
 
Join Date: Oct 2003
Location: Lord of HFBOARDS
Country: United Nations
Posts: 68,634
vCash: 131
Quote:
Originally Posted by okgooil View Post
You are right. Variables in hockey are to correlated with each other. I think Soccer is the extreme, that is a sport where they can't even seem to come up with a good statistic for who is the best offensive player, let alone all around best player.

Bottom line some sports lend them selves to stats others don't. Take something like darts, I mean clearly you can break every element down. The basic score keeping does just that.

Any way, doesnt' mean advanced stats are meaningless in hockey, but you have to be realistic. Will take a lot more work to get there from here than it did in baseball.
Never said they were meaningless--everyone on this board who plays fantasy pool will tell you stats are important--some stats more then others. But it comes down to the process of interpreting that stats to get viable and pliable information that can then be used to understand a player. Hockey is not there yet. IT is about balance. On one hand we need to look at what I call deep stats or what other call fancy stats and on the other hand--look what happens on the ice. Balance information out to get a better process of what is going wrong with this team.

jumptheshark is offline   Reply With Quote
Old
01-12-2015, 09:36 PM
  #13
McGhostbuster
Who ya gonna call?
 
McGhostbuster's Avatar
 
Join Date: Apr 2007
Location: Edmonton
Posts: 7,713
vCash: 50
It's not that hockey can't be statistically analysed.

Hockey is very dynamic like you mentioned.

Baseball is very static.

It's like comparing elementary math vs. doctorate levels of calculus but hockey number munchers are still using that grade 3 knowledge to come up with the numbers.

No one right now has the time or technology to really get deep and dirty in it.

McGhostbuster is online now   Reply With Quote
Old
01-12-2015, 09:38 PM
  #14
sweetcrude
Registered User
 
sweetcrude's Avatar
 
Join Date: Jan 2006
Country: Canada
Posts: 621
vCash: 500
Quote:
Originally Posted by Cloned View Post
Will it ever be as useful or as reliably predictive of player/team performance as in baseball? No, not ever. I feel 100% confident in saying that. Put it this way: in baseball you can actually measure WAR (wins above replacement level) for every position on the field. And it actually correlates pretty well with team performance. In hockey? Good luck.
Agreed.

Games like hockey inherently have greater complexity within it as compared to others. Whereas it may be likely to find metrics that potentially indicate certain "conclusions" about the game / team(s) / player(s), they need to be treated with judgment. Good judgment, that is. Metrics may not in fact mean anything unless they begin to demonstrate causation instead of correlation only.

The difficulties I have in listening to people speak about the subject, or write about it, are:
1) hardly anyone has the tools, nor the background, to do the analysis being performed
2) people inherently trust that those performing the analysis have performed the mathematics correctly
3) Important parameters may not even be currently considered to be analyzed
4) Popular metrics used are rarely scrutinized seriously
5) Complex systems necessitate inherently complex analyses, with likely decreasing probability of finding / revealing strict causation

and more...

I'll leave it there.

sweetcrude is offline   Reply With Quote
Old
01-12-2015, 09:40 PM
  #15
Declassified
Valkyrie Pedalworks
 
Declassified's Avatar
 
Join Date: Jul 2006
Location: Ottawa
Country: Canada
Posts: 3,981
vCash: 145
the crux of the issue for me is that advanced stats tell you nothing that watching and scouting won't tell you.

Declassified is offline   Reply With Quote
Old
01-12-2015, 10:22 PM
  #16
oilinblood
Registered User
 
Join Date: Aug 2009
Posts: 3,691
vCash: 515
Quote:
Originally Posted by Declassified View Post
the crux of the issue for me is that advanced stats tell you nothing that watching and scouting won't tell you.
This. Watching how players deal with facing certain circumstances, seeing how they use their teammates and environment, with an without the puck...can not be replaced by stats. Witnessing theur abilities to read their time and space.

a defenseman is being pressured by a localized odd man by an opposing team that is entering his zone, he breaks it up and has no immediate pass option (or very risky pass option where he is outnumbered in his own zone). He makes a clean chip off the boards and out of the zone, banking the puck to a fifty,fifty position near a teammate. Even if the other team gets that puck, takeaway, its the cleaner play to get the puck out and make the other team regroup onside whil your team can get focussed and reset.
Stats will say a different story.


Last edited by oilinblood: 01-12-2015 at 10:28 PM.
oilinblood is offline   Reply With Quote
Old
01-13-2015, 03:26 AM
  #17
Cloned
Dial M for Michelle
 
Cloned's Avatar
 
Join Date: Aug 2003
Posts: 25,013
vCash: 500
Quote:
Originally Posted by Ghostbuster View Post
It's not that hockey can't be statistically analysed.

Hockey is very dynamic like you mentioned.

Baseball is very static.

It's like comparing elementary math vs. doctorate levels of calculus but hockey number munchers are still using that grade 3 knowledge to come up with the numbers.

No one right now has the time or technology to really get deep and dirty in it.
Actually, I'm not sure you CAN ever get to the point where hockey analysis gives you reliable, meaningful information that predicts outcomes.

To expand on the comparison with baseball, is there ever a situation in baseball that you CANNOT account for in terms of a defined, discrete outcome? Fly ball out, ground ball RBI, runners on base, strike out, etc. There isn't very much observational bias at all.

Take hockey now. What constitutes a scoring chance? What is a QUALITY scoring chance? Can a non-scoring chance shot lead to a scoring chance, and if so how should that non-scoring chance shot be accounted for? Does a hit lead to a scoring chance? Is the hit even properly recorded as a hit? How many situations cannot be accounted for as a defined, discrete outcome due to the dynamics of the game?

I'll extend even further. You could argue that hockey has discrete outcomes just like baseball. Goal, icing, offsides, save, etc. But what leads up to the goal? Does a MISSED scoring chance lead to a goal the other way? Is a missed pass a giveaway, or a takeaway by the other team, or both, or neither? How would you quantify that? The dynamics are immensely complex, probably too complex to be defined by a number.

In baseball, you can pretty much tell what happened in the game by looking at the scoring sheet. Baseball is a set of discrete, predictable events that occur in sequential order. It's a game where each individual battle between two opponents (i.e. pitcher and batter, batter and fielder) can be defined pretty easily without much observational bias. If I wanted to find out a run was scored I can look at each event individually and account for everything that was happening at that moment. I can go back to look at the tiniest details, like pitch velocity, at the moment of each event.

Maybe someone will prove me wrong and they'll be able to come up with analytical ways to define the game of hockey that are predictable and reliable. But I can't see it happening. I mean, even in baseball, which is pretty much a statistical dream due to its nature, there are arguments about the reliability of certain stats. And that's without having to deal with the complexity and bias involved in hockey analytics. It might just be an impossible feat.


Last edited by Cloned: 01-13-2015 at 03:35 AM.
Cloned is offline   Reply With Quote
Old
01-13-2015, 07:31 AM
  #18
Mr Sakich
Registered User
 
Mr Sakich's Avatar
 
Join Date: Mar 2002
Location: Motel 35
Posts: 8,506
vCash: 500
hockey is different than baseball but it is similar to soccer and basketball. The fancy stats for all these fluid sports are progressing.

As the quality of the information gets better (tracking devices in the uniforms), the predictive quality of the stats will improve. Too early to dismiss analytics.

Mr Sakich is online now   Reply With Quote
Old
01-13-2015, 08:21 AM
  #19
Dorian2
Positional Bias.
 
Dorian2's Avatar
 
Join Date: Jul 2009
Location: Edmonton
Country: Canada
Posts: 8,295
vCash: 99
Just reading the title gives me a headache.

Advanced stats (to me) are what they are, a vehicle to manipulate the outcome that you want. Just like any statistics.

I'll stick to the basics.

Dorian2 is online now   Reply With Quote
Old
01-13-2015, 09:11 AM
  #20
Fourier
Registered User
 
Fourier's Avatar
 
Join Date: Dec 2006
Location: Waterloo Ontario
Country: Canada
Posts: 11,465
vCash: 500
I think there is a significant amount of confusion caused by improperly distinguishing between what these stats can tell you and what they can't. The problem I have had all along is the significant over reach in what a statistic says about the play on the ice and in particular about an individuals contribution to the play. This over reach tends to result in the whole concept being discredited.

Personally I think there is a fair bit of value in many of these "advanced stats" when taken in context. Unfortunately the context is often missing in the most egregious manner. But I also think that the Stats we tend to see are not likely the ones individual teams use explicitly in making their own decisions. Teams have the advantage of resources to break down game film to get the information they specifically desire.

Virtually all statistics have their good points and their limitations. For example goals scored is obviously one of the most basic stats in hockey. If I told you that a certain player scored 117 goals over a four year period in the years playing just 2002-2003 through 2006-2007 in just 276 games (roughly 35 goals per year prorated) I think we would all agree that this guy is a top flight goal scorer. However if I ask you if this guy has an elite shot or at least a very good one, things get dicey. Having an elite shot is a characteristic we connect with goal scores and I think we would all feel that most guys capable of scoring 35 goals a year can really shoot the puck. But this dudes name is Ryan Smyth so that additional piece of information gives those of us who watched him play the proper context to answer the question above.

Fourier is online now   Reply With Quote
Old
01-13-2015, 09:54 AM
  #21
missinthejets
Registered User
 
Join Date: Dec 2005
Posts: 4,030
vCash: 500
Statistics are only ever as useful as the person analysing them.

It's a tool, if used right it can help, it's not a substitution for actual scouting but it can be a helpful supplement to it.

missinthejets is offline   Reply With Quote
Old
01-13-2015, 10:20 AM
  #22
nki
Registered User
 
Join Date: Sep 2006
Posts: 2,743
vCash: 500
You can't build a team with statistics. It's one of many tools in the belt. If you want to see what type of statistical analysis actually matters within the context of pro-hockey, listen to Darryl Belfry. Skills coach for many of the top players with his business (P. Kane, Tavares, MacKinnon etc.), worked in LA's development, and currently a development consultant with Maple Leafs. I've listened to all of his videos and he's a real smart guy as you might expect, hope the videos don't get taken down.



And here's a few of his tweets on Sabres and Oilers:

Quote:
To the degree that the @BuffaloSabres parlay their optimum draft positioning over this three year window into the "Pittsburgh Penguins Model" or they become the "Edmonton Oilers" of recent years will in large part be determined on the skill habits of their immediate core and current placeholders. The path to competitive respectability will rest on what they do between now and before Connor McDavid arrives.

If the 2014-2015 Sabres are going to continue, for a second year, to ignore the clear skill application principles to be successful in todays @NHL, then Tim Murray must get every player under 22 years old out of the Buffalo environment and back to Junior/AHL immediately (Zadorov, Ristolainen, Girgensons and Reinhart). What they are learning playing in @BuffaloSabresare skill application habits that will pull them further into the abyss, most recently experienced by @edmontonoilers.

Every team in the @NHL has personnel upgrades they want to make to either give them more depth, size, speed, etc and/or some special systematic wrinkle that they want to define their team play. However, to be successful in today's @NHL, you must adhere/foster/cultivate a few translatable skill application principles which transcend and supersede all personnel and/or system your team plays. While I won't list all the principles and skill blends here, here are a couple to illustrate:

1. In the Defensive Zone, the primary player defending at the puck, must use body position, stick position and purposeful angling to take control of the space and force the puckcarrier into a secondary movement (stop or turn), must contain the puck with a good stick and control the hips of the puckcarrier providing a seal contact opportunity.

In the Offensive Zone, the player with the puck, must take control of the feet of the defender, extend the possession to beat their check in such a way that it forces a defensive switch of responsibilities, recognize that switch and be able to make a 10 foot play inside the rotation.

2. When Defending the Rush, you must consistently outnumber the opposition (back pressure support) on entry in an effort to defend the line and push the point of entry (dump or carry) outside the dots.

When Attacking off the rush, your D must consistently make a play in the Hot Zone and utilize, as a group, any combination of false gap, speed behind the puck, isolation 2 on 1 entry and D support on entry.

The following illustration clip is the first extended sequence of possession that the Leafs had in the zone in the third period, coupled in sequence by multiple offensive possessions by the Sabres.

http://youtu.be/G9pVEKB-7X0

Once you shift from ... controlling space and making plays as a mentality.... to a ...get it, get it out, get it over the redline, get it in and wing it to the net mentality...this is the result. Now if the Sabres go the next 72 games playing with the latter mentality and its associated skill sets, that will become who you are as a team. Which is not projectable and reminiscent of seeing Oiler 1st Overall Selections stepping into a similar environment in Edmonton and start chipping pucks in deep when they have favourable offensive gaps to work with.

As a team, irrespective of personnel or system, there must be a concerted focus to both execute these principles consistently and build the skill set, movement qualities that drive that execution. At that point, of course, you can stack any number of support systems, personnel upgrades etc.

The best example of this is the Myers/Georges pairing. Myers, escapes pressure and makes a play to Georges and re-hinges underneath to establish a false gap. Georges punts the puck up the strong side for a knife dump in. The placeholder (Georges), must facilitate the core player (Myers) in executing translatable skill patterning. While Georges works hard and competes all over the ice, his approach to control space and his puck play doesn't translate to todays game. Therein lies the problem.

I wish it were as easy as "working harder," but it is not. Work ethic is not at the core of the @BuffaloSabres problematic start, nor is it a massive talent discrepancy in relation to the rest of the league, rather a blatant disregard to the skill habits that dictates how space and possession is controlled.

As you look ahead to next year and the infusion of a generational player to the roster, I ask, whom is this player going to play with? It will invariably be with some combination of current "core" players, "unseasoned prospects" and placeholders. The Sabres challenge is to prepare that misfit bunch awaiting the generational player to understand and consistently execute projectable skill execution habits.

What gets lost in the rise of the @LAKings is the work of the Development Dept. Whose role on the surface appears to be in prospect to player conversion, but in reality, it is to create translatable skill principles as an undercurrent of performance expectation throughout the organization.

If the @BuffaloSabres want to avoid the arduous road still trudging along in Edmonton, they will need to embrace this season as an opportunity to build the "translatable skill culture" with the group that will be holding over as a stable and unified platform for their hot shot prospects to grow from. That process must be fully engaged in Buffalo, Rochester and with their entire development staff. As much as they have a scouting challenge, they have a development challenge that must be met head on and with unified implementation.

If they are insistent upon ignoring these principles, then they need to ship the kids out immediately, otherwise they will perpetuate what they are learning which will stunt the growth pattern of the entire organization. At the very least a concerted effort must be in giving McDavid a fighting chance, or if done properly a running start."
http://beerleagueheroes.weebly.com/h...ows-his-hockey


Last edited by nki: 01-13-2015 at 10:25 AM.
nki is offline   Reply With Quote
Old
01-13-2015, 10:40 AM
  #23
Master Lok
Registered User
 
Join Date: Jul 2003
Location: Edmonton
Posts: 7,072
vCash: 500
Quote:
Originally Posted by Declassified View Post
the crux of the issue for me is that advanced stats tell you nothing that watching and scouting won't tell you.
Except that people have an inherent bias. Different people see different things. For example, I usually make a point of listening to analysts like Ray Ferraro because he points out things on the ice that make sense to me as important points. I usually ignore analysts like Jamie McLennan or Ryan Rishaug or Aaron Ward because after listening to them, I don't value their opinion of what they see in the game.

I see statistics as an attempt remove that bias.

Master Lok is offline   Reply With Quote
Old
01-13-2015, 12:22 PM
  #24
SPIRIT
Registered User
 
SPIRIT's Avatar
 
Join Date: Mar 2014
Posts: 410
vCash: 50
Quote:
Originally Posted by nki View Post
You can't build a team with statistics. It's one of many tools in the belt. If you want to see what type of statistical analysis actually matters within the context of pro-hockey, listen to Darryl Belfry. Skills coach for many of the top players with his business (P. Kane, Tavares, MacKinnon etc.), worked in LA's development, and currently a development consultant with Maple Leafs. I've listened to all of his videos and he's a real smart guy as you might expect, hope the videos don't get taken down.



And here's a few of his tweets on Sabres and Oilers:



http://beerleagueheroes.weebly.com/h...ows-his-hockey
Really good post.

It's an interesting discussion.

As someone who maths and does data analysis for a career... I do not - never have - understand the practicality of the hockey analytic.

The above insight is interesting, but at the same time there seems to be a compulsion to add complexity to what should already be simple coaching practice.

Develop a system that works, find players to fill roles, continue improving the quality of pieces you can fit in your overall puzzle.

Huge Darnell Nurse fan... unfortunately there is not a statistic yet that captures "hulking, nasty, intimidating".

Maybe his very presence enacts some sort of -10%-to-all-Corsi-while on ice mechanic, like some sort of game of alpha-male Magic the Gathering.

There is one angle of analytics no one has touched upon yet, and as my above gaming reference might foreshadow, it is rooted in... simulation over speculation!

NHL by EA games improve every year - this year, NHL 15 (for the Xbox 360, apparently the ps4/xb1 versions were a bit watered because of their new engine).

The player development and scouting mechanics in that game are... remarkable! Players are broken down by categories, categories are given star-ratings by potential... scouting is done through random algorithms, you can improve upon the success of the number generation through improving your "staff"...

Here is the kicker... boot up NHL13 and look at our prospect stats. They're bang on! Bust city! Through simple rank-and-appropriate metric recording, EA was able to apparently dominate our scouting staff.


So, to my point...

You have an engine that is flexible enough to simulate what the actual outcome of a game might be based on individualized statistics playing themselves out over the course of a truly dynamic, random engine.

Sure, game by game basis is not overly compelling, but if you could trend the data over thousands of attempts you would definitely see patterns crystallize.

Hey! 3 tiny center-men! We haven't made the players in 839838382 out of 899838382 seasons!

Sure beats the heck out of hind-sight assessments of 'fancy stats' - it actually gives you a chance to potentially look forward with a statistical backing that is greater than fairy-dreams.

SPIRIT is offline   Reply With Quote
Reply

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 10:25 PM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"

vBulletin Copyright ©2000 - 2016, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. 2016 All Rights Reserved.