HFBoards

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

Team Building

Reply
 
Thread Tools
Old
02-18-2009, 05:13 PM
  #1
Pred303
Registered User
 
Join Date: Oct 2004
Location: Murfreesboro, Tn.
Country: United States
Posts: 2,815
vCash: 500
Team Building

posted this post on np.com and din't really get much responce, just thought i'd share it from an old hockey guy;

*warning. very long, rambling post about things some might not find interesting*

HOCKEY 404, TEAM BUILDING/POINT PREDICTION

have been pm'ed by a person asking how i come up with the prediction that we're 'only about an 85 point team' and have in the past fairly accurately predicted just about the point total we wound up at. so i thought i'd throw out so old hockey stuff from the past that some gm's did to 'predict' about where their team was going to be and what moves they might need to make. back in the early 80's i followed the oilers. the oilers had an old GM whose name escapes me now, that was generally considered to be a wizard building a team. he gave a series of interviews back then explaining what he did pre-season to best guess what lineup combination would equal about what final point finish for a team. something i mentally still use to SWAG (scientific wildazz guess) to guess what point total a lineup made of specific players should/most likely will ultimately probably produce.

the system he used basically went like this.
...he first divided players into 4 position categories. 1)scoring line type forwards, 2)grinding/checking line type forwards, 3)defensemen, 4)goalies.
...next he ranked each player in each postion category into whether he thought he was a 1)star, 2)an elite player, 3)an above average guy, 4)an average player, 5) a below average player, 6)a weak player
... for each of those 4 categories of player positions, each 'type of player' he had a number that he thought translated into a specific number of probable team wins, simply based on role and talent level.
...he then added up the expected daily 19 man 'win' total expected and from that could pretty accurately predict the number of wins any specific lineup should/would probably produce.

for example..
..for scoring line forwards he thought star players would equal about 8+ more wins over the course of a full season, that elite players would probably equal about 6 wins, above average guys 4 wins, average guys 2 wins, below average 0 more wins, and weak players would take away about 2 expected wins.
...for grinder forwards, he thought star level type grinders would translate into about 4 wins, elite 3, above average 2, average 1, below average 0, and weak grinders cost you 1 game each
..for defensemen, he thought stars would equal about 8+ extra wins, elite ones 6, above average 4, average 2, and below average 0, and weak ones -2
..for goalies, he thought stars would translate into 10+ extra wins, elite ones 8, above average 6, average 4, below average 0, and weak ones -4

confusing sounding stuff of course, but really simple if you can honestly judge talent and potential analytically instead of emotionally or thru hope.

for example let's look at our pre-season lineup we expected, what 'expectations' were and plug in his expected win numbers...
scoring line guys.. arnott 4(above average), erat 4(above average), legwand 4(above average), dumont 4(above average)jones 0(below average), hornqvist 0(below average).. so a net total of 16 wins.
grinder line type guys.. fiddler 1(average), bonk 1(average), nichol 1(average), tootoo 1(average), belak 0(below average)smithson 0(below average).. so about a total of 4 wins.
defensemen... weber 4(above average), suter 4(above average), hamhuis 4(above average), zanon 2(average), devries 0(below average), koistinen 0(below average).. so maybe 14 wins from the blueline.
goalies.. ellis 6(above average), rinne 4(average).. so 10 odd wins (note they have basically flipflopped)

so with everyone completely healthy and playing at just about expectation level you come up with about 44 expected wins (88 points). of course, you always assume everyone will not be completely healthy so even if everyone plays completely to potential, you have to subtract some number of points from the projected total to come up with the adjusted projected total based on players fitting their roles. he said he typically subtracted 3-5 points to account for probable missed time due to injuries.. which is basically where i came up with 85 points this year. and 91 last year (91 actual), and 110 the year before (110 actual) and 100 the year before that (106 actual), and 90 the year before that (91 actual).

as a further example, look at the 05-06 lineup we had and why the forecast was so much higher...

kariya 6(elite), sullivan 6(elite), hartnell 4(above average), legwand 4(above average), walker 4(above average), erat 4(above average).
perreault 1(average), johnson 1(average), smithson 1(average), hall 1(average), tootoo 1(average), hordichuk 1(average)
timonen 4(above average), zidlicky 2(average), markov 2(average), eaton 2(average), suter 0( below average), york 0(below average)
vokoun 8(elite)

so going into that season using his methodology there was an expectation that lineup could win as many as 52 games. backing out probable injuries one could realistically guess that this team would put up about 100-102 points. (106 actual with the late adds of witt, weber, and especially sillinger)

it always surprises me how accurately this methodology can predict probable point totals. of course there are always extenuating circumstances. some people will outperform, some will under perform, injuries and adds will happen.. but it's still about as accurate as one can guess.

one thing a person can use this methodlogy to look at, is how trades or adds might affect final point totals to a specific team or possible trade or pickup.. for example, adding a superstar like ovechkin would in theory add about about 8 wins as a star to our current team if we picked him up in free agency.. it's always a domino type effect when you add player, in that that player replaces someone, so you have to back out whatever they would have brought you, and you also have to figure how one player might bump another and improve say your grinding line, etc.. so say an elite player replaces an above average player you get a net add of about 4 points (6 wins - 4 wins). but if that bumped guy, can play on a grind type line, maybe he becomes an above average grid type player replacing an average one so maybe you pick up an additional win because of that as well. but, if it takes losing two above average players, to afford a superstar, well the net, net is about the same. any more than that, and you even tend lose out in the long run if it makes you have to drop off anymore than that to sign a 'star'. (as pittsburg and atlanta are prime examples)

but i've always found it to be a pretty useful planning tool to try to maybe help quantify the unquantifiable.

apologize for the rambling nature and length as i've tried to type around phone calls and work.

Pred303 is offline   Reply With Quote
Old
02-19-2009, 10:32 AM
  #2
jstreet
Registered User
 
jstreet's Avatar
 
Join Date: Jan 2004
Location: smashville
Country: United States
Posts: 7,739
vCash: 1500
Have you seen THESE FIGURES?

This predicts we have a 12% chance of making the playoffs

jstreet is offline   Reply With Quote
Old
02-21-2009, 03:18 AM
  #3
sparkle twin
I luv u summertime
 
sparkle twin's Avatar
 
Join Date: Jul 2002
Location: Hendersonville, TN
Country: United States
Posts: 2,920
vCash: 500
Send a message via ICQ to sparkle twin Send a message via AIM to sparkle twin
And here I thought SWAG was just stuff we all get. I didn't know it was so versatile.


So even if we had Radulov we probably would only have about 2-4 more wins. And if Sully was healthy all year another 2-4 wins. So possibly 4-8 wins, 8-16 pts. Added to our 85 is 93-101 (I didn't take any pts out because I figured those 2 would replace Jones/Hornqvist and they had 0 wins/pts). Hopefully it would have been more than 93 pts because that might still have put us in the 7/8 spot and playing det/sj....again.


Hopefully someday Weber can step up into that elite or star category and get us a few extra wins. Really I think the Weber/Suter pair can be an elite pair. Would be nice if that got us an extra 8 wins.

sparkle twin is offline   Reply With Quote
Old
02-22-2009, 03:02 PM
  #4
Enoch
This is my boomstick
 
Enoch's Avatar
 
Join Date: Jul 2003
Location: Cookeville TN
Country: United States
Posts: 12,528
vCash: 500
Spectacular post. I think Arnott is in the elite range, but when talking about top 6 forwards, I think Hornqvist may be in the weak range right now.....which would even out. This year I think Weber and Suter both have arguments for being in the elite range, with Suter (IMO) being elite. Koistinen would be on the weak side of things...

__________________
- Enoch -
Enoch is offline   Reply With Quote
Old
02-22-2009, 04:07 PM
  #5
darth5
Rowsdower!
 
darth5's Avatar
 
Join Date: Mar 2002
Location: Smashville, TN
Country: United States
Posts: 2,411
vCash: 500
very interesting approach, p303. Modeling like this can help you project the normal expectation-- i.e., season impacting injuries, extremely hot streaks, unknown goal tending emerges, significant player trades, etc would change the model and invalidate the original assumption. I would go so far as to posit that the predictability of this team is influenced by the risk averse management model it operates under, also. The Forsberg trade was the single biggest risk DP undertook in the franchise's existence. Don't see that changing without intervention from ownership, which is unlikely at this stage.

darth5 is offline   Reply With Quote
Old
02-24-2009, 06:29 AM
  #6
Pred303
Registered User
 
Join Date: Oct 2004
Location: Murfreesboro, Tn.
Country: United States
Posts: 2,815
vCash: 500
Quote:
Originally Posted by Enoch View Post
Spectacular post. I think Arnott is in the elite range, but when talking about top 6 forwards, I think Hornqvist may be in the weak range right now.....which would even out. This year I think Weber and Suter both have arguments for being in the elite range, with Suter (IMO) being elite. Koistinen would be on the weak side of things...

well of course, the truly difficult part of this type model, is to correctly classify players. because what you have to do is predict what type of season you think they will have and of course be very good at evaluating ability.

the evaluation of a player into a predictive category is of course subjective to many variables. among scoring line forwards for example, it cann't simply be if he scores 20-29 he's an above average guy, 30-35 an elite guy, and 36+ he's a star. it has to inculde all the intangibles as well and his overall game. for example, kovachuk might well score 40+ goals every year. but maybe his lack of defensive responsibility prevents him from falling into that star category in this model. maybe in fact he's net net only an elite type player because of this.

same goes for our team. maybe arnott's 30 goal scoring status makes him an elite player in terms of goalscoring, but his overall game reduces that net worth to only being an above average player.

one thing i can guarantee you, is that pretty much all GM's use this type of predictive model in planning for what their team should look like. some are of course far better at it than others.

and yes yoda, there are hundreds, if not thousands of variables. things that can happen to change all original expectations. but like all planning tools, it's a place to start. some players will have better than expected years, some will have worse. (as enoch notes in our situation, you will tend to be surprised at how those cancel each other out). but. if you are good at this evaluation and if you were largely correct in your original estimate, you will be surprised at how accurate this can be predicting what will happen with a particular lineup in terms of point totals over the course of a season.

Pred303 is offline   Reply With Quote
Old
02-24-2009, 04:44 PM
  #7
nine_inch_fang
Registered User
 
Join Date: Oct 2004
Location: Spring Hill, TN
Country: United States
Posts: 1,506
vCash: 500
Send a message via Yahoo to nine_inch_fang
Another major problem that one could run into with this model is trying to build a video game type team. The Rangers and Leaves come to mind.

If you have an elite or above average scoring line player but no openings on your top six then you really can't add the 4 to 6 wins for that player because they may not actually be utilized properly/to their potential on a grind/checking line.

nine_inch_fang 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 04:47 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.