HFBoards

HFBoards (http://hfboards.hockeysfuture.com/index.php)
-   Winnipeg Jets (http://hfboards.hockeysfuture.com/forumdisplay.php?f=29)
-   -   Garret: Evaluating 2way players... Please judge me (http://hfboards.hockeysfuture.com/showthread.php?t=1295407)

garret9 11-20-2012 05:14 PM

Garret: Evaluating 2way players... Please judge me
 
Oh great... Garret is doing his stupid stats work again...

So out of boredom (or just procrastinating on lab work) I decided to try and take some peeks at who were the most effective two-way players in the game last year. The key word is effective, in that they were getting results. I was wondering if I could get some of your guys’ comments on my method and my results.

The way I evaluated these players was:
*they had to have played at least 30 games
*they have been mostly lining up against tough opponents/competition
*they had to score at a respectable rate for their team and position
*they had to been beating their competition and by a reasonable amount
*they couldn’t have their numbers being inflated by ridiculous offensive zone usage


Forwards:
*only players with 30+ games played
*>0.00 RelQoC (they aren't being lined up against 3rd/4th lines)
*<60 OZS% (no offensively inflated Sedins)
*>1.00 pts/60 (scoring at a respectable rate)
*>10 RelCorsi (they are solidly beating their competition)
ANA: Getzlaf, Perry
BOS: Seguin, Horton, Bergeron, Marchand
CBJ: Prospal
CGY: Moss, Stempniak
COL: Landeskog, O`Reilly, Downie
DAL: Benn
DET: Datsyuk
EDM: Hall
LAK: Williams, Kopitar
MIN: Setoguchi, Cullen
MTL: Cole, Pacioretty
NJD: Ponikarovsky
NSH: Legwand, Hornqvist
NYI: Tavares
NYR: Hagelin
OTT: Alfredsson
PHI: Jagr
PHX: Whitney, Hanzal, Vrbata
SJS: Thornton, Pavelski, Couture
STL: Steen
TOR: Grabovski
VAN: Kesler, Booth
WPG: Ladd, Little, Wheeler


I then tried the same parameters for D and only got three (Karlsson, Chara and Byfuglien). So, I then lightened the load to >0.80 pts/60 and >0 RelCorsi. Here were my results.
Defense:
*only players with 30+ games played
*>0.00 RelQoC
*<60 OZS%
*>0.80 pts/60
*>0 RelCorsi
ANA: Visnovsky
BOS: Chara
BUF: Ehroff
CBJ: Nikitin
CGY Giordano
CHI: Leddy, Seabrook
DAL: Goligoski
DET: White
EDM: Petry
FLA: Garrison
LAK: Doughty
MTL: Subban
NSH: Weber
NYI: Streit
NYR: McDonagh
OTT: Karlsson
PHI: Carle, Timonen
PHX: Rozsival, Ekman-Larsson
PIT: Martin, Letang
SJS: Boyle, Burns
STL: Shattenkirk, Pietrangelo
TBL: Hedman
VAN: Edler, Bieksa, Hamhuis
WPG: Enstrom, Byfuglien
WSH: Wideman

If you are wondering why Bogosian missed this list, he would have made it if he had been paired with Hainsey the full year instead of Stuart, who pulled him down hard.


I found that list to have too many offensive only defensemen. Now TBH, I rate offense as more important skill than defense, as a guy who can push the puck out of the zone reduces more scoring chances against than a guy who blocks lots of shots but keeps giving the other team the puck back. So I experimented with zone starts and finishes, and I’m curious to what you guys think of the change. What I did is I took any guy off the list if they ended their shift in the offensive zone less than they started. In other words, I got rid of the guys who couldn’t keep the puck in the other team’s zone when they didn’t score or couldn’t get the puck out of their own zone. So I kept anyone who’s OZF>OZS. Due to neutral zone information missing, I gave a leeway of 0.5% OZF-OZS. All other parameters were the same as last time.
BOS: Chara
CBJ: Nikitin
CHI: Seabrook
DAL: Goligoski
EDM: Petry
LAK: Doughty
MTL: Subban
NSH: Weber
NYI: Streit
NYR: McDonagh
PHX: Rozsival, Ekman-Larsson
STL: Pietrangelo
TBL: Hedman
VAN: Bieksa, Hamhuis
WPG: Enstrom, Byfuglien


So I thought, what would happen if I did the same to my original Forwards list.
ANA: Getzlaf, Perry
BOS: Bergeron
CBJ: Prospal
CGY: Moss, Stempniak
COL: O`Reilly
DAL: Benn
LAK: Williams, Kopitar
MIN: Cullen
MTL: Cole, Pacioretty
NJD: Ponikarovsky
NSH: Legwand
SJS: Thornton, Pavelski
STL: Steen
VAN: Kesler
WPG: Ladd, Little

It's really interesting who get's booted off the original list when you do that.

Thoughts? Comments? Feel free to provide

garret9 11-20-2012 05:24 PM

I will say right of the bat... there are some obvious problems:
It is strongly based on player usage...

Example:
One 2way player that missed the list was Toews, and only because of zone usage. Because, although he is a 2way player, Chicago really pushes him for offensive usage.

sully1410 11-20-2012 06:38 PM

I gotta ask dude...how do you still have a job or pass your classes or w/e is you do?

This **** must take forever to do! How do you find the time?

Unless your actually that homeless guy that i see all the time infront of 7-11 on Douglas st.(your from the island right...if not, apologies, and sub 7-11 on Douglas for portage place) that panhandles for money to use the Internet cafe down the street. If that's the case...it explains everything!

garret9 11-20-2012 06:47 PM

Quote:

Originally Posted by sully1410 (Post 55922405)
I gotta ask dude...how do you still have a job or pass your classes or w/e is you do?

This **** must take forever to do! How do you find the time?

Unless your actually that homeless guy that i see all the time infront of 7-11 on Douglas st.(your from the island right...if not, apologies, and sub 7-11 on Douglas for portage place) that panhandles for money to use the Internet cafe down the street. If that's the case...it explains everything!

My life: Wife, UBC(dual honors science degree), hockey, then work pt on w/e...
This actually didn't take me long to do. Behindthenet is realy user friendly to navigate once you know a few things.

And yes I previously lived in Vic for many years and I even think I know which homeless guy you're talking about.

sully1410 11-20-2012 07:21 PM

Quote:

Originally Posted by garret9 (Post 55922639)
My life: Wife, UBC(dual honors science degree), hockey, then work pt on w/e...
This actually didn't take me long to do. Behindthenet is realy user friendly to navigate once you know a few things.

And yes I previously lived in Vic for many years and I even think I know which homeless guy you're talking about.

I was debating between that guy and darth vader that plays violin in front of the bay centre.

GermanJetsFan 11-21-2012 02:42 AM

Garret? Is there a tutorial for those kind of stats or something like that? I don't understand at least half of them.

garret9 11-21-2012 04:02 AM

Quote:

Originally Posted by GermanJetsFan (Post 55929827)
Garret? Is there a tutorial for those kind of stats or something like that? I don't understand at least half of them.

Sure thing. I'll give a lil' breakdown but link to Gabriel Desjardins FAQs on advance statistics. Desjardins is a man originally from Winnipeg who runs behindthenet.com (the website where most advance stats guys gets their data) and has worked for as a consultant to multiple sports networks and NHL teams. He also founded Arctic Ice Hockey.

Part 1 - Plus/Minus statistics
I think every fan in hockey understands basic plus/minus; it's just a goal differential of when you're on the ice. There are more advanced plus/minus' and that's what Gabe talks in the link, but I usually don't use those because they can be affected by goalies and luck too much.
So plus/minus is the goals going on when you are on the ice. We all know this but Corsi is related so I wanted to review.

Part 3 - Corsi Number
Plus/minus is a goal differential, while Corsi is a shot differential (usually includes missed shots and blocked shots). As I noted above plus/minus is affected by the goalies on either side, the shooters percision and luck; what Corsi does is remove a lot of that noise, so the differential better correlates to who is outplaying who. Corsi is named after a Buffalo Sabres' goalie coach, Jim Corsi, who invented it. So basically, think of it as plus/minus but includes shots, goals, blocked shots and near misses. Also, we don't use raw whole numbers, but divide so the value equals the amount per 60 minutes of icetime. That way when you compare a 4th liner to a 1st liner, it evens out the difference from additional ice time. We usually do that with all stats. Corsi, like plus/minus, is positive when you are outshooting your opponent, and negative when you are being outshot.
Relative Corsi (RelCorsi) is a players Corsi relative to the teams average. This is the best for when compairing players on different teams since it removes the factors of playing on a really good or really bad team.

Part 2 - Quality of Competition
Quality of Competition (QoC) is what it sounds like. It's a measurement to determine whether or not you faced against the other teams top line, 4th line, or anywhere in between. In Gabe's writeup he uses a variation of plus/minus of the opponents the player lines up against to calculate their QoC. Nowadays (the article was written 5 years ago) we use the opponents Corsi. So basically all QoC are averages of a stat of your opponents you are on the ice against, but weighted for the amount of time you play against them. So 1 min against a topline and 2 mins against the 4th line is lower Qoc than 2 mins against a topline and 1 min against the 4th line. A larger (more positive) QoC means you mostly faced top lines, a smaller (more negative) QoC means you mostly faced 4th liners.
QoC (or CorsiQoC) is the avg Corsi of the opponents he faces, weighted to the time they play against each other.
RelCorsiQoC (or RelQoC) is the same thing but uses RelCorsi instead of Corsi. This is always better for determining usage IMHO.

Part 4 - Zone Starts (or Offensive/Defensive Zone Faceoff Ratio)
Offensive Zone Starts (OZS) is a measurement of the amount of times the coach puts you on the ice for a faceoff in the offensive zone, or the other teams defensive zone. This is used a lot to determine usage. Coaches tend to give their scorers mostly OZS while their more defensive players DZS (defensive zone starts). Combined with RelQoC this gives a good idea of a persons workload when it comes to scoring or Corsi. A person who usually starts in the DZone will have more trouble scoring than a person who usually starts in the OZone.
OZS% shows what percentage are you starting in the OZ relative to OZ + DZ starts. We don't include neutral zone starts in the calculations.
OZF% shows what percentage of your shifts end in the OZ... but it tends to not correlate as well with skill or usage. I was fiddling with using it as a skill determiner above, which was what I was asking about.

Part 5 - Penalty +/- (or Penalty Differntial)
Just another differential. This one is how many penalties you draw vs how many you take. Some players take more penalties than others. Some players draw penalties from other teams by being a pest or by being so skilled/fast that the other team consistently screws up and has to hook/slash. PPs are good and PKs are bad, so we look at the differential of these to determine if you are helping the team or hurting the team with your ability to draw and make penalties.

Part 7 - Goals and Points per 60 Minutes
As I mentioned before, we tend to make stats per 60 minutes of ice time. The reason for this is I may have more goals than you, but it may be because coach gives me more icetime, not because I score faster or better than you. We tend to seperate primary and secondary assists because secondary assits tend to fluctuate for a player hugely throughout a carreer and isn't a good indicator of skill (this is more true for forwards than dmen).
G/60 - goals per 60, A1/60 - primary assists per 60, pts/60 (or p/60) points per 60


There are other stats, but I have yet to use any of them...

garret9 11-21-2012 04:49 AM

Decided to add a few things for context. When comparing players and stats, unless indicated otherwise, stats guys always look at even strength (or 5v5) alone. This levels the playing field more when analyzing or compairing players. Also, when compairing players, we usually only look at players who have played more than a certain amount of games. Players who play small samples allows luck to play a larger influence in their performance stats than skill does. When I look at "the best of ___ players" it usually only looks at those with 30 or more games played.

Now to help the learning I'll show how/why I use stats and give examples in the league, and one of each of the Jets forwards and defensemen. I tend to seperate stats in my mind into two pieces: usage and performance. Two players may be used similarly but perform differently due to different skill levels. Two players may have performed similarly but one was actually more skilled but had tougher usage.

Situation/usage stats.
I pretty much only use OZS% and RelQoC here.
OZS: Lower than 50 OZS% means you gotta go uphill, while higher than 50% means you gotta go downhill which is much easier. The Jets highest F + D were Wellwood(60.5) and Byfuglien(53.4), the Jets lowest were Slater(28.5) and Hainsey (46.9). The leagus highest are the Sedins(79.6 and 78.6). Noel pushed the top 2 lines and the top pairing to more offensive situations with increasing their OZS, and he pushed the 3rd line (GST) and second pairing D to more defensive situations with decreasing their OZS.
RelQoC: As I explained it shows the toughness of your opponents you face, where higher is tougher. Highest Jets were Slater(0.900) and Byfuglien(0.674), lowest were Jones and Stapleton (-0.913 and-0.549). Second highest in the league was Jokinen (2.056), Bisonatte was 8th lowest (-1.837). Noel gave tough QoC to Little and Slater's lines so that Kane's lines would predominately face weak competition so Kane would score more.
The problem happens with evaluating the strenght of these effects. IMHO, it's affects each player different, since players have different skillsets. That makes it hard to ask what is tougher lots of DZS with easy competition or lots of OZS with tough competition.

Performance stats.
RelCorsi: the way I think of RelCorsi is are they beating their competition, and if so by how much... given a long enough period of time almost anyone who has a positive Corsi will outscore their competition. We also use it as a way to approximate puck possesion. If you have more time with the puck, you'll get more shots on net. Jets highest were Ladd and Enstrom(17.1 and 14.5) while lowest Slater and Flood (-27.9 and -9.2). D.Sedin has the leagues highest with 22.5. Obviously Slater's would be higher than 2nd worse in the NHL if he wasn't in the DZ so much and faced weaker competition, and obviously Sedin's would be lower if he wasn't starting in the OZ so much... this is why the usage stats are important.
g,a1,p/60: this is their scoring rate, so I think it's pretty self explanatory why this shows a players performance. Kane was 14th in the league for g/60 and Wheeler was 26th in the league for a1/60. For D Buff had teams best g/60 and Enstrom had the best A1/60.


For you or anyone who want's to check out the stats yourself:
behindthenet

ps241 11-21-2012 06:41 AM

Good analysis garret I always learn something from your posts.

Burgeron has always been my favorite 2 way player in the league.

Edit: Garret that link to behind the net is great and that table is so easy to manipulate. I appreciate your tutorial and catagory breakdowns you are starting to convert me. I admit that I didn't watch the game much after the Jets left so the advanced stats are new to me but I am aquiring a taste.

GermanJetsFan 11-21-2012 01:26 PM

Awesome Garret :laugh:

thank you very much.

garret9 11-21-2012 04:34 PM

Although I love compliments, I was looking for holes haha.

If there's a few thoughts here and there of wrongful information, then it may be just be a difference of opinion... But if there is consensus on a problem, then I can try to tweek it.

I don't watch too much, non-Jets, so any views help.
I know just from some common thoughts on hfboards some problems could be:
*Wheeler is one of the best 2way guys and shouldn't of been dropped in the last step.
*Byfuglien is offense only and shouldn't be on a 2way list
*how is ___ not on here while ___ isn't.
*how the eff did Datsyuk get the boot on the last step or is he getting old...
Etc.

*PS those aren't neccessarily my views, just predicting some responses haha

garret9 11-21-2012 07:05 PM

So change in plans; I don't need your responses.
I've had some stuff escalate that need my attention, and one outcome from it is most likely a severe reduction in HFBoards as well(and AIH as well). Anyways, I'll probably be around still from time-to-time but (most will be saying thankfully), I won't be really doing my stuff as before.

King Woodballs 11-21-2012 07:27 PM

Quote:

Originally Posted by garret9 (Post 55948403)
So change in plans; I don't need your responses.
I've had some stuff escalate that need my attention, and one outcome from it is most likely a severe reduction in HFBoards as well(and AIH as well). Anyways, I'll probably be around still from time-to-time but (most will be saying thankfully), I won't be really doing my stuff as before.

When she due?

scelaton 11-21-2012 07:56 PM

Quote:

Originally Posted by garret9 (Post 55948403)
So change in plans; I don't need your responses.
I've had some stuff escalate that need my attention, and one outcome from it is most likely a severe reduction in HFBoards as well(and AIH as well). Anyways, I'll probably be around still from time-to-time but (most will be saying thankfully), I won't be really doing my stuff as before.

Quite the contrary, Garret, everyone here is very thankful for your contributions. In content, analysis and just plain decency, you're definitely one of this board's top 6 forwards, with high RelCorsi:).
Seriously, hope everything is OK on your end.

ps241 11-21-2012 08:46 PM

Yea garret I hope all is well?

garret9 11-21-2012 09:08 PM

Haha just one last thing:
All is well
My wife isn't pregnent
Just workload has gone sky high and a really important project has fallen to me

King Woodballs 11-21-2012 09:25 PM

Quote:

Originally Posted by garret9 (Post 55951111)
Haha just one last thing:
All is well
My wife isn't pregnent
Just workload has gone sky high and a really important project has fallen to me

Well that was a swing and a miss :laugh:
All the best

Bob E 11-21-2012 11:46 PM

Quote:

Originally Posted by garret9 (Post 55948403)
So change in plans; I don't need your responses.
I've had some stuff escalate that need my attention, and one outcome from it is most likely a severe reduction in HFBoards as well(and AIH as well). Anyways, I'll probably be around still from time-to-time but (most will be saying thankfully), I won't be really doing my stuff as before.

Oh, man! I'm gonna miss your analysis, big time.

And i had my list - of guys who should be on this 2way list - ready. Guys like:

Travis Zajac

Jordan Staal

Nick Backstrom

Marion Hossa

Henrik Zetterberg

David Backes

Mikko Koivu

Huffer 11-22-2012 08:38 AM

Quote:

Originally Posted by garret9 (Post 55948403)
So change in plans; I don't need your responses.
I've had some stuff escalate that need my attention, and one outcome from it is most likely a severe reduction in HFBoards as well(and AIH as well). Anyways, I'll probably be around still from time-to-time but (most will be saying thankfully), I won't be really doing my stuff as before.

Hopefully you will have time to come back once things settle down. Your quantitative analysis skills will be missed.

sully1410 11-22-2012 10:19 AM

Quote:

Originally Posted by garret9 (Post 55948403)
So change in plans; I don't need your responses.
I've had some stuff escalate that need my attention, and one outcome from it is most likely a severe reduction in HFBoards as well(and AIH as well). Anyways, I'll probably be around still from time-to-time but (most will be saying thankfully), I won't be really doing my stuff as before.

The great thing about this place is....it'll still be here in all it's homer glory when you get back.

If you put half as much work into your important project as you do with our stats, you'll knock it out of the park.

I thought it was a good guess with the pregnancy KW.

truck 11-23-2012 01:18 PM

Cole and Pacioretty those names are the two that surprised me the most.

garret9 12-16-2012 07:42 PM

Quote:

Originally Posted by Bob E (Post 55954209)
Oh, man! I'm gonna miss your analysis, big time.

And i had my list - of guys who should be on this 2way list - ready. Guys like:

All of these guys didn't make the RelCorsi cutoff:
http://www.behindthenet.ca/nhl_stati...67+17+18+19+20
But also...


Travis Zajac (RelCorsi -3.8, Corsi -4.35)
Only played 15 games... sample size too small to look at.

Jordan Staal (RelCorsi 4.6, Corsi 12.02)
He fell off of the list because of problems with the formula. He has a solid Corsi, but because he's on such a strong team his RelCorsi isn't that great. RelCorsi is good because it reduces team effects, but being on an elite team like the Penguins kinda killed him a bit... next season will be a better test.

Nick Backstrom (RelCorsi 3.75, Corsi 8.61)
Too sheltered from tough forward competition to be considered. -0.272 RelQoC as the Laich line and 3rd line took all the tough minutes.

Marion Hossa (RelCorsi 1.9, Corsi 7.49)
Those numbers aren't actually that great considering he was pushed to the OZS 59.0% and didn't face tough competition (0.29 RelQoC). He had similar numbers in Corsi and pts/60 as Wellwood with similar usage.
Also, loses in the OZS>OZF department.

Henrik Zetterberg (RelCorsi 6.1, Corsi 10.38)
See Staal.
Also, loses in the OZS>OZF department.

David Backes (RelCorsi 5.6, Corsi 10.45)
Exact same reasoning as Staal.

Mikko Koivu (RelCorsi 6.1, Corsi -8,17)
Opposite of Backes and Staal. He had one of the least negative corsis on a team that EVERYONE (except one) was negative.


Better late than never :P
Only 2 exams left to go!!


All times are GMT -5. The time now is 02:41 AM.

vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
HFBoards.com, A property of CraveOnline, a division of AtomicOnline LLC ©2009 CraveOnline Media, LLC. All Rights Reserved.