HFBoards

Go Back   HFBoards > NHL Eastern Conference > Atlantic Division > Montreal Canadiens
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
Notices

Sabremetrics/Microstats in Hockey

Reply
 
Thread Tools
Old
03-30-2012, 08:38 PM
  #51
Kjell Dahlin
Registered User
 
Kjell Dahlin's Avatar
 
Join Date: Jan 2010
Location: Québec, Québec
Posts: 1,998
vCash: 500
Quote:
Originally Posted by Lafleurs Guy View Post

(...)

Problem is that giveaways and takeaways vary hugely from rink to rink. And that leads to another big problem in analytics for a fluid sport such as hockey... subjectivity. What constitutes a giveaway for one person doesn't for another... if the NHL can't get it right themselves, then it casts doubt on the ability to 'count things' doesn't it?

(...)
True but I tried a little test...

You mentioned in a previous post: "... Size and grit ARE attributes that GMs simply can't ignore. And I'm not sure how you capture this via analytics...". I suggested the following parameters (hits, PIM, TkA-GvA) to answer this question:

--------------
Is player A grittier than player B.
--------------

I used raw data: no regression, I did not run tests to exclude exterior factors, no kNN, heck I did not even use the MS Excel solver! Nothing: just raw data.

--------------
Player's grit = f (hits, PIM, TkA-GvA)
--------------

So... someone who never watched a game would rank our players (from a grit perspective), that way:

Ryan White 1,00
Mike Blunden 0,83
Brad Staubitz 0,81
Erik Cole 0,78
Louis Leblanc 0,76
Blake Geoffrion 0,73
Travis Moen 0,71
Rene Bourque 0,70
Lars Eller 0,70
Mathieu Darche 0,69
Brian Gionta 0,68
Max Pacioretty 0,68
Aaron Palushaj 0,65
David Desharnais 0,64
Petteri Nokelainen 0,64
Tomas Plekanec 0,60
Scott Gomez 0,58

--------------------------------------------------------------------

Alexei Emelin 0,69
Josh Gorges 0,60
P.K. Subban 0,50
Yannick Weber 0,50
Raphael Diaz 0,49
Tomas Kaberle 0,47
Chris Campoli 0,47

I separated the forwards and the defensemen because I did not correct the inherent worst TkA-GvA for the Ds. I removed Markov, Saint-Denis and Engqvist because they did not play enough games.

Conclusion?

Despite the numerous flaws (subjectivity, inconsistency in the way the data is collected...) associated with the set of data provided by nhl.com, the results are not bad: someone who never watched a game would not look foolish when ranking our players from a grit perspective. It took me 5 minutes to do this and again: I used raw data - no regression, I did not run tests to exclude exterior factors... If I did this by the book, the results would be more precise.


PS I know (thanks to Mathletic and MathMan!) that "... on average the raw stat of hits for is negatively correlated with winning...". This little test was just an attempt to answer this simple question: is player A grittier than player B.

Kjell Dahlin is offline   Reply With Quote
Old
03-30-2012, 09:00 PM
  #52
Lafleurs Guy
Registered User
 
Lafleurs Guy's Avatar
 
Join Date: Jul 2007
Posts: 23,244
vCash: 500
Quote:
Originally Posted by Talks to Goalposts View Post
Its important to note that for these kinds of analytical discussions "skill" and "talent" refer to any abilities a player has that helps to repeatably produce the results desired. So ability to make space with elbows, to the extent it helps drive results, counts as a skill.
Right. But where it gets problematic is when you start trying to build a team together. Making it mesh. If a guy has great puck possession how much of it is skill vs. his linemate being somebody who's physical?

We built a team of small players up front with our core. Maybe their puck possession stats were all great on other clubs but put them together and you might not get the same result. Why? Because none are big guys and can be intimidated or beaten off the puck. And we've seen that even when a player has good puck possession stats it doesn't necessarily translate to points.

Lafleurs Guy is offline   Reply With Quote
Old
03-30-2012, 11:00 PM
  #53
EllertoKostitsynGoal
Registered User
 
EllertoKostitsynGoal's Avatar
 
Join Date: Nov 2011
Location: Mtl
Country: Canada
Posts: 1,055
vCash: 500
Quote:
Originally Posted by Talks to Goalposts View Post
2. Amount of Goals/Points doesn't necessarily have a direct correlation with value. There are plenty of NHL players that score a lot at the expense of defense. And if you give up more than you create then you're hurting your team even if you're one of the top scorers. A player's value has to correlate with how much they help you outscore the opposition rather than just how much they help you score.
This can't be said enough. People love shiny numbers so much that they forget that winning hockey games isn't technically about scoring goals. The goal is actually to score more goals than the opposition. Scoring goals is a big part of that but so is preventing goals against.

Wich comes back to puck possession, the more you have the puck, the more chances you have to score. The more you have the puck, the less chances the opposition has to score. There are all kind of usefull skills you can have in hockey and puck possession is one of them. Especially, since a pretty high % of goal scored aren't necessarily pretty passing perfect plays but a lot of goalmouth scrambles or kind of fluky goals that happens when you can push the play in the right direction.

About Gomez, he was actually scoring like a top 6 foward at ES (wich is a 30+ ES pts pace) before he came back from his second injury. No goals (no PP point either I think, wich hurted the totals) but he basically kept that pace on 1rst assist alone. He actually lead the league in ES 1rst assist per minutes of icetime before coming back, still leads our team too.

Like Mathman said, it's hard to say what exactly happened after that though I don't think having him play in a system where he's not allowed to carry the puck is really helping him (and everyone else really). We might have underrated Gionta (talking about this year since Gomez didn't really play with Paches, works for last year too) who's a terrific two way player. Or he might just have stepped into an elevator shaft and his days as an ES contributor are over. We'll likely only know if we buy him out and a team signs him on the cheap next year.

EllertoKostitsynGoal is offline   Reply With Quote
Old
04-01-2012, 06:30 PM
  #54
Lafleurs Guy
Registered User
 
Lafleurs Guy's Avatar
 
Join Date: Jul 2007
Posts: 23,244
vCash: 500
Quote:
Originally Posted by MathMan View Post
In the long term, pretty much. Look up at the various links I posted for any number of posts that demonstrate this.
Saying... go read a bunch of pages isn't really an answer man. I've seen your links before and I haven't really seen that question answered. If you have a place where it shows the actual correlation I'd love to see it. Don't send me to several different sites to find it.


Quote:
Originally Posted by MathMan View Post
Not to turn this into another pointless Gomez debate, but essentially: because Gomez doesn't suck nearly as much as advertised.
And I disagree with you strongly. The results speak for themselves.


Quote:
Originally Posted by MathMan View Post
Here's another point that's very important to realize. In hockey, there is not ever going to ever be any such thing as a method that says "if you do X, then you are 100% guaranteed to win", regardless of whether it's via analytics or not. There is actually an important lesson here that studying analytics will tell you, though it's certainly possible to learn this very important lesson by other means:
Well, if you score more goals than the other team, you'll win the game.
Quote:
Originally Posted by MathMan View Post
It's hockey. **** happens.
And some things can't be measured. If you've got teammates who don't like each other for example it can effect on ice chemistry.


Quote:
Originally Posted by MathMan View Post
More formally, Hockey is a game with a lot of variances, what might be called "luck" in a mathematical sense. Analytics will help you maximize your chances of winning games in the future. They will tell you what things will, in the long run, lead to better odds of winning. But better odds is all you are ever going to get.
Well, is it luck if your captain is an idiot who sleeps with a teammate's wife? (not using a specific example there) Is it luck if your captain ducks fighting another team's tough guy but then fights the younger brother of a teammate? (That is a specific example)

What goes on in the dressing room can impact what happens on the ice. In baseball a guy can be a jerk and still be a great hitter. He's not passing the puck to teammates or standing up for them on the ice... it's more of an individual game. Chemistry is an intangible taht's really ared to measure. Some players will work really well together and some won't. Hull and Oates are made for each other as an example because their skill complitment one another.

Quote:
Originally Posted by Roke View Post
I don't agree with the baseball assertion. The Blue Jays under JP Riccardi de-emphasized scouting and ended up not doing well - though it didn't help that their GM couldn't multi-task, gave out horrible contracts, and hamstrung their drafting and development by only drafting College guys and basically not signing international free agents.

In baseball, the best organizations (Boston, present-day Toronto, Tampa Bay come to mind) use a combination of both. Scouting tends to be focused a lot on skill and mechanics while you compliment that with the statistical analysis of a player's performance. If the two don't agree on a player or team that's an impetus to look closer and try to figure out why. There's a reason the Toronto Blue Jays at least doubled their scouting staff when Alex Anthopoulos took over as GM - scouting still has value.
Well even Billy Beane learned that. You need to look at the behind the scenes stuff too. And baseball is much friendlier towards analytics than hockey.

Quote:
Originally Posted by Talks to Goalposts View Post
I can see the case for intimdation being a factor but I think the premise that if a player has a talent for that it won't show up quantitatively. If your measuring overall play and your intimidation factor causes opposing player to play poorly it will show up in terms of you playing well. So like skating or stick handling if its an ability that affects the game in your favour it will show up in a measure of overall play. If it doesn't actually cause the opposition to play measurably worse then it wasn't actually a factor in the game.

And if you are good a intimidation but don't get good overall results then the rest of your game is probably so poor that intimidation is pointless for the team because your giving it away elsewhere.

I think there is a real danger of double counting attributes when you want to count things like intimidation over and above ability to measurably effect the game.



Corsi isn't supposed to be used as a straight measurement of how good a player is, that would be absurd for a number of reasons. Its used as a measure of dominance within circumstances. For example Subban and Weber had similar corsi/fenwick etc. last season but Subban's circumstances were by far and away more difficult.

Its also a matter of thinking less about individual brilliance and more defining a player's worth in the context of how well his team does with him on the ice and working backwards from that. Its a matter of the idea that team's that are good at this are good teams for whom we can predict that they will do well in the future, lines that are good at this are good lines that we expect good things from and that the value of a player is in how much their presence makes teams and lines good.


Your Hull example actually a good one for illustrating some of the concepts inherent to what we're are discussing here.

1. When we're discussing possession in this context its really a shorthand for a bigger concept than just having the puck on your stick. Its more of a complex of possession and territorial play involved in controlling who has the puck and where. With the overall goal of your team having the puck in scoring position more often than your opponent. So a forward that's excellent at getting in scoring position like Hull is going to have "possession" value even if he doesn't handle the puck much himself. On the other side you can have defensive defensemen that have "possession" value even if they don't handle the puck much if they prevent opportunities for the opposition.

2. Amount of Goals/Points doesn't necessarily have a direct correlation with value. There are plenty of NHL players that score a lot at the expense of defense. And if you give up more than you create then you're hurting your team even if you're one of the top scorers. A player's value has to correlate with how much they help you outscore the opposition rather than just how much they help you score. For example, Kessel might be one of the top even strength scorers in the league this year but I don't think he's near the top for even strength contribution to winning. This of course ignores special teams but that's its own situation and should be considered somewhat independently from even strength play.

3. "Finishing" talent (likelihood of scoring with a given chance) does exist at the NHL level but a) its really hard to say for sure who actually has it and who is just "hot" or "cold." b) there doesn't seem to be as much of a spread on it, if you're good enough to make the NHL chances are you're part of the very large mass of players that aren't noticeably better or worse at it.

So for one thing its hard to say for sure if this sort of thing would undervalue a Brett Hull without having any real information about how it actually values Brett Hull. And if it did rate a top goal scorer like Hull low that should also lead to some consideration on how much value he does provide.
My main point is that it's dangerous to simply look at a spreadsheet, look at puck possession and then make a decision on who to build with. I don't know about Hull, but we have seen players who don't produce with high puck possession stats. If you get caught up in looking at puck possession and measuring players that way, you miss out on productive players.

Some folks are arguing that because he shot the puck alot he was a puck possession guy. Maybe but maybe not. Hull was the king of the one timer and was always a 'shoot first' type guy. Gomez is the opposite... hangs onto the puck looking to pass to someone. Hull would shoot the puck a lot and had a pretty good shot to boot. Shot totals and goals certainly matter when you're looking at how effective a scorer a player is going to be.
Quote:
Originally Posted by Kjell Dahlin View Post
True but I tried a little test...

You mentioned in a previous post: "... Size and grit ARE attributes that GMs simply can't ignore. And I'm not sure how you capture this via analytics...". I suggested the following parameters (hits, PIM, TkA-GvA) to answer this question:

--------------
Is player A grittier than player B.
--------------

I used raw data: no regression, I did not run tests to exclude exterior factors, no kNN, heck I did not even use the MS Excel solver! Nothing: just raw data.

--------------
Player's grit = f (hits, PIM, TkA-GvA)
--------------

So... someone who never watched a game would rank our players (from a grit perspective), that way:

Ryan White 1,00
Mike Blunden 0,83
Brad Staubitz 0,81
Erik Cole 0,78
Louis Leblanc 0,76
Blake Geoffrion 0,73
Travis Moen 0,71
Rene Bourque 0,70
Lars Eller 0,70
Mathieu Darche 0,69
Brian Gionta 0,68
Max Pacioretty 0,68
Aaron Palushaj 0,65
David Desharnais 0,64
Petteri Nokelainen 0,64
Tomas Plekanec 0,60
Scott Gomez 0,58

--------------------------------------------------------------------

Alexei Emelin 0,69
Josh Gorges 0,60
P.K. Subban 0,50
Yannick Weber 0,50
Raphael Diaz 0,49
Tomas Kaberle 0,47
Chris Campoli 0,47

I separated the forwards and the defensemen because I did not correct the inherent worst TkA-GvA for the Ds. I removed Markov, Saint-Denis and Engqvist because they did not play enough games.

Conclusion?

Despite the numerous flaws (subjectivity, inconsistency in the way the data is collected...) associated with the set of data provided by nhl.com, the results are not bad: someone who never watched a game would not look foolish when ranking our players from a grit perspective. It took me 5 minutes to do this and again: I used raw data - no regression, I did not run tests to exclude exterior factors... If I did this by the book, the results would be more precise.


PS I know (thanks to Mathletic and MathMan!) that "... on average the raw stat of hits for is negatively correlated with winning...". This little test was just an attempt to answer this simple question: is player A grittier than player B.
Those results aren't that bad actually. Now how do you weigh 'grit' vs. things like scoring, puck possession, defensive ability etc? And how does chemistry between two players factor in here?


Last edited by Lafleurs Guy: 04-01-2012 at 06:39 PM.
Lafleurs Guy is offline   Reply With Quote
Old
04-02-2012, 12:50 AM
  #55
Kjell Dahlin
Registered User
 
Kjell Dahlin's Avatar
 
Join Date: Jan 2010
Location: Québec, Québec
Posts: 1,998
vCash: 500
Quote:
Originally Posted by Lafleurs Guy View Post

(...)

Those results aren't that bad actually. Now how do you weigh 'grit' vs. things like scoring, puck possession, defensive ability etc? And how does chemistry between two players factor in here?
I know it was a rhetorical question but here I go anyhow!

I like the Bill James' quote provided by Roke (post#12): "... So if we can't tell who the good fielders are accurately from the record books, and we can't tell accurately from watching, how can we tell? By counting things."

The first step would be to find a way to count an abstract reality. For instance we would need to transform "defensive ability" in a countable thing. A little bit like I did with "grit". Heck Mathletic mentioned that even the vaporous concept of "intangible" may become quantitatively measurable! Extract: "... There have been other advances in quantification of "intangibles". I remember a conference at the MIT Sloan Conference on a group of researchers who have analyzed speech of different players in order to predict future behaviour as well as another team measuring and assessing various leadership skills. For instance, players who touch teammates more than others like Garnett IIR ...".

As mentioned by Mathman, the context (quality of opposition, situation on the ice...) is important but, given the appropriate budget, it's only a matter of men/women power imo.

As for "... Now how do you weigh 'grit' vs. things like scoring, puck possession, defensive ability etc?", once it's a reality you can count, there are plenty of tools/tests (ANOVA, kNN*1, average tests, regression...) you can run to see how strong the connection is between a given parameter (example: scoring) and the measured variable – here the variable would be "winning %". Based on the strength of the connection, a weight is given to each statistically significant parameters.

The model would look like this: V = f(B1, B2, ..., Bn) where "V" would be the variable "winning%".

As for "... And how does chemistry between two players factor in here?". We are chatting about a tool designed to help the decision makers here (chemistry can be "easily" observed - the decison makers are still part of the process!) and I saw very effective models (in my line of work; not hockey related) with less than 3 parameters. I think it would be possible to quantify "chemistry between two players" but I am not sure it would be relevant. You think it can become significant?


*1 I never used or studied kNN (k Nearest Neighbours) but I often read new researches that use it; I think it could be applied to hockey.


Last edited by Kjell Dahlin: 04-02-2012 at 01:00 AM.
Kjell Dahlin is offline   Reply With Quote
Old
04-02-2012, 12:27 PM
  #56
CGG
Registered User
 
Join Date: Jan 2005
Location: 416
Country: Canada
Posts: 3,323
vCash: 500
Interesting conversation. I know Billy Beane and the A’s pulled out all sorts of detailed micro stats, but what I took away from the book is far more general – finding “value” players, that can contribute more than their salary would suggest.

The A’s basically dumbed everything down to on-base percentage (and to a lesser extent, slugging percentage) to figure out how productive a player would be, ignoring many other things like stolen bases, speed, age, appearance, etc. They did a great job of finding “defective” players that no one else wanted but could still produce.

Most interestingly, they figured out that everything boiled down to runs for and runs against – that losing a guy like Damon might mean their defense allowed 15 more runs in a season, but you didn’t need to replace him with an $8 million defensively sound centrefielder, it would be much easier to go out and get 15 more runs yourselves.

They replaced Damon (below average OBP), Giambi (great OBP) and some other guy (below average OBP) with three completely unrelated cheaper guys who all had moderately decent OBP’s but were flawed in some way – basically none of them could play defense. But the increased offense they would create would help offset their terrible defense.

This is where this sort of thing can come into hockey – flawed players that no one else wants can still contribute to wins, and do so at a very cheap price. In Montreal’s case it isn’t a self-imposed budget, it’s a salary cap. Players like Marc-Andre Bergeron and David Desharnais fit this criteria perfectly – they both put up points, and are both cheap, despite being “flawed” – in this case, too small for the average hockey team / scout. Pierre Dagenais was another good example – terrible hockey player, but okay on the PP and terrific in shootouts. Ignoring shootout proficiency is just one of many problems this year’s Montreal Canadiens have gotten themselves into.

Gainey tried to replace the offense lost by Kovalev-Koivu-Tanguay-Ryder with simple replacements in Cammalleri-Gomez-Gionta and to a lesser extent, Moen. The problem is, the offense lost wasn’t worth the $19.8 million those 4 new forwards were making. They could have got much more value out of keeping Koivu and Ryder (and Higgins) instead of getting Gomez. Lesser known free agents, increased roles for younger guys, undrafted European guys, and keeping Streit on the blue line could have helped fill the offensive void left by Kovalev and Tanguay and/or helped reduce the goals against. Adding a solid defensive third line center instead of using Lapierre or Metropolit might have also helped reduce goals against, freeing up Plekanec from having to kill penalties.

This is what I want the new GM to do – find guys who actually produce more than their salaries would indicate. We’ve overpaid enough guys. Rookies and young guys on entry level contracts fit the bill (i.e. Subban). So do Desharnais and Pacioretty for next year. Role players who do a few things well (like kill penalties or take face offs) even if they’re not “complete”. Get a few more cheap talented players, then supplement with high priced free agents to fill up your salary cap room. Don’t be afraid of old guys either – it makes more sense to pay a 36 year old decent winger as a stop gap for 1 year than to sign a Gionta for 5 years.

Analyze draft picks too – figure out the chances of a 4th round pick making the NHL. Don’t be afraid to use picks wisely, for example trading one for the rights to a UFA before July 1. We certainly have enough picks coming in the next 2 years, might as well use a few of them for sure things.

CGG is offline   Reply With Quote
Old
04-02-2012, 05:52 PM
  #57
MathMan
Registered User
 
Join Date: Jan 2006
Posts: 17,097
vCash: 500
A general rule on "things that can't be measured": if it has an impact on the game then the impact can be measured. If something is so difficult to detect to the point that it has no measurable impact on the game, then by definition it's not worth worrying about because it doesn't change anything.

On the subject of "chemistry": statheads will happily isolate performance for various combos of players and given players X and Y, will often come up with charts describing the performance of X with Y, X without Y, and Y without X.

A personal opinion: the notion of "chemistry" as understood by most people is somewhat overrated. There might be value in line stability (probably worth a study), but I think it's unlikely that a unit comprised of X+Y+Z will be significantly better together than it would be by replacing Z with an equivalent player with less "chemistry". This is not to say that such examples won't exist, but I don't think it is nearly as important as it is described.

Anecdotal example: everyone thought Kovalev and Tanguay couldn't play together because both were puck hogs, yet somehow they made it work extremely well. Another anecdotal example: Henrik Sedin having his career-best production the year that Daniel misses the most games.

This would not be a bad subject for a study, but unfortunately since the proponents of "chemistry" see it as an objections to analytics rather than a subject of study, they're unlikely to want to meet the burden of proof that should be there in this situation.

Which brings me to a bit of commentary about the burden of proof, actually: it is generally the case, in these discussions, that the people supporting analytics will show the research and all the factors that have been shown to matter. Then the people denying analytics will say "what about X, Y, and Z", purporting these factors to be either immeasurable or not covered by the model.

Assuming X, Y and Z haven't already been covered (as is most often the case), it does not follow that the burden of disproving X, Y and Z as factors should be on analytics; the way it works is that if you want to assert some factor is significant, then you are the one who needs to prove its significance, so if you want to claim that, say, "heart" is important, you need first to define "heart" in some measurable fashion (ideally objectively) and then demonstrate that teams that have it are measurably more successful than teams that don't, everything else being equal.

IOW, it's generally backwards to say "you must prove that X doesn't matter". The proper approach should be to prove that it does.

This is, in a nutshell, what hockey analytics try to do -- prove which factors show sustained success, and which factors either don't matter or aren't controllable by the players/teams.

MathMan is offline   Reply With Quote
Old
04-03-2012, 12:11 AM
  #58
Lafleurs Guy
Registered User
 
Lafleurs Guy's Avatar
 
Join Date: Jul 2007
Posts: 23,244
vCash: 500
Quote:
Originally Posted by MathMan View Post
A general rule on "things that can't be measured": if it has an impact on the game then the impact can be measured. If something is so difficult to detect to the point that it has no measurable impact on the game, then by definition it's not worth worrying about because it doesn't change anything.
Who's to say it didn't? Just because it can't be analyzed via analytics doesn't mean that the effect is not there. Again, you can't distinguish the force of a hit from Zedeno Chara. When you measure hits in hockey all hits are created equal and anyone who's played hockey KNOWS that it's not the case.

Until you can explain this, you're going to have a real problem getting people to accept that general rule of yours.
Quote:
Originally Posted by MathMan View Post
On the subject of "chemistry": statheads will happily isolate performance for various combos of players and given players X and Y, will often come up with charts describing the performance of X with Y, X without Y, and Y without X.

A personal opinion: the notion of "chemistry" as understood by most people is somewhat overrated. There might be value in line stability (probably worth a study), but I think it's unlikely that a unit comprised of X+Y+Z will be significantly better together than it would be by replacing Z with an equivalent player with less "chemistry". This is not to say that such examples won't exist, but I don't think it is nearly as important as it is described.

Anecdotal example: everyone thought Kovalev and Tanguay couldn't play together because both were puck hogs, yet somehow they made it work extremely well. Another anecdotal example: Henrik Sedin having his career-best production the year that Daniel misses the most games.

This would not be a bad subject for a study, but unfortunately since the proponents of "chemistry" see it as an objections to analytics rather than a subject of study, they're unlikely to want to meet the burden of proof that should be there in this situation.

Which brings me to a bit of commentary about the burden of proof, actually: it is generally the case, in these discussions, that the people supporting analytics will show the research and all the factors that have been shown to matter. Then the people denying analytics will say "what about X, Y, and Z", purporting these factors to be either immeasurable or not covered by the model.

Assuming X, Y and Z haven't already been covered (as is most often the case), it does not follow that the burden of disproving X, Y and Z as factors should be on analytics; the way it works is that if you want to assert some factor is significant, then you are the one who needs to prove its significance, so if you want to claim that, say, "heart" is important, you need first to define "heart" in some measurable fashion (ideally objectively) and then demonstrate that teams that have it are measurably more successful than teams that don't, everything else being equal.

IOW, it's generally backwards to say "you must prove that X doesn't matter". The proper approach should be to prove that it does.

This is, in a nutshell, what hockey analytics try to do -- prove which factors show sustained success, and which factors either don't matter or aren't controllable by the players/teams.
Except that's not really what anyone is arguing about. We all know what analytics is trying to do. Nobody has a problem with that. The question is... what are it's limitations? You've said if something can't be measured it doesn't have value and I'd say that you're wrong.

Those 'intangibles' may show up somewhere else like in the win column but there are so many variables it's impossible to know. Last season Tim Thomas had the highest save percentage in the modern era. How much of that was due to players not wanting to get their heads kicked in by going to the net? You can always measuer how many times teams went to the net but you CAN'T measure how many times they DIDN'T go to the net for fear of getting smoked. Maybe it's a big reason why they won the games they did. There's just too many variables for us to be able to measure it. That doesn't mean that we can't observe what happens on the ice and come to a conclusion that size, grit and tough play CAN have a positive effect on winning games.

Lafleurs Guy is offline   Reply With Quote
Old
04-03-2012, 12:24 AM
  #59
Ohashi_Jouzu
Registered User
 
Ohashi_Jouzu's Avatar
 
Join Date: Apr 2007
Location: Halifax
Country: Japan
Posts: 22,862
vCash: 500
Quote:
Originally Posted by Lafleurs Guy View Post
Who's to say it didn't? Just because it can't be analyzed via analytics doesn't mean that the effect is not there. Again, you can't distinguish the force of a hit from Zedeno Chara. When you measure hits in hockey all hits are created equal and anyone who's played hockey KNOWS that it's not the case.

Until you can explain this, you're going to have a real problem getting people to accept that general rule of yours.

Except that's not really what anyone is arguing about. We all know what analytics is trying to do. Nobody has a problem with that. The question is... what are it's limitations? You've said if something can't be measured it doesn't have value and I'd say that you're wrong.

Those 'intangibles' may show up somewhere else like in the win column but there are so many variables it's impossible to know. Last season Tim Thomas had the highest save percentage in the modern era. How much of that was due to players not wanting to get their heads kicked in by going to the net? You can always measuer how many times teams went to the net but you CAN'T measure how many times they DIDN'T go to the net for fear of getting smoked. Maybe it's a big reason why they won the games they did. There's just too many variables for us to be able to measure it. That doesn't mean that we can't observe what happens on the ice and come to a conclusion that size, grit and tough play CAN have a positive effect on winning games.
As someone who has just been sitting back and reading so far, I'd say this post puts you "in the lead" right now. Still think that there are plenty of microstats whose quantity, regardless of "quality", which could/should be tracked if the resultant "rankings" seem even halfway consistent with "educated" opinion or "eye tests". For example, I agree with you about not all hits being equal, but even the stat itself says something about the frequency that the player initiates contact. I think that does say at least something about a player's "physicality", but you're right that you have to see the impact/result to gauge for yourself.

Basically, I think that advanced microstats in the hands of the right hockey mind would result in pretty reliable decision-making when it comes to line/roster adjustments. Used by themselves by a hockey illiterate, though, and I bet "success" would be more hit/miss or 50/50, because there are always statistical red herrings and impossibly complex relationships in the mix.

Ohashi_Jouzu is offline   Reply With Quote
Old
04-03-2012, 01:04 AM
  #60
Teufelsdreck
Registered User
 
Join Date: Sep 2005
Posts: 14,753
vCash: 500
An "error" in hockey gan be negated in several ways: good goaltending, support from a teammate, offsides, penalties away from the play, and just plain poor shooting,

Teufelsdreck is offline   Reply With Quote
Old
04-03-2012, 01:05 AM
  #61
coolasprICE
Registered User
 
coolasprICE's Avatar
 
Join Date: Mar 2008
Location: Montreal
Country: Canada
Posts: 9,491
vCash: 500
Andrei Kostitsyn agrees with this thread

coolasprICE is offline   Reply With Quote
Old
04-03-2012, 01:16 AM
  #62
Lafleurs Guy
Registered User
 
Lafleurs Guy's Avatar
 
Join Date: Jul 2007
Posts: 23,244
vCash: 500
Quote:
Originally Posted by Ohashi_Jouzu View Post
As someone who has just been sitting back and reading so far, I'd say this post puts you "in the lead" right now. Still think that there are plenty of microstats whose quantity, regardless of "quality", which could/should be tracked if the resultant "rankings" seem even halfway consistent with "educated" opinion or "eye tests". For example, I agree with you about not all hits being equal, but even the stat itself says something about the frequency that the player initiates contact. I think that does say at least something about a player's "physicality", but you're right that you have to see the impact/result to gauge for yourself.

Basically, I think that advanced microstats in the hands of the right hockey mind would result in pretty reliable decision-making when it comes to line/roster adjustments. Used by themselves by a hockey illiterate, though, and I bet "success" would be more hit/miss or 50/50, because there are always statistical red herrings and impossibly complex relationships in the mix.
I think analytics/microstats can be useful in looking at players in a different light. I would never dispute this as I like what James has done in baseball. We SHOULD be investing in this kind of analysis.

My problem is the weight that these stats are given by some on this board. Some see it as the be/all end/all and I don't think that's the case. Saying a player is great because he's got puck possession skills doesn't necessarily translate to him being a good player and I think Gomez is a good example of this. Brett Hull I suspect would be a good example of it as well if we had the stats to look at with him too.

I also don't agree with the argument that if something isn't measurable it's not important... Just because you don't know how to measure a factor in winning doesn't mean that it's not there.

The Chara hits example is a good one.

Look at Chris Pronger. Guy took 3 teams to the finals in consecutive seasons and some of those clubs had no business being there. His offense plays a part but he also happens to be the meanest blueliner this side of Chara. Is it coincidence that the Oilers get to the finals (let alone make the playoffs) with him in the lineup? What are the characteristics of Pronger that make him so good? Is it just points or are there intangibles to his game that are difficult to measure? He's not Paul Coffey so it's not just offense that gets the Oilers there. Could it be that opponents are more hesitant to behave the way they normally would but are intimidated just slightly enough that they aren't as effective?

You can say small sample size and luck all you want but Pronger was a huge reason why those clubs did so well. And I don't see how anyone who's seen him play would argue that intimidation isn't a big part of his game.

Lafleurs Guy is offline   Reply With Quote
Old
04-03-2012, 01:32 AM
  #63
Mathletic
Registered User
 
Mathletic's Avatar
 
Join Date: Feb 2002
Location: St-Augustin, Québec
Country: Canada
Posts: 12,235
vCash: 500
Quote:
Originally Posted by Lafleurs Guy View Post
I also don't agree with the argument that if something isn't measurable it's not important... Just because you don't know how to measure a factor in winning doesn't mean that it's not there.
I'm not sure you understand what is meant. The fact Chara hits hard and makes players think twice before going to the net is implicitly measured by his defensive rating. The rating itself doesn't say this guy hits that hard while this other player hits that hard. However, the rating takes into consideration that a player won't drive the net because Chara is on the ice.

Now, you may break defensive rating into smaller parts like blocked shots, hits, not turning the puck over and so on. In the end, the effectiveness of the hits, giveaways and so on are swallowed by the defensive rating itself.

If a certain aspect of player's play is not measured by his defensive rating, then it holds no importance to defense itself.

From a logical point of view you can see this as defense being made up of hits, chemistry with linemate, blocking shots and all you want. If what you think is part of defense is not measured by the defensive rating, then it can't be part of the defensive skillset of a player.

For Pronger, it's the same story. If intimidation actually works then it will be measured by his defensive rating. It doesn't mean that the defensive rating measures intimidation in of itself but measures it implicitly.

Mathletic is offline   Reply With Quote
Old
04-03-2012, 01:39 AM
  #64
Kjell Dahlin
Registered User
 
Kjell Dahlin's Avatar
 
Join Date: Jan 2010
Location: Québec, Québec
Posts: 1,998
vCash: 500
Quote:
Originally Posted by Lafleurs Guy View Post
I think analytics/microstats can be useful in looking at players in a different light. I would never dispute this as I like what James has done in baseball. We SHOULD be investing in this kind of analysis.

My problem is the weight that these stats are given by some on this board. Some see it as the be/all end/all and I don't think that's the case. Saying a player is great because he's got puck possession skills doesn't necessarily translate to him being a good player and I think Gomez is a good example of this. Brett Hull I suspect would be a good example of it as well if we had the stats to look at with him too.

I also don't agree with the argument that if something isn't measurable it's not important... Just because you don't know how to measure a factor in winning doesn't mean that it's not there.

The Chara hits example is a good one.

Look at Chris Pronger. Guy took 3 teams to the finals in consecutive seasons and some of those clubs had no business being there. His offense plays a part but he also happens to be the meanest blueliner this side of Chara. Is it coincidence that the Oilers get to the finals (let alone make the playoffs) with him in the lineup? What are the characteristics of Pronger that make him so good? Is it just points or are there intangibles to his game that are difficult to measure? He's not Paul Coffey so it's not just offense that gets the Oilers there. Could it be that opponents are more hesitant to behave the way they normally would but are intimidated just slightly enough that they aren't as effective?

You can say small sample size and luck all you want but Pronger was a huge reason why those clubs did so well. And I don't see how anyone who's seen him play would argue that intimidation isn't a big part of his game.
I need to go to bed so I will limit myself to this part "... I also don't agree with the argument that if something isn't measurable it's not important... Just because you don't know how to measure a factor in winning doesn't mean that it's not there...".

That's not exactly what Mathman mentioned.

Quote: "A general rule on "things that can't be measured": if it has an impact on the game then the impact can be measured. If something is so difficult to detect to the point that it has no measurable impact on the game, then by definition it's not worth worrying about because it doesn't change anything...".

I see it that way: if you test a 5 parameters model and its results are 94% in sync with reality, then there is no point in adding 12 others parameters to reach a 98% level of conformity.

As I mentioned in regard to your comment about "chemistry between two players", some concepts may be harder to quantify but maybe some of these concepts are not statistically significant. I will partially quote Mathman: "... If something is so difficult to detect to the point that it has no measurable impact on the game…" (end quote) then chances are it is not statistically significant.

Kjell Dahlin is offline   Reply With Quote
Old
04-03-2012, 02:52 AM
  #65
Ohashi_Jouzu
Registered User
 
Ohashi_Jouzu's Avatar
 
Join Date: Apr 2007
Location: Halifax
Country: Japan
Posts: 22,862
vCash: 500
Quote:
Originally Posted by Kjell Dahlin View Post
I need to go to bed so I will limit myself to this part "... I also don't agree with the argument that if something isn't measurable it's not important... Just because you don't know how to measure a factor in winning doesn't mean that it's not there...".

That's not exactly what Mathman mentioned.

Quote: "A general rule on "things that can't be measured": if it has an impact on the game then the impact can be measured. If something is so difficult to detect to the point that it has no measurable impact on the game, then by definition it's not worth worrying about because it doesn't change anything...".

I see it that way: if you test a 5 parameters model and its results are 94% in sync with reality, then there is no point in adding 12 others parameters to reach a 98% level of conformity.

As I mentioned in regard to your comment about "chemistry between two players", some concepts may be harder to quantify but maybe some of these concepts are not statistically significant. I will partially quote Mathman: "... If something is so difficult to detect to the point that it has no measurable impact on the game…" (end quote) then chances are it is not statistically significant.
While I find stats interesting, I completely agree with the bolded part. Having said that, I think Lafleur Guy's response would be similar to my own. How do you determine the 94/98% conformity? With the same kind of formula that treats all saves, goals, passes, and hits equally? Also, how flexible is such a system when it comes to dealing with players whose key attributes are intangibles (use the "intimidation" example again, I guess, but options include "leadership", "experience", "clutchness", etc), moreso than the absolute frequency with which they generate various countable/measurable results (and/or their combinations explored)?

Or another way to put it perhaps. Is there something that you can't count that possibly contributes significantly to the opponents' measurable events? Put back into (slightly verbose) English, how do you figure out what combination of stats you can count and combine to "sufficiently" describe the observable fact that Chara's size, strength, and willingness to be physical severely limits the number of quality scoring chances that opposing teams typically get in certain areas when he's on the ice, for example?

Ohashi_Jouzu is offline   Reply With Quote
Old
04-03-2012, 03:19 AM
  #66
Mathletic
Registered User
 
Mathletic's Avatar
 
Join Date: Feb 2002
Location: St-Augustin, Québec
Country: Canada
Posts: 12,235
vCash: 500
Quote:
Originally Posted by Ohashi_Jouzu View Post
While I find stats interesting, I completely agree with the bolded part. Having said that, I think Lafleur Guy's response would be similar to my own. How do you determine the 94/98% conformity? With the same kind of formula that treats all saves, goals, passes, and hits equally? Also, how flexible is such a system when it comes to dealing with players whose key attributes are intangibles (use the "intimidation" example again, I guess, but options include "leadership", "experience", "clutchness", etc), moreso than the absolute frequency with which they generate various countable/measurable results (and/or their combinations explored)?

Or another way to put it perhaps. Is there something that you can't count that possibly contributes significantly to the opponents' measurable events? Put back into (slightly verbose) English, how do you figure out what combination of stats you can count and combine to "sufficiently" describe the observable fact that Chara's size, strength, and willingness to be physical severely limits the number of quality scoring chances that opposing teams typically get in certain areas when he's on the ice, for example?
not that your message is addressed to me but I'll answer anyways. I'll try to answer in general terms as much as possible and not go too technical.

There are statistical tests that tell you how much you explain what you are studying. In hockey we are mostly studying what explains goals for and goals against. GF and GA basically explain roughly 96% ... IIRC ... a team's record. For example, you can plug GF and GA in the pythagorean formula

http://en.wikipedia.org/wiki/Pythagorean_expectation


and on average you get close to 96% to the team's actual record.

Whatever ratings will basically work from top down if I can put it this way ... whether it's a win shares system, point contribution, WAR, adjusted +/- and so on ... They will calculate how much of an impact a player had, say defensively, on his team compared to a replacement player. That is, how much better is Chara defensively compared to a border line AHL/NHL d-man for example.

Ratings, in general, don't compute that a guy made 65 hits so he gets that much credit for defense plus 105 blocked shots, so he gets that many more credits and so on.

So, in general, a defensive rating will implicitly include intimidation for example or whatever intangible you may think of. Since, if Chara actually scares player to the point that they won't drive the net as much than if player X was on the ice ... who's borderline AHL/NHL ... instead of Chara then it will reflect in his defensive stats whether rates of shots/against, goals/against, adjusted +/- or whatever system you use.

Also, whatever intangible you may think of, you have to first sit down and think of what it really is. Simply saying well that thing experience can't be measured, so analytics don't work won't lead you far.

I'll take experience as an example. Exerience can be measured. The n' of years a player has been in a league, how many minutes he's played with a teammate, how many playoffs games he has played, so on and so forth.

That said, you have to approach analytics with an open-mind. If you think clutchness is a huge factor in a game, you might be disappointed or come out of your reflection with a different understanding of clutchness if you stop and think about what it really is.

I'll give you an example. On saturday afternoon Kobe Bryant and the Lakers were playing New Orleans. Kobe was terrible all day long. He went for something like 5-25. However, the score was something like 94-92 Hornets with 30 seconds left. Kobe takes the ball and shoots a 3 to win the game for the Lakers. Was that a clutch performance? Had Kobe had just an average game, the Lakers would have won easily by that point. Inversely, a player who would have had a great game and whose team shouldn't even be in the game but misses the last shot. Do you consider that as a clutch performance?

In general, analytics will tell you that clutch plays a small part of the game. If that, it probably doesn't even work the way you think it does. The home team on average is at a disadvantage in clutch situations. On average, players hit fewer money shots with a game tied at home than they do on the road.

Also, the sample size to determine who's clutch and who's not is so damn small that it's pretty much impossible to determine who's clutch and who's not. You may run psychological tests for that and whatnot but at that level, it's much more important to use a player who'll bring you to the final game than it is to get a player who might make that final shot more easily but nowhere near good enough to help you get there. Also, in general studies show that better players are simply more clutch.

Mathletic is offline   Reply With Quote
Old
04-03-2012, 08:28 AM
  #67
Talks to Goalposts
Registered User
 
Join Date: Apr 2011
Location: Edmonton
Country: Canada
Posts: 3,641
vCash: 500
Quote:
Originally Posted by Ohashi_Jouzu View Post
While I find stats interesting, I completely agree with the bolded part. Having said that, I think Lafleur Guy's response would be similar to my own. How do you determine the 94/98% conformity? With the same kind of formula that treats all saves, goals, passes, and hits equally? Also, how flexible is such a system when it comes to dealing with players whose key attributes are intangibles (use the "intimidation" example again, I guess, but options include "leadership", "experience", "clutchness", etc), moreso than the absolute frequency with which they generate various countable/measurable results (and/or their combinations explored)?

Or another way to put it perhaps. Is there something that you can't count that possibly contributes significantly to the opponents' measurable events? Put back into (slightly verbose) English, how do you figure out what combination of stats you can count and combine to "sufficiently" describe the observable fact that Chara's size, strength, and willingness to be physical severely limits the number of quality scoring chances that opposing teams typically get in certain areas when he's on the ice, for example?

I think you've got a completely backwards. The hockey stats work generally has been done backwards from finding out what creates wins. The current thinking goes wins are from goal differential and goal differential is driven by differential in scoring chances + goaltending. And scoring chance differential correlates very strongly with fenwick (shots+missed shots) differential which also correlates strongly with puck possession time whenever that has been measured.

So to look at Chara by this kind of thinking is to largely not care if he prevent scoring
chances through physical strength, intimidation or puck skills. The fact that he does it is sufficient for our purposes. And Chara is consistently one of the few defenders that has a very strong corsi relative to his team while playing really tough minutes.

This sort of thing is pretty much why, beyond the fact that the real-time stats are tracked terribly, not much of microstats people care about things like hits.


Also nit-picking on edge cases is missing another point. As a system the new stats don't need to be perfect to be useful. They just have to be a better description of hockey than older methods. And looking at things like corsi and quality of competition is a lot better than relying on the older stuff. Traditional plus minus is a joke and counting goals and assists as a mark of value has way more problems than looking at shot differentials.

Talks to Goalposts is offline   Reply With Quote
Old
04-03-2012, 09:05 AM
  #68
HH
GO HABS GO!
 
HH's Avatar
 
Join Date: Mar 2005
Location: Montreal
Country: Canada
Posts: 17,302
vCash: 500
Lol crazy I see this thread after just seeing Moneyball.

HH is offline   Reply With Quote
Old
04-03-2012, 09:53 AM
  #69
onice
Registered User
 
onice's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 5,718
vCash: 500
Quote:
Originally Posted by MathMan View Post
This actually brings to mind another point that, I must admit, occasionally bugs me.

If one can think of an objection to microstats within thirty seconds (what about shot quality? what about teammates? what about those guys who always play against the top lines?), that doesn't give leave to dismiss the whole thing. The odds are very good that the objection has already been thought of by analysts, examined, and either had a metric introduced or was dismissed as not making enough difference to matter on aggregate.
Microstats are a tool like other tools and putting too much emphasis on them blinds you to the game.

You are a perfect example of that. Your stats told you (they probably still tell you) that Gomez is playing well and - how did you put it - oh yeah, he has been unlucky. Unlucky, how do you calculate unlucky?

This is nonsense that the boys with the computers and spreedsheets come up with when their stats don't explain what is happening in the game.

Taking a tool from baseball and trying to apply it to hockey is utter insanity. The two games share nothing. They have nothing in common.

I can understand using microstats for broad strokes, for another point of view and I would leave it at that. Microstats are simply a way of putting a scientific facade on intuition.

A perfect example of what can go wrong with going gaga over microstats is you explaining the discrepancy between your stats & Gomez's putrid results with that scientific concept: luck.

onice is offline   Reply With Quote
Old
04-03-2012, 10:23 AM
  #70
coolasprICE
Registered User
 
coolasprICE's Avatar
 
Join Date: Mar 2008
Location: Montreal
Country: Canada
Posts: 9,491
vCash: 500
Quote:
Originally Posted by onice View Post
Microstats are a tool like other tools and putting too much emphasis on them blinds you to the game.

You are a perfect example of that. Your stats told you (they probably still tell you) that Gomez is playing well and - how did you put it - oh yeah, he has been unlucky. Unlucky, how do you calculate unlucky?

This is nonsense that the boys with the computers and spreedsheets come up with when their stats don't explain what is happening in the game.

Taking a tool from baseball and trying to apply it to hockey is utter insanity. The two games share nothing. They have nothing in common.

I can understand using microstats for broad strokes, for another point of view and I would leave it at that. Microstats are simply a way of putting a scientific facade on intuition.

A perfect example of what can go wrong with going gaga over microstats is you explaining the discrepancy between your stats & Gomez's putrid results with that scientific concept: luck.
zing.

His words precisely were that Gomez is destined to return to the 55-60 point mark. Those whom stick to a singular and rigid method of evaluation and speculation will end-up looking silly.

It's like the analyzer who is solely relying on observant evaluation. Doomed for failure just as much as those who rely solely on spreadsheets.

coolasprICE is offline   Reply With Quote
Old
04-03-2012, 10:57 AM
  #71
MathMan
Registered User
 
Join Date: Jan 2006
Posts: 17,097
vCash: 500
Quote:
Originally Posted by Lafleurs Guy View Post
Who's to say it didn't? Just because it can't be analyzed via analytics doesn't mean that the effect is not there.
Admittedly, analytics are only concerned about things that help win games, not things that make the players or the fans feel better (unless the players feeling better also leads to winning more games, which is measurable).

On the other hand, if player X has "intangibles" that help win hockey games, then the effects of the intangibles will be measured, if not the intangibles themselves. Analytics are not terribly concerned about style; they are concerned about effects. Whether Chara achieves defense via physicality or positioning is not relevant; what is relevant is that defense is achieved and defense is measurable (and that is certainly not "intangible").

I repeat myself, but this is important and it is actually unrelated to analytics: "intangibles" that do not actually have an impact on winning games that is significant enough to be noticeable might as well not exist. And intangibles that do have an impact will be measured as part of that player's performance.

It's also significant that there seems to be a fair bit of waffle about the definition of "intangible". Is it merely stuff like "leadership" and "good in the room", or are we including stuff that does not fall under that traditional heading such as "Chara hits harder than Desharnais", which really have more to do with playing style than intangibles?

Quote:
Originally Posted by Lafleurs Guy View Post
Those 'intangibles' may show up somewhere else like in the win column but there are so many variables it's impossible to know. Last season Tim Thomas had the highest save percentage in the modern era. How much of that was due to players not wanting to get their heads kicked in by going to the net?
This is measurable by examining shots positions. To my knowledge however, the Bruins did not have an inordinate percentage of perimeter shots relative to "dangerous" shots.

MathMan is offline   Reply With Quote
Old
04-03-2012, 11:02 AM
  #72
MathMan
Registered User
 
Join Date: Jan 2006
Posts: 17,097
vCash: 500
Quote:
Originally Posted by onice View Post
You are a perfect example of that. Your stats told you (they probably still tell you) that Gomez is playing well and - how did you put it - oh yeah, he has been unlucky. Unlucky, how do you calculate unlucky?
Very easily. When uncontrolled variance results in lower-than-expected performance then we call it "unlucky" -- as opposed to when lower-than-expected performance is due to factor the individual controls.

A basic understanding of probabilities is pretty much necessary for analytics. The science of "luck" is really quite precise, at least in the abstract -- the practical question of how much "luck" there is in hockey has a margin of error, of course, but it's probably higher than most people realize or are willing to admit.

Of course, a big part of the problem is that "luck" in the mathematical sense is a somewhat different concept than "luck" in the almost-pejorative sense commonly used by hockey fans. "Lucky" and "unlucky" has emotional connotations in general discourse, which is why there's a bit of a movement to call it something different like "variance". In the context of evaluating a player, though, there really is a surprising amount of things that player doesn't control and wall under either "context" or "variance". But variation of shooting percentages at either end, which has been shown to regress to the mean strongly, tends to be the bulk of "variance".


Last edited by MathMan: 04-03-2012 at 11:38 AM.
MathMan is offline   Reply With Quote
Old
04-03-2012, 11:34 AM
  #73
onice
Registered User
 
onice's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 5,718
vCash: 500
Quote:
Originally Posted by MathMan View Post
Very easily. When uncontrolled variance results in lower-than-expected performance then we call it "unlucky" -- as opposed to when lower-than-expected performance is due to factor the individual controls.

A basic understanding of probabilities is pretty much necessary for analytics. The science of "luck" is really quite precise, at least in the abstract -- the practical question of how much "luck" there is in hockey has a margin of error, of course, but it's probably higher than most people realize or are willing to admit.

Of course, a big part of the problem is that "luck" in the mathematical sense is a somewhat different concept than "luck" in the almost-pejorative sense commonly used by hockey fans, which is why there's a bit of a movement to call it something different like "variance". In the context of evaluating a player, though, there really is a staggering amount of things that player doesn't control and aren't explained by context that would fall under "variance". But variation of shooting percentages at either end, which has been shown to regress to the mean strongly tends to be the bulk of it.

God, I wish there were a few GMs in the league who had your faith in this pseudo science: maybe our new GM could unload Gomez & Kaberle to them and get something worthwhile in return.

onice is offline   Reply With Quote
Old
04-03-2012, 11:51 AM
  #74
MathMan
Registered User
 
Join Date: Jan 2006
Posts: 17,097
vCash: 500
Quote:
Originally Posted by onice View Post
God, I wish there were a few GMs in the league who had your faith in this pseudo science: maybe our new GM could unload Gomez & Kaberle to them and get something worthwhile in return.
If Gauthier had had any notion of this, he wouldn't have traded for Kaberle, he wouldn't have fired Martin, he wouldn't have traded Cammalleri for a dud like Bourque, and the Habs would probably be in the playoffs right now.


Last edited by MathMan: 04-03-2012 at 12:02 PM.
MathMan is offline   Reply With Quote
Old
04-03-2012, 01:07 PM
  #75
Lafleurs Guy
Registered User
 
Lafleurs Guy's Avatar
 
Join Date: Jul 2007
Posts: 23,244
vCash: 500
Quote:
Originally Posted by Kjell Dahlin View Post
I know it was a rhetorical question but here I go anyhow!

I like the Bill James' quote provided by Roke (post#12): "... So if we can't tell who the good fielders are accurately from the record books, and we can't tell accurately from watching, how can we tell? By counting things."
I understand the philosophy... I do. But there are still limitations in what it shows.

Not sure if there are any philosophy students but I'll give an example of Immanuel Kant's 'veil of perception' here.

In a nutshell Kant argues that our senses (sight, touch, sound, taste, smell) allow us to perceive the world. Those senses are a gift in the sense that it allows the perceiver to process information out there outside of ourselves in the world. However, we are LIMITED to perceiving the world only via these senses. We can never know the 'thing in itself' beyond those five senses. Just as our senses give us the ability to perceive things in the real world, they limit us to only perceiving things within those parameters.

Imagine a person who is completely colour blind. This person would never know that an object was red or green... He wouldn't even know colours existed except that he's told by others that they do.

That's kind of what analytics is like. It gives you the ability to measure some things and draw conclusions from the data. But it's limited. It doesn't capture everything.

Moreover, there's the problem of 'counting things' themselves. There is subjectivity involved in what constitutes a takeaway or giveaway... one person may count it one way and for another person it's a completely different value set. That's why you see discrepencies from arena to arena. Those discrepencies can have a fairly significant effect on the data depending on who wide the discrepencies are. In some arenas almost everything is a takeaway, in others... they are barely counted at all. Who decides the standards here?

Relying solely on analytics is like being colour blind. You close yourself off from what's happening on the ice when you're only studying a spreadsheet. Mathman has argued that different people may argue over whether a certain player is effective or not and stats are what tell the story. Well... it depends on what stats you are looking at. Scott Gomez apparently has great puck possession skills. Well, I'd agree that he takes the puck and gains the opposing zone like a champ. But then nothing happens. I know this because I've WATCHED him. Apparently Gomez had a good defense rating too... well, I've seen him completely quit on plays as well. You could argue that the number of times where he's been defensively responsible is far more than the number of times that he wasn't effective... but he's QUIT on the play. I know this as well because I've seen it happen.

Someone above said it was a pet peeve of his when people said 'you can make stats say anything you want'... well, in some ways you can. Numbers are just that... raw data. But there's a whole slew of numbers you can look at. If Gomez isn't putting up points then I'd say that his CORSI doesn't matter to me a whole lot. At the end of the day, you have to produce. Totals DO matter. I get the whole 'luck' argument, but come on... at some point those totals matter. And Gomez is no longer a benefit to our club. That's why everyone can't wait to get rid of him.


Last edited by Lafleurs Guy: 04-03-2012 at 01:13 PM.
Lafleurs Guy 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 05:51 AM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"
Contact Us - HFBoards - Archive - Privacy Statement - Terms of Use - Advertise - Top - AdChoices

vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. ©2014 All Rights Reserved.