New club: Hockey Statistics -- Reference and Discussion

[...]

I wouldn't ever lump polling into the same category as NHL stats collection.

[SwampD]

3 minutes ago, ... said:

A scourge on society?  You could have stopped there, having said all you need to have said.

There is no irony.  I just explained why to the layman, no less someone who thinks data collection is a "scourge on society", a peek into the inner world can lead to erroneous conclusions.  If you were to poke around the inner world of surgeons, for example, you might find fault with robot-assisted surgeries or non-invasive techniques.  And then you would prefer for heart-bypasses the patient is sliced open from foot to sternum.

On the third point, without collecting the first set of data and analyzing it, they have no idea what additional data, or changes to the collection process, they need to make.  That's how it works.  It's science.  

Yes, I’m sure that any issue that arises with robot assisted surgery is promptly reported. Any hospital administrator would be happy to shut down that revenue stream to make sure they get it right.[/s]

 

[SwampD]

I guess for anyone who has asperger’s or ocd big data is wonderful. It sends them down the same pathways of the internet day after day and makes sure they see the same feeds in their Pintrest profile or Prime homepage.

[North Buffalo]

9 minutes ago, ... said:

I wouldn't ever lump polling into the same category as NHL stats collection.

No not same level of flaws, but polling uses large sample sizes to mitigate flaws and still has them.  Observational data still has issues of bias of observers that is again hard to mitigate let alone ID.

[North Buffalo]

14 minutes ago, SwampD said:

Yes, I’m sure that any issue that arises with robot assisted surgery is promptly reported. Any hospital administrator would be happy to shut down that revenue stream to make sure they get it right.[/s]

 

Those errors still happen and I bet minor ones are are under reported.  I hear about them from my wife an OR nurse and can often be attributed to the user/Dr.  Its the human element that is the problem.. the user/observer and the human object of observation... the uncontrolled variable... not saying Stats are not really valuable... again teasing out their limitations can be tough.  Otherwise couldnt we figure out who best a player(s) based on skill set could be matched with. Stats might give us an idea of who to try.. still too many personality and unseen skills to be sure.

[Randall Flagg]

8 hours ago, SwampD said:

Very cool. Thanks for the legwork.

My only reaction is, “All this analysis is done from data observed in real-time?!?!”,... really?!

And the solution is coming up for ways to “correct” for incorrect data? How about going back and re-analyzing games accurately. I mean, it only costs $139 dollars a season to watch every game as many times as you want.

While I still believe stats are useful, this information strengthens my eye test/analytics weighting, and It’s going to take a lot for me to change it.

In 2008 it was, at least. as recently as 2017, I found an article that basically confirms it's impossible to find this information anywhere haha.

For what it's worth, whatever you think of the data, it still outperforms all basic counting stats in all the ways that have been outlined over the years, and I dont think anyone is miscounting goals, assists, or points! If it didn't, we wouldn't use it. 

[Randall Flagg]

35 minutes ago, ... said:

I wouldn't ever lump polling into the same category as NHL stats collection.

Orders of magnitude difference there. The "error analysis" done by polling versus that done by any hockey nerd, or any real scientist, are simply not comparable from what I've seen. (I'm sure not all polls are conducted this way, I know a guy like Nate Silver is likely rigorous)

[North Buffalo]

8 minutes ago, Randall Flagg said:

Orders of magnitude difference there. The "error analysis" done by polling versus that done by any hockey nerd, or any real scientist, are simply not comparable from what I've seen. (I'm sure not all polls are conducted this way, I know a guy like Nate Silver is likely rigorous)

Some pollers are more rigorous no doubt.  Having worked in the field many years ago and problem still to this data is weeding out unlikely voters and the skew resulting from non responders.

[shrader]

Isn't it a bit redundant to have both the club and then also this thread?  I'm not trying to be a Richard here, but I'm not really sure what is the real goal of having all the clubs.  I like the idea of having a stand alone source for all of this, but isn't it supposed to do just that, stand alone?

[SwampD]

4 minutes ago, shrader said:

Isn't it a bit redundant to have both the club and then also this thread?  I'm not trying to be a Richard here, but I'm not really sure what is the real goal of having all the clubs.  I like the idea of having a stand alone source for all of this, but isn't it supposed to do just that, stand alone?

I tried to post it there but nobody answered over there.

[shrader]

1 minute ago, SwampD said:

I tried to post it there but nobody answered over there.

Maybe it's the thread starter taking his ball and going home in truly confusing fashion?

[Randall Flagg]

I'm not joining the club because last night was enough stats reading for the next year or so. Also, I'd never remember to check it anyway

[Lanny]

PDOcast with Dmitri Filipovic's most recent episode was a really good listen. Guest was Andrew Thomas:

 

 https://hockeypdocast.com/2019/07/11/ep-306-the-next-frontier/

[spndnchz]

Words...

[TrueBlueGED]

On 7/11/2019 at 12:38 AM, Randall Flagg said:

I am dreadfully sorry for how long this is, I didn't mean it to get that way

Life as a grad student on break, right? 

On 7/11/2019 at 9:19 AM, North Buffalo said:

Data collection, how and by whom... ie their inherent bias is always an underrated problem that is tough to weed out and in many respects intractable because it is the only way to collect it and tough to detect flaws even when standards are applied.  Part of the reason political polling is so inaccurate.

Define inaccurate. Polls aren't perfect, of course, and it also depends on what the poll is asking....but they're pretty accurate.

On 7/11/2019 at 10:05 AM, Randall Flagg said:

Orders of magnitude difference there. The "error analysis" done by polling versus that done by any hockey nerd, or any real scientist, are simply not comparable from what I've seen. (I'm sure not all polls are conducted this way, I know a guy like Nate Silver is likely rigorous)

Care to expand? The major pollsters are incredibly rigorous and put a ton of effort into getting an accurate picture. Also, Nate Silver isn't a pollster, he simply uses them as the foundation of his prediction models.

[North Buffalo]

4 minutes ago, TrueBlueGED said:

 

Define inaccurate. Polls aren't perfect, of course, and it also depends on what the poll is asking....but they're pretty accurate.

 

Asking, question bias, picture in time.. those that dont answer,lie.... all issues especially in politics.  Wilder in VA for Governor we assumed had a 5-7 point negative in polls because of racial bias. And it played out he won if I remember by 5 points despite polls saying he was up by 13. Talking tracking polls.

[TrueBlueGED]

36 minutes ago, North Buffalo said:

Asking, question bias, picture in time.. those that dont answer,lie.... all issues especially in politics.  Wilder in VA for Governor we assumed had a 5-7 point negative in polls because of racial bias. And it played out he won if I remember by 5 points despite polls saying he was up by 13. Talking tracking polls.

A result from the 90s!?!?!?! ?

Really though, polling isn't perfect, and you're of course going to find polls where the actual result falls outside the margin of error. But that's not the norm. Polls as a whole are pretty accurate, even if not fool-proof. I also bristle at the notion that pollsters aren't methodologically rigorous and putting good-faith efforts into accuracy, though I realize that wasn't your assertion.

[North Buffalo]

54 minutes ago, TrueBlueGED said:

A result from the 90s!?!?!?! ?

Really though, polling isn't perfect, and you're of course going to find polls where the actual result falls outside the margin of error. But that's not the norm. Polls as a whole are pretty accurate, even if not fool-proof. I also bristle at the notion that pollsters aren't methodologically rigorous and putting good-faith efforts into accuracy, though I realize that wasn't your assertion.

I didnt say the pollsters werent methodical and even the callers try hard, but accounting for the unknown is difficult and then in politics the lying on answers or answering one way on issues but voting another way based on socio ethno norms are difficult to weed out and a 1-2% error rate can be a huge difference in polls.  Also determining who actually will vote, turnout is a huge issue and which sides ground game is most effective.  All huge problems for pollsters to accurately assess. And ground games can vary widely from one election to next, so prior elections are not always a good barometer, unless its the same candidate and much of the same staff.

[TrueBlueGED]

10 minutes ago, North Buffalo said:

I didnt say the pollsters werent methodical and even the callers try hard, but accounting for the unknown is difficult and then in politics the lying on answers or answering one way on issues but voting another way based on socio ethno norms are difficult to weed out and a 1-2% error rate can be a huge difference in polls.  Also determining who actually will vote, turnout is a huge issue and which sides ground game is most effective.  All huge problems for pollsters to accurately assess. And ground games can vary widely from one election to next, so prior elections are not always a good barometer, unless its the same candidate and much of the same staff.

Yea, my apologizes. I was trying to respond to both you and Flagg (who I understood to be questioning the methodological rigor of pollsters) and it got all muddled up. I've been in a hospital waiting room since 6am and my brain is slowly melting. I don't disagree with any of the challenges you just rattled off. 

Edited July 12, 2019 by TrueBlueGED

[North Buffalo]

You okay?

[TrueBlueGED]

3 minutes ago, North Buffalo said:

You okay?

I'm fine, thanks. My fiancee's mom is having a valve replaced and while inside they found damage to a second valve that they're working on now. They just told us to go get dinner if we haven't already because it'll be awhile. 

[North Buffalo]

2 minutes ago, TrueBlueGED said:

I'm fine, thanks. My fiancee's mom is having a valve replaced and while inside they found damage to a second valve that they're working on now. They just told us to go get dinner if we haven't already because it'll be awhile. 

Ah... food good... take a cat nap even if its in a car... I live by then.. ill say my prayers... 

[triumph_communes]

Daily dosage of grains of salt:

 

The curious story of Justin Schultz:

 

 

RAPM charts still require CONTEXT-- the models don't explain it all

[pi2000]

Some TRpm data for the past season.   

TRpm is the difference between the players actual +/- and what their expected +/- is based on their teams performance, their number of games played and % of minutes played.   

A TRpm of 0 on a good team means you're an equal contributor to the team's success (when compared to your teammates).   

A TRpm of 0 on a bad team means you're an equal contributor to the team's failure.

eg, Brett Pesce had an actual +/- of +35.  CAR's even strength goal differential was +22,  over 82 games that's +0.2683/gm.    Pesce played 73 games, and averaged about 20min of ice time (33% of 60min).     Over 73 games, we would expect CAR to have an even strength goal differential of 73 * 0.2683 = 19.59.   Pesce played 33% of total minutes, so his expected +/ would be 19.59 * 0.33= +6.5.   However, his actual +/- was +35... a difference of +28.5.... meaning that CAR had a much better even strength goal differential with him on the ice than without.    

In summary, this stat tells you which players helped or hurt their team the most.... eg, on a bad team, negative TRpms identify the players who are weighing the team down... those players most responsible for the teams' lack of success.    Likewise, a large positive TRpm on a bad team identifies players who might be attractive trade targets ?

Min 20 games played.

Top 25:

CAR Brett Pesce             28.31
OTT Mark Stone              26.81
CAR Teuvo Teravainen        23.43
PIT Brian Dumoulin          22.88
DAL Alexander Radulov       21.16
CBJ Josh Anderson           19.61
CGY Mikael Backlund         19.43
CGY Mark Giordano           19.02
DET Tyler Bertuzzi          18.60
VGK Nate Schmidt            18.19
COL Nathan MacKinnon        18.16
TOR Ron Hainsey             18.00
COL Nikita Zadorov          17.78
TBL Ryan McDonagh           17.76
WPG Andrew Copp             17.61
CAR Sebastian Aho           17.61
EDM Matt Benning            17.38
NSH Nick Bonino             17.34
LAK Jake Muzzin             17.00
ARI Niklas Hjalmarsson      16.54
PHI Claude Giroux           16.51
VAN Troy Stecher            16.26
CHI Duncan Keith            16.07
DAL Tyler Seguin            15.96
OTT Dylan DeMelo            15.90

 

Bottom 25:

PHI Wayne Simmonds         -15.62
FLA Mike Hoffman           -15.68
WPG Bryan Little           -15.69
VAN Ben Hutton             -15.79
TOR Patrick Marleau        -15.81
NYI Jordan Eberle          -15.96
CGY Sam Bennett            -15.97
MTL Matthew Peca           -16.00
VGK Max Pacioretty         -16.19
NYI Mathew Barzal          -16.35
EDM Adam Larsson           -17.19
WSH Chandler Stephenson    -18.04
NSH Kevin Fiala            -18.05
TOR Auston Matthews        -18.23
CBJ Brandon Dubinsky       -18.91
LAK Ilya Kovalchuk         -19.24
CBJ Zach Werenski          -19.25
LAK Drew Doughty           -19.81
SJS Justin Braun           -20.11
CAR Warren Foegele         -21.30
VAN Erik Gudbranson        -22.21
BUF Rasmus Ristolainen     -23.04
PIT Phil Kessel            -26.49
WPG Patrik Laine           -28.02
PIT Evgeni Malkin          -31.50

 

[TrueBlueGED]

On 7/15/2019 at 1:31 PM, triumph_communes said:

Daily dosage of grains of salt:

 

The curious story of Justin Schultz:

 

 

RAPM charts still require CONTEXT-- the models don't explain it all

In this case, the context is Schultz had a career year and had considerable drop-off in 2017-18: 

Then again in 2018-19: 

But hey, good on you for picking out the obvious outlying season in an ill-fated attempt to show Risto will blossom elsewhere. Schultz was great in 2016-17, but it's pretty clear that season is not representative of who Schultz is as a player. 

Again, the argument isn't that context is irrelevant, it's that context is not transformative.