HFBoards

Go Back   HFBoards > General Hockey Discussion > The History of Hockey
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
Notices

The History of Hockey Relive great moments in hockey history and discuss how the game has changed over time.

Estimated Ice Time

Reply
 
Thread Tools
Old
07-30-2012, 01:06 PM
  #76
ssh
Registered User
 
Join Date: May 2008
Posts: 94
vCash: 500
Here's what the old model shows:
 TotESPPPK
Tremblay31.93924.9063.3933.641
Laperriere27.66923.0201.7682.880
Harris26.47423.2500.3472.878
Savard22.70421.4930.4740.737
Harper19.60518.8410.3770.386
Hillman13.68013.3470.0000.334
Lapointe7.3137.3130.0000.000

ssh is offline   Reply With Quote
Old
07-30-2012, 01:46 PM
  #77
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Details

Quote:
Originally Posted by ssh View Post
Here's what the old model shows:
 TotESPPPK
Tremblay31.93924.9063.3933.641
Laperriere27.66923.0201.7682.880
Harris26.47423.2500.3472.878
Savard22.70421.4930.4740.737
Harper19.60518.8410.3770.386
Hillman13.68013.3470.0000.334
Lapointe7.3137.3130.0000.000
Pairings were J.C. Tremblay / Ted Harris, Laperriere/mainly Savard-some Harper.

Rousseau on the PP balances the Harris shortage on the pairings.

Savard used to play a forward position at times on the PK, especially in 3 on 5 or 3 on 4 situations.

Canadiens1958 is online now   Reply With Quote
Old
07-30-2012, 02:07 PM
  #78
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Alternative

Quote:
Originally Posted by seventieslord View Post
what's your point? How does this disprove the relationship between the amount of goals for and against each player had, and the time they spent on the ice?

Equally valid estimates without relying on GF and GA. Will hold for the next two seasons although 1970-71 you have to adjust for the Canadiens going to a 3F/2D PP.

Canadiens1958 is online now   Reply With Quote
Old
07-30-2012, 02:24 PM
  #79
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 28,212
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Equally valid estimates without relying on GF and GA.
Why on earth would you ever think that?

seventieslord is offline   Reply With Quote
Old
07-30-2012, 02:39 PM
  #80
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Because......

Quote:
Originally Posted by seventieslord View Post
Why on earth would you ever think that?
Because replacing simple with complex increases the potential for error and generates extra work. You are welcome to your preference.

Canadiens1958 is online now   Reply With Quote
Old
07-30-2012, 04:36 PM
  #81
Bear of Bad News
Mod Supervisor
 
Bear of Bad News's Avatar
 
Join Date: Sep 2005
Location: San Angeles
Posts: 4,273
vCash: 663
Quote:
Originally Posted by Canadiens1958 View Post
Because replacing simple with complex increases the potential for error and generates extra work. You are welcome to your preference.
Okay. Now demonstrate the "equally valid estimates" claim. Please be rigorous.

Bear of Bad News is offline   Reply With Quote
Old
07-31-2012, 09:12 PM
  #82
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Ed Westfall played his first three NHL seasons as a defenseman:
Yes he did. But we're not talking about his first three NHL seasons.

Quote:
Originally Posted by Canadiens1958 View Post
My point is/was that HR is far from perfect in their classifications.
Agreed, which is why I don't blindly accept them.

I will also humbly suggest that your memory is also far from perfect in its classifications, especially when discussing specific seasons.

Quote:
Originally Posted by Canadiens1958 View Post
Iain never replied about the overall shortages - below 1216 games and overages - above 1216 games for the remaining teams during the 1968-69 season.
I have now. And they're irrelevant, because my method takes the total ice time available to a team's skaters and allocates it to each of the team's skaters. Whether or not the team used a full complement of skaters in any game is irrelevant, because the actual games played by the skaters are used. I don't know why anyone would do it any other way.

Quote:
Originally Posted by Canadiens1958 View Post
Not so, ice time without overtime is constant for skaters for a 76 game schedule with a slight extra for 6th attacker situations. Distributed over differing numbers < 1216 changes how the ice time is distributed amongst the players.
See above.

Quote:
Originally Posted by Canadiens1958 View Post
Equally valid estimates without relying on GF and GA. Will hold for the next two seasons although 1970-71 you have to adjust for the Canadiens going to a 3F/2D PP.
Such as? Please demonstrate, with rigour, as requested above.

Iain Fyffe is offline   Reply With Quote
Old
07-31-2012, 09:14 PM
  #83
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Because replacing simple with complex increases the potential for error and generates extra work. You are welcome to your preference.
Complex can also, of course, remove more error than simple can. Complex can in fact be designed to remove specific errors that are detected in simple, as long as you're not afraid of a little work. You are welcome to your preference.

Iain Fyffe is offline   Reply With Quote
Old
08-01-2012, 06:55 PM
  #84
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
1970-71 Canadiens Estimated TOI - Defensemen

1970-71 Canadiens estimated TOI for defensemen only. Very basic arithmetic involved, reflecting how the team was structured, coached and how it adapted to situations.

1970-71 Montreal Canadiens data:

http://www.hockey-reference.com/teams/MTL/1971.html

Laperriere injury, Savard return:

http://news.google.com/newspapers?id...pg=850%2C83185

Savard injury history:

http://www.hockey-reference.com/play...savarse01.html

Data - GP and PIM by participating Canadiens defensemen:

Defenseman-Games Played /PIM

J.C. Tremblay - 76 / 23
Guy Lapointe - 78/107
Terry Harper - 78/116
Jacques Laperriere - 49/20
Serge Savard - 37/30
Pierre Bouchard - 51 /50
Bob Murdoch - 1/2

Details - Serge Savard missed the first 11 games and the last 30 games of the season.

Jacques Laperriere played the first 10 games of the season then missed 27 games into early January 1971 and 2 games post January 30,1971.

Laperriere and Savard only played 12 games as part of the projected 5 defenseman rotation of Tremblay, Lapointe, Harper, Laperriere, Savard. Bouchard played games as the fifth defenseman when either Savard or Laperrierre were out. Bob Murdoch played the last game of the season. Murdoch game:

http://www.flyershistory.com/cgi-bin....cgi?H19700541

Team PP and PK data:

286 PPO / 24.83% success rate. Estimated PPMPG 5:30, Mainly 3F/2D power play

319 PPOA / 84.64% success rate. Estemated PKMPG 7:00. Mainly 2F/2D penalty kill.

120 defenseman minutes per game, breaks down to 11 PP minutes, 14 PK minutes, 95 ES minutes.

Estimated TOI for Defensemen - 1970-71 Canadiens regular season. To GP

J.C. Tremblay 29:09
Guy Lapointe 28:27
Terry Harper 26:14
Jacques Laperriere 26:30
Serge Savard 26:02
Pierre Bouchard 11:33
Bob Murdoch 15:00

Comments: difference in the success rate of the PP and PK created a situation where the Canadiens as a team and the defensemen spent more on ice time on th PK than the PP. Individual PIMs reduced on ice time for the highly penalized defensemen Harper and Lapointe. Sidebar - coaching change early in the season led to a situation of conflicts at times and some veterans were benched during games, notably J.C. Tremblay.

Thank you.

Canadiens1958 is online now   Reply With Quote
Old
08-01-2012, 08:41 PM
  #85
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Compared to my results, you have Lapointe 3 minutes per game higher, and Savard 3 minutes per game lower. And Bouchard 4 minutes lower. The rest are more or less the same, within a minute or so. That's the thing about these estimates: they are unlikely to contain big surprises. But they're systematic and don't rely upon inspection to arrive at the results.

But what is your method exactly? You mention simple arithmetic, but you don't actually tell us what you're doing. Can you provide situational numbers (ES, PP, SH), which are of course much more useful than just the total? Could you do a couple of other examples, say the 1985/86 Hartford Whalers and the 1977/78 Cleveland Barons? Or, indeed, all NHL teams for 1992/93?

Iain Fyffe is offline   Reply With Quote
Old
08-01-2012, 09:42 PM
  #86
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Basics

Quote:
Originally Posted by Iain Fyffe View Post
Compared to my results, you have Lapointe 3 minutes per game higher, and Savard 3 minutes per game lower. And Bouchard 4 minutes lower. The rest are more or less the same, within a minute or so. That's the thing about these estimates: they are unlikely to contain big surprises. But they're systematic and don't rely upon inspection to arrive at the results.

But what is your method exactly? You mention simple arithmetic, but you don't actually tell us what you're doing. Can you provide situational numbers (ES, PP, SH), which are of course much more useful than just the total? Could you do a couple of other examples, say the 1985/86 Hartford Whalers and the 1977/78 Cleveland Barons? Or, indeed, all NHL teams for 1992/93?
There are a few basics that have to be considered for each situation.

Let's take the Lapointe / Savard difference that you mention. Both were equally proficient on the PP BUT Savard was returning from a near career ending leg fracture, so he was nursed back into the line-up and rarely put at risk on the point. His time would have been managed to avoid unnecessary icetime. Bouchard reflects how Canadiens optimized the four other defencemen when he played. He did not appear in 27 games yet was with the team the whole season so he would not get much quality time - no PP or PK time in close or critical situations. The time gets distributed to the other dmen, notably Lapointe.

ES/PP/PK. Look at the raw data. PP = a team that has a 25% success rate would see 75 complete PP opportunities and 25 short opportunities that likely average around one minute each. Work the global average per 100 or do a HSP sampling of PPG times for 10 goals. Same approach for PK. Once you have the per game PP and PK subtract the total from 120 to get the ES. Then distribute amongst concerned players. For each player you will have a sum of ES+PP+PK-PIM+/- situational adjustment.

1977-78 Cleveland:

http://www.hockey-reference.com/teams/CLE/1978.html

If you can get the game day roster info like I was able to do for the Canadiens 1970-71 season then distributing the workload is easy. The calculations are straight forward but more time consuming since you are looking at 12 dmen as opposed to 7.

Likewise for the other teams or season. From what I have seen from your data if you factor in injury management apportion ice time for the 5th and 6th dman at a lower rate than the top 4 you will have a better estimate.
.

Canadiens1958 is online now   Reply With Quote
Old
08-01-2012, 09:53 PM
  #87
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Likewise for the other teams or season. From what I have seen from your data if you factor in injury management apportion ice time for the 5th and 6th dman at a lower rate than the top 4 you will have a better estimate.
.
I'm still no closer to understanding what your method is. When you say "distribute" ice time, do you mean simply by inspection? Or are you actually doing some sort of calculation?

Anything that affects a player's ice time will also affect the number of goals he is on the ice for. These things are not independent, but are heavily interdependent. As such any factor that affects one affects the other. That's why the goal data forms a good basis to build ice time estimates on.

Iain Fyffe is offline   Reply With Quote
Old
08-01-2012, 10:38 PM
  #88
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Two Comments

Quote:
Originally Posted by Iain Fyffe View Post
I'm still no closer to understanding what your method is. When you say "distribute" ice time, do you mean simply by inspection? Or are you actually doing some sort of calculation?

Anything that affects a player's ice time will also affect the number of goals he is on the ice for. These things are not independent, but are heavily interdependent. As such any factor that affects one affects the other. That's why the goal data forms a good basis to build ice time estimates on.
Assume that there are no penalties or goals in a game and a team has three defensive pairings. Does not matter if it is a 60 minute game generating 120 defensive minutes or a 65 minute game generating 130 defensive minutes. Is there an expectation that each of the defensemen play exactly 120/6 or 130/6 minutes or would: x+a / x / x-a be a more accurate expectation? Goals may be one or a good base but it is not the only one.

Situational playing time. The trend in hockey and football. Playing time or player usage is determined by the position of the faceoff or snap location. Alain Vigneault in Vancouver micro manages this. Which would give you a better TOI estimate if looking at an Alain Vigneault coached team.Relying on goals vs relying on faceoff position on the ice.?

Canadiens1958 is online now   Reply With Quote
Old
08-01-2012, 11:11 PM
  #89
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Assume that there are no penalties or goals in a game and a team has three defensive pairings. Does not matter if it is a 60 minute game generating 120 defensive minutes or a 65 minute game generating 130 defensive minutes. Is there an expectation that each of the defensemen play exactly 120/6 or 130/6 minutes or would: x+a / x / x-a be a more accurate expectation? Goals may be one or a good base but it is not the only one.
Is it impossible for you to describe exactly what you're doing to arrive at your estimates? That's all I'm asking right now. I don't want opaque comments, just tell me exactly how you arrived at the numbers you provided above. Provide specifics, not vague generalities.

Iain Fyffe is offline   Reply With Quote
Old
08-01-2012, 11:11 PM
  #90
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 28,212
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Assume that there are no penalties or goals in a game and a team has three defensive pairings. Does not matter if it is a 60 minute game generating 120 defensive minutes or a 65 minute game generating 130 defensive minutes. Is there an expectation that each of the defensemen play exactly 120/6 or 130/6 minutes or would: x+a / x / x-a be a more accurate expectation? Goals may be one or a good base but it is not the only one.

Situational playing time. The trend in hockey and football. Playing time or player usage is determined by the position of the faceoff or snap location. Alain Vigneault in Vancouver micro manages this. Which would give you a better TOI estimate if looking at an Alain Vigneault coached team.Relying on goals vs relying on faceoff position on the ice.?
Regardless of this, you have yet to explain why your estimate should be taken more seriously than one that proportionally assigns ice time to them proportionally based on the goals for and against each player in each situation.

seventieslord is offline   Reply With Quote
Old
08-01-2012, 11:31 PM
  #91
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Limits

Quote:
Originally Posted by Iain Fyffe View Post
Is it impossible for you to describe exactly what you're doing to arrive at your estimates? That's all I'm asking right now. I don't want opaque comments, just tell me exactly how you arrived at the numbers you provided above. Provide specifics, not vague generalities.
Definitely possible but definitely proprietary. Respect my proprietary rights like I respect yours.

Thank you for this consideration and for validating my efforts.

Canadiens1958 is online now   Reply With Quote
Old
08-01-2012, 11:43 PM
  #92
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Objective

Quote:
Originally Posted by seventieslord View Post
Regardless of this, you have yet to explain why your estimate should be taken more seriously than one that proportionally assigns ice time to them proportionally based on the goals for and against each player in each situation.
Never claimed that your assertion was my objective.

That said, view it as vanilla ice cream vs the flavour of the month that comes with various components and sparkles. Vanilla ice cream has been around since ice cream was invented and will always be around.

As stated previously - see the Alain Vigneault example up thread, I have doubts that the method you seem to favour will work for such situations. If it does more power to it.

Canadiens1958 is online now   Reply With Quote
Old
08-01-2012, 11:45 PM
  #93
Bear of Bad News
Mod Supervisor
 
Bear of Bad News's Avatar
 
Join Date: Sep 2005
Location: San Angeles
Posts: 4,273
vCash: 663
Quote:
Originally Posted by Canadiens1958 View Post
Definitely possible but definitely proprietary. Respect my proprietary rights like I respect yours.

Thank you for this consideration and for validating my efforts.
It's hard to validate your efforts if we can't see your efforts.

I could publish the NHL ice time statistics, tell you that I came up with them through a combination of games played, goals scores, height and weight, and the number of eggs that they ate over the past year, and that look! They match exactly. But I can't tell you how I did it, and please respect that.

Would you take that seriously?

Bottom line - unless your method is reproduceable, it's not "equally valid" to a method that works and is reproduceable.

Bear of Bad News is offline   Reply With Quote
Old
08-01-2012, 11:46 PM
  #94
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Definitely possible but definitely proprietary. Respect my proprietary rights like I respect yours.

Thank you for this consideration and for validating my efforts.
This is a joke, right? I've explained the basis of my method. I haven't provided every detail, but the logical underpinnings are clear. And my original method was published, in full, over a decade ago.

You have basically said "here are some number I came up with" and expect us to treat them as equally valid. To quote you from earlier in the thread:

Quote:
Originally Posted by Canadiens1958 View Post
So there is no independent verification of the study and no one has been able to replicate the data.

Bolded above. Correlation that was produced was to be expected since the known data may have been the inspiration and source for the study. Since we do not know the details of the study ........
You'll jump on someone else's method that hasn't been "independently verified" but then refuse to share your own method, in anything but the vaguest of vagaries.


I missed this one the first time through the thread:

Quote:
Originally Posted by TheDevilMadeMe View Post
I have no idea why Iain refuses to release the exact assumptions and calculations that go into his formulas
As I've said in this thread, I don't have all seasons done yet, and don't yet know the best way to disseminate the estimates yet. In other words: be patient. I have a lot of things on the go.

Iain Fyffe is offline   Reply With Quote
Old
08-01-2012, 11:48 PM
  #95
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
Never claimed that your assertion was my objective.
You claimed there were "equally valid" methods, and when pressed for an example, this is what you gave us. So you expect us to treat it as "equally" valid without actually explaining what you're doing in any level of detail. I have my suspicions, but that's all they are because you refuse to provide any sort of detail.

Iain Fyffe is offline   Reply With Quote
Old
08-02-2012, 12:13 AM
  #96
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Expectations

Quote:
Originally Posted by Taco MacArthur View Post
It's hard to validate your efforts if we can't see your efforts.

I could publish the NHL ice time statistics,
tell you that I came up with them through a combination of games played, goals scores, height and weight, and the number of eggs that they ate over the past year, and that look! They match exactly. But I can't tell you how I did it, and please respect that.

Would you take that seriously?

Bottom line - unless your method is reproduceable, it's not "equally valid" to a method that works and is reproduceable.
But the NHL holds the copyright for the stats you refer so it would be up to them to review the type of claim you describe. The NHL is protected. Iain has the copyright for his proprietary knowledge and is protected. Releasing proprietary knowledge without copyright protection is folly.

Effectively posters here have indicated that my method is worthy of consideration for reproduction and copyright. So I am much further ahead than I expected.

Canadiens1958 is online now   Reply With Quote
Old
08-02-2012, 01:28 AM
  #97
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 28,212
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
But the NHL holds the copyright for the stats you refer so it would be up to them to review the type of claim you describe. The NHL is protected. Iain has the copyright for his proprietary knowledge and is protected. Releasing proprietary knowledge without copyright protection is folly.

Effectively posters here have indicated that my method is worthy of consideration for reproduction and copyright. So I am much further ahead than I expected.
I tip my hat to you.

seventieslord is offline   Reply With Quote
Old
08-02-2012, 02:01 AM
  #98
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Method

Quote:
Originally Posted by Iain Fyffe View Post
This is a joke, right? I've explained the basis of my method. I haven't provided every detail, but the logical underpinnings are clear. And my original method was published, in full, over a decade ago.
Reviewed your method.

Defence: 22 MPG for the first pair, 18 for the second, 17 for the third.

For the defensemen,you are measuring from a three pairing base. I measure from a two pairing base. You total 57 minutes with 3 minutes to cover a 7th defenseman and other situations.

I use a 120 minute base as opposed to 60. I also split the pairings since the players that comprise a pairing are never equal. Example, a pairing where one player has 30 PIM in a season and another has 150 PIM in a season is weighed to reflect the difference since penalty box time reduces on ice time.I also weigh for other factors such as missed games within a pairing, veteran / rookie / call-up factors are considerations that are weighed as well. Beyond two pairings I define and weigh whether a team has a form of a five man rotation plus fillers/call-ups or actually three pairings.

Canadiens1958 is online now   Reply With Quote
Old
08-02-2012, 07:48 AM
  #99
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 3,469
vCash: 500
Quote:
Originally Posted by Canadiens1958 View Post
But the NHL holds the copyright for the stats you refer so it would be up to them to review the type of claim you describe. The NHL is protected. Iain has the copyright for his proprietary knowledge and is protected. Releasing proprietary knowledge without copyright protection is folly.
Did I take the red pill or the blue pill?

Quote:
Originally Posted by Canadiens1958 View Post
Effectively posters here have indicated that my method is worthy of consideration for reproduction and copyright. So I am much further ahead than I expected.
Actually, all we've been doing is questioning whether it's worthy of any sort of consideration, which we won't know until we some kind of detail. You say you distribute ice time to players, but give no indication of the basis on which you do so. For all we know, you're just making the numbers up.

Quote:
Originally Posted by Canadiens1958 View Post
Reviewed your method.

Defence: 22 MPG for the first pair, 18 for the second, 17 for the third.
That's just the normalizing factor, and it's also no longer the method I use. That was the first go at it, over 10 years ago (holy crap).

It's a relatively minor adjustment. The logical basis of the method is using goals the player was on the ice for as an indication of the amount of time the player spent on the ice, because these goal figures are influenced by every player on the ice, reducing the individual's direct effect on the numbers.

Quote:
Originally Posted by Canadiens1958 View Post
Beyond two pairings I define and weigh whether a team has a form of a five man rotation plus fillers/call-ups or actually three pairings.
Okay, but you still haven't told us the basis you use to assign the minutes. You've told us some things you consider, but haven't told us where the numbers come from.

Iain Fyffe is offline   Reply With Quote
Old
08-02-2012, 10:32 AM
  #100
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,658
vCash: 500
Basics

Quote:
Originally Posted by Iain Fyffe View Post

That's just the normalizing factor, and it's also no longer the method I use. That was the first go at it, over 10 years ago (holy crap).

It's a relatively minor adjustment. The logical basis of the method is using goals the player was on the ice for as an indication of the amount of time the player spent on the ice, because these goal figures are influenced by every player on the ice, reducing the individual's direct effect on the numbers.


Okay, but you still haven't told us the basis you use to assign the minutes. You've told us some things you consider, but haven't told us where the numbers come from.
Did tell you where the numbers come from. Will walk you thru examples.

Let's look at the the PK minutes element. During the 1970-71 season J.C. Tremblay, Terry Harper and Guy Lapointe played 76, 78, 78 games respectively. They had 23, 116 and 107 PIMs respectively. Jacques Laperrierreand Serge Savard played 49 and 37 games respectively. Their PIMS totaled 20 and 30 minutes. In terms of PIM/PG there is a ratio that is generated that ranks the players in PIMs. From 1 lowest to 5 highest.

For the team there is data, provided about the number of PK situations over the course of the season and the teams success ratio. This data is used to calculate and estimate the PK minutes per game that the team had to distribute amongst the defensemen.
We are looking at distributing the PK minutes amongst the five primary defensemen. Three options - a two pairing split with a % for the fifth and/or sixth defenseman. a first pairing/second pairing/ third pairing ratio or a five unit split with a small % for the sixth defenseman. The chosen option does not have to be the same for all segments of the season.

The chosen option is dictated by the the actual games played and the number of defensemen on the regular season roster. The Canadiens had 7 defensemen appear on their regular season roster including Bob Murdoch who only played one game. Reviewing the injury data for the Canadiens defensemen indicates that there are four distinct segments.Start of the season until Laperriere was hurt which almost coincided with the return of Serge Savard from a previous season injury.Laperriere missed games. The team had all five for about a dozen games then Savard missed the last 30 games. The most appropriate of the three options is selected for each segment 1st,2nd,4th segments are viewed as two pairing split plus while the 3rd segment is viewed as a five unit split with a small % for Bouchard.

Regardless of the choice, the availability of each to actually kill penalties has to be weighed. J.C. Tremblay was the most available so he gets credit with extra time over the average in proportion to his availability while the others get extra time or lose time in proportion to availability. No rule of thumb exists since PIMs vary from team to team, season to season. Basic calculations are done and the results for each segment are totaled to reflect the season.

PP minutes. First determine the type of PP a team ran 3F/2D or 4F/1D.1970-71 Canadiens ran a 3F/2D PP season long. Next step look at the available defensemen for the PP. Laperriere was available < 2/3 of the season while Savard was available < 1/2 of the season. Harper and Bouchard were not PP quality but did get late time when the team would be transitioning to ES.The Laperriere/Harper pairing was a mainstay entering their eight season used against the oppositions best forwards when ever possible. Given that Lapointe and Tremblay were available 100 and >97% of the season,PP time is weighed in their favour, proportionate to the availability and skills to the others.This is done for the same four segments explained in the PK section above then totaled for the season.

ES time. I see ES time as a function of PK and PP time. As illustrated above, J.C Tremblay had high availability for PK and PP time so this would lower his ES availabilitv time since he would be rested post PK and PP. Guy Lapointe would be rested post PP but he would not get as much rest time post PK since he would have rested in the penalty box.

Subtracting the PK and PP total time from total game time yields the ES time. The ES time is viewed as two groups. ES dictated by PPO and PPOA, sum estimated at 1 minute per opportunity. Second ES group is the remaining ES time not dictated by PPO and PPOA.Players time is apportioned and weighed per segment then totaled to reflect the season.

The final estimated tally for each defenseman, 1970-71 Canadiens, was presented up thread.


Last edited by Canadiens1958: 08-02-2012 at 10:35 AM. Reason: wording
Canadiens1958 is online now   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 07:27 AM.

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.