2017-18 stats and underlying metrics thread [Mod: updated season]

Discussion in 'Winnipeg Jets' started by garret9, Aug 2, 2014.

Thread Status:
Not open for further replies.
View Users: View Users
  1. garret9

    garret9 AKA#VitoCorrelationi

    Joined:
    Mar 31, 2012
    Messages:
    20,358
    Likes Received:
    889
    Trophy Points:
    139
    Location:
    Vancouver
    Home Page:
    In fact, Corsi out performs possession.

    Which makes sense... Hockey (from offense perspective) in steps is:
    1) Get possession
    2) Maintain possession
    3) Create opportunity
    4) Shoot
    5) Repeat
     
  2. garret9

    garret9 AKA#VitoCorrelationi

    Joined:
    Mar 31, 2012
    Messages:
    20,358
    Likes Received:
    889
    Trophy Points:
    139
    Location:
    Vancouver
    Home Page:
    This is factually false and it is easily proven that Corsi is useful:
    Remembering Dellow: A few graphs to convince you on Corsi

    Now, yes, there are outside factors... but that just means that Corsi becomes even better if you adjust for that....

    Teams do play different with a lead, and we can and have made Corsi better by adjusting for that:
    Adjusted Possession Measures

    Teams actually don't actually play that differently with different refs to any significant importance. That hasn't shown any credibility.

    Teams play differently depending on rest, and we can and have made Corsi better by adjusting for that:
    Schedule Adjustment for Counting Stats

    There is more than one player on the ice, and we can look at how player's impact linemates and competition using models like XPM which simply adjusted Corsi for shot quality, linemates, competition, coaches, and zone starts.

    These "too moo many variables" are really just "variables". These variables can be accounted for.

    Corsi was already shown to have value. A team with a good Corsi at game 20 that is winning is more likely to continue winning than a team with bad Corsi at game 20 that is winning. At a player level, the same thing is true if you replace "winning" with "out scoring."
    To say something that is "a signal on whether success is likely to be maintained or not" is useless would be just flat out wrong.

    Shot volume is a major part of what causes teams to win. It's not the only part, but a major part.
    [​IMG]
     
    Last edited: Jan 5, 2018
  3. Halberdier

    Halberdier Registered User

    Joined:
    May 14, 2016
    Messages:
    1,798
    Likes Received:
    1,012
    Trophy Points:
    94
    Do you mean that the statistical correlation between possession and actual scoring is even less than 0.18? That might well be as straight counter attacks, often with odd man rush are so much more valuable than just possessing the puck against defense that has had enough time to make their formation straight.

    You don't do anything with possession if you can't get high quality scoring changes in large enough quantities. You can still be interested to measure the possession and o-zone time as that nevertheless tells you more about the game.

    Also you do nothing with Corsi, if that consists only from low quality perimeter shots.
     
  4. garret9

    garret9 AKA#VitoCorrelationi

    Joined:
    Mar 31, 2012
    Messages:
    20,358
    Likes Received:
    889
    Trophy Points:
    139
    Location:
    Vancouver
    Home Page:
    Who only takes low quality perimeter shots?

    Aside:
    Does low quality perimeter shots reduce the probability of producing the next shot?
    Does low quality perimeter shots increase the probability of the other team producing the next shot?
    Is there overlap in the talent being talent in producing more shots while preventing opponent shots with producing better shots and preventing better shots?
    *Spoiler* The answer is yes.
     
    Last edited: Jan 5, 2018
  5. Ducky10

    Ducky10 President Patrik Laine Fan Club Sponsor

    Joined:
    Nov 14, 2014
    Messages:
    9,634
    Likes Received:
    5,029
    Trophy Points:
    111
    You don't do anything with possession if you don't, you know.....shoot.

    I'd say the chart above makes a pretty convincing statement about volume being important.
     
  6. Ducky10

    Ducky10 President Patrik Laine Fan Club Sponsor

    Joined:
    Nov 14, 2014
    Messages:
    9,634
    Likes Received:
    5,029
    Trophy Points:
    111
    Chiarot?
     
  7. Joe Hallenback

    Joe Hallenback Registered User

    Joined:
    Mar 4, 2005
    Messages:
    9,910
    Likes Received:
    1,243
    Trophy Points:
    139
    Shooting is important

    I mean I can hear myself screaming inside my head at my kids high school team to Shoot
     
  8. garret9

    garret9 AKA#VitoCorrelationi

    Joined:
    Mar 31, 2012
    Messages:
    20,358
    Likes Received:
    889
    Trophy Points:
    139
    Location:
    Vancouver
    Home Page:
    Yup.

    It's a weird debate. It's not like anyone has argued that shot differentials tell the whole story... just that it's kinda important to shoot, so having more is good than bad.
     
    YWGinYYZ and csk like this.
  9. Halberdier

    Halberdier Registered User

    Joined:
    May 14, 2016
    Messages:
    1,798
    Likes Received:
    1,012
    Trophy Points:
    94
    Of course not. But let's say it might be more valuable to keep the possession until you get a high quality scoring change and only then shoot than to just shoot from every angle all the time and risk losing the possession. That is crucial in 3-on-3, but there are several hockey teams I know that have been using this approach (trying to be very selective on shooting) also 5-on-5 with really good results. Don't actually know if there are many teams in NHL even trying that though. In the end keeping puck possession also tends to lower the CA, so Corsi against gets affected in the process, that's true.

    There are a lot of different takes about how to play the hockey. Most common ways will be very similar to each others, so therefore correlation between Corsi and real results will be more higher among these "median hockey style teams". And again, this doesn't mean that you could not have an approach to the hockey where you trade a lot of quantity over quality. You might also try to force your opponent to shoot a lot from the perimeter, which might result them "winning the Corsi", but losing for example those shooting heat maps hands down. I don't know if we won Corsi against Colorado last match, but looking at where those shots were taken, it's not a surprise we did not get too much 5-on-5 goals against them.

    Above chart is awesome, though lacks any sources and includes infamous scientific factor called "Luck" with huge percentage, as the variance evens out when the sample size increases. Also 44% is not looking quite right to me.



    Raw Corsi doesn't convince me as a super relevant stat just because it lacks the quality factor. I have never opposed using super advanced "Corsi++++++" that is adjusted by all kinds of quality factors, though using the name "Corsi" with them is more than a stretch. It's like using the name of "+/-" of something like "WAR-adjusted +/-". I'm only against the blind raw Corsi (and raw rel Corsi) gospel and claims that "real goals are irrelevant / too random" that are like a plague in all hockey forums. Corsi is invaluable tool if you know your playing style, can estimate your quality factors at least roughtly (eg. long time SH% etc.) and take those into account. Or just see how the Corsi evolves during your gameplay and use that as a some sort of "first quick indicator" of upcoming problems. Like it's quite bad if your line gets 3 shots in a game against 15 by opponent. There almost can't be a quality factor that could turn that kind of Corsi difference a success givesn that both are pro teams, but only sometimes you can get away with that.

    To summarize you can't look just Corsi of two teams or (rel Corsi) of two players and tell that this team or this guy is better. You need to look real results (goals -> wins -> points) for that. You can combine real results and development of Corsi and then more or less effectively quesstimate about the future success of those teams or those players.
     
  10. Halberdier

    Halberdier Registered User

    Joined:
    May 14, 2016
    Messages:
    1,798
    Likes Received:
    1,012
    Trophy Points:
    94
    Sometimes a team that is forced to do that by an opposite team that is superior in defending the center of the d-zone. But rarely that's "only", but often that's "mostly".

    Yes for all those questions. And those two first are exactly the reason why some teams have been more keen to increase the quality than the quantity and actually be (maybe even too) selective with their shots. They did quite exhaustive and scientific statistical studies about the subject in Finland and then at least one successful coach and his successor did apply that quite religiously at some point winning some medals (3 silver and 2 gold in 5 years) during the process. Though during the process they (Tappara Tampere) lost against high volume perimeter shooter team Ässät Pori (Joel Armia, Antti Raanta), but that was at least partially caused by injury of couple of their top guys (including Alexander Barkov) and so called prison rules which did allow Ässät to prevent Tappara from entering the center of offensive zone almost completely, at least with their helmets on their heads, and playing without helmet was prohibited already by then. Tappara did lose three consecutive FEL playoffs finals before finally winning two consecutive championships (the first with Patrik Laine), being most of the time very selective shooters often losing the Corsi handily (though probably being much better with sCorsi). This season they have again adopted new tactics, and they have been much less selective shooting high Corsi / low SH% team and are currently 3rd in the league.

    There are high event teams, low event teams, high quality event teams and low quality event teams. There are teams that are relatively low event but high quality Corsi for teams that allow relatively high quantity but low quality Corsi events against. Every possible style might lead to success.
     
  11. Ducky10

    Ducky10 President Patrik Laine Fan Club Sponsor

    Joined:
    Nov 14, 2014
    Messages:
    9,634
    Likes Received:
    5,029
    Trophy Points:
    111
    But nobody has said you can just look at the Corsi of two teams or two players and decide who is better. I think that's been stated numerous times, as recently as two posts ago.
     
  12. Halberdier

    Halberdier Registered User

    Joined:
    May 14, 2016
    Messages:
    1,798
    Likes Received:
    1,012
    Trophy Points:
    94
    It's said all the time all over the forums. Maybe not this much in this thread and never by ones like Garret who actually knows about the subject.

    Corsi has really been like a silver bullet for many, and ignoring real results another silver bullet for many. Prime example being the article linked in OP with that absurd "90% is Corsi + luck" claim.

    Also when the season did go on and Jets had bad Corsi, yet they won quite easily many games there was a lot of doomsday mentality over the subject. There were many having a consensus about the subject how the winning had to be unsustainable and only because of great goaltending by Helle. Yet in the reality there was only 0.18 average statistical correlation between Corsi and real goals when couple of seasons had been analyzed statistically by Whilee. What that does mean is that you might have a team that consistently loses Corsi (not like 100-0, but let's say like 45-55) and still "magically" "with pure luck and goaltending" happens to win most of their games. That is considered witchcraft, sustained unsustainability or just pure luck by so many people on these forums. And that is what I'm opposing, and that is why I'm not at all into comparing raw Corsi between different teams and players.
     
  13. lomiller1

    lomiller1 Registered User

    Joined:
    Jan 13, 2015
    Messages:
    4,193
    Likes Received:
    567
    Trophy Points:
    94

    The formula for producing goals in the NHL is to take a "low quality" perimeter shot and look for tips and rebounds. Teams that hold onto the puck looking for the perfect scoring opportunity generally don't do very well.
     
  14. garret9

    garret9 AKA#VitoCorrelationi

    Joined:
    Mar 31, 2012
    Messages:
    20,358
    Likes Received:
    889
    Trophy Points:
    139
    Location:
    Vancouver
    Home Page:
    The graph is from an athletic article I wrote, which has been posted a few pages ago, and predominately comes through multiple sources that overlap and (informally) peer reviewed each other.

    "Luck" is actually fairly scientific in this case, which is natural variance not being evenly distributed plus a portion of stochastic system. There are two very different ways that have shown exactly 38% "luck" in 82 game season standings.
    (I made the graph only 33% to purposefully underestimate luck)

    1) Desjardins pointed out that even if NHL games were decided by a 50/50 coin toss, there would be some teams better than others in the standings. No team controls their destiny as hockey is a tug-a-war. Probability is just a weighted coin toss (albeit moving weighting) coin toss. So, we expect some percentage of the standings to be pushed by the coin flip as well.
    2) Wiessbock used machine learning to determine the theoretical limit to predicting the totatility of a full season.
    (If you want to see some fun on the shortened season with luck, check this out)

    It's not that goals are irrelevant, just less telling than Corsi. Goals are powerful, and each goal contains about 5x the information as a shot (ie: higher correlation coefficient), but their rarity causes a weighted shot differential with weighting goals heavier have about a +0.98 correlation coefficient with non-weighted.

    Raw Corsi is superior to raw goals and is entirely useful. Garnering a shot is part of the process.

    You want to maximize your number of opportunities (shot volume), maximize your chance in each opportunity (shot quality), and capitalize on those chances (finishing talent). The defensive side of the sport is merely the reciprocal. You want to minimize your opponent's number of opportunities (shot volume), minimize your opponent's chance in each opportunity (shot quality), and stop their chances (goaltending). All other actions either lead up to or are part of these inputs in outscoring the opposition.
     
    Last edited: Jan 5, 2018
  15. garret9

    garret9 AKA#VitoCorrelationi

    Joined:
    Mar 31, 2012
    Messages:
    20,358
    Likes Received:
    889
    Trophy Points:
    139
    Location:
    Vancouver
    Home Page:
    Not even mostly... the truth is the distribution in shots is quite similar between teams. In fact, if you add ONLY shot distribution to Corsi as a shot quality factor, it is often *less* predictive than Corsi.

    Why is that?

    It's because while you can outperform your shots in the short duration through shooting more in prime areas or allowing fewer in the prime areas, but you are unlikely to consistently do so.

    Every team that vastly outperformed their shot differentials were because of shot location - possession teams and/or being counter-attack team son the rush. All of these teams were the classic teams that fell in the standings significantly either in the season or the following.
     
  16. Halberdier

    Halberdier Registered User

    Joined:
    May 14, 2016
    Messages:
    1,798
    Likes Received:
    1,012
    Trophy Points:
    94
    If there is a good net front presence and correct timing that "low quality" perimeter is a high quality perimeter shot by definition.

    That poses a problem for shot heat maps that I am otherwise avid fan of.

    Without net front presence or screen, low quality perimeter shot remains low quality perimeter shot.

    Personally I try to keep it simple and just throw the puck towards the net if there is some traffic, as I'm a sub-beer league level defenceman lacking talent and being kind of combination of mobility of "sub beer league Hendricks" and vision and hands of "sub beer league Tanev". If I had vision and hands I would have other, more high quality options available from time to time, but given that there is some traffic in fron of the net, it's never completely wrong to just send the puck there.
     
    Last edited: Jan 6, 2018
  17. kcin94

    kcin94 Registered User

    Joined:
    Jul 17, 2011
    Messages:
    603
    Likes Received:
    161
    Trophy Points:
    61
    The end. Right here. It can be said over and over again that statistics is a tool and can never be used alone. Yet just as much as people use the number stupidly as you claim, there are others on the other side that dismiss the numbers as "useless" because they don't tell everything
     
  18. Aavco Cup

    Aavco Cup "I can make you cry in this room"

    Joined:
    Sep 5, 2013
    Messages:
    37,630
    Likes Received:
    10,354
    Trophy Points:
    142
    This article includes these graphics about Ehlers Zone entry prowess, which were previously posted in twitter so I'm including them as being outside the paywall.

    [​IMG]
    [​IMG]


    How Winnipeg's new top line of Perreault-Little-Ehlers...

    Since losing Mark Scheifele to injury on December 27, the Winnipeg Jets have the NHL's second-best record.
    That Winnipeg is 11-2-3 in 16 games without its No. 1 centre is a testament to many factors. Goaltending, special teams, and puck luck have all played a role but, to me, the biggest reason for Winnipeg’s success has been the play of its new top line.
    In Winnipeg’s 16 games since Scheifele’s injury, Bryan Little, Nikolaj Ehlers, and Mathieu Perreault have dominated possession at even strength, earning 60.7 per cent of the shot attempts, 64.3 per cent of high danger scoring chances, and 62.7 per cent of expected goals. Without Scheifele, these three players (and not Blake Wheeler) have led Jets forwards in even strength ice time per game (Wheeler is fourth.) They've also split tough minutes with Adam Lowry's line(when it was healthy) and, while doing so, have the highest relative Corsi of any line the Jets have iced all season long.
    To put that relative Corsi stat another way: strictly in terms of shot attempts, Perreault/Little/Ehlers have been better than their teammates by a larger margin than any Jets line since opening night.
     
    SLAYER, MardyBum and YWGinYYZ like this.
  19. CaptainChef

    CaptainChef Registered User

    Joined:
    Jan 5, 2014
    Messages:
    7,278
    Likes Received:
    289
    Trophy Points:
    79
    Location:
    Bedroom Jetsville
    Interesting. I heard PMo raving at how far Ehlers game has improved since Scheif went down, but those numbers are staggering. I guess we all could see that that the Laine-Little combo was terrible, but who'd of ever guessed that switching Perrault & Laine would have had that much effect.
     
  20. Aavco Cup

    Aavco Cup "I can make you cry in this room"

    Joined:
    Sep 5, 2013
    Messages:
    37,630
    Likes Received:
    10,354
    Trophy Points:
    142
    Perreault could be our most important player. We need him healthy in the playoffs.
     
    libertarian and Saidin like this.
  21. Gm0ney

    Gm0ney Unicorns salient

    Joined:
    Oct 12, 2011
    Messages:
    9,687
    Likes Received:
    1,715
    Trophy Points:
    127
    Location:
    Winnipeg
    I could see Connor coming to dominate successful zone entries like Ehlers - hopefully in the next year or two (or sooner...I won't complain).
     
  22. DashingDane

    DashingDane Registered User

    Joined:
    Dec 16, 2014
    Messages:
    1,438
    Likes Received:
    917
    Trophy Points:
    113
    Location:
    Los Angeles
    Really good article! So... basically we should keep that line together :popcorn:
     
  23. csk

    csk Registered User

    Joined:
    Nov 5, 2015
    Messages:
    2,658
    Likes Received:
    248
    Trophy Points:
    66
    Location:
    Winnipeg, MB
    It's really about the Perreault-Little duo, as they did damn fine with Wheeler as well.
     
  24. DashingDane

    DashingDane Registered User

    Joined:
    Dec 16, 2014
    Messages:
    1,438
    Likes Received:
    917
    Trophy Points:
    113
    Location:
    Los Angeles
    Fixed it for you ;). I think that's a very simplistic view you have. Ehlers has been good with Wheeler as well... and with Chef, Frenchie, Laine ect.. It all depends on players being in a good rhythm/hot. No one will be good for a full season. We could flip it and say they also all had bad games with Wheels. Nothing in hockey is black and white.

    On a side note. I don't get the whole it's player A fault the line is a good statement. You need all three players to be good or it won't work. Sure a good player makes a line better but hockey is a team sport so you won't accomplish anything unless they all work well together (look at McJesus). We keep hearing Chef + Wheels = gold or Chef +Laine = gold but all conveniently ignore the third forward that might not put up as many pts but contributes in many other ways.
     
    Mortimer Snerd likes this.
  25. winnipegger

    winnipegger Registered User

    Joined:
    Dec 17, 2013
    Messages:
    5,928
    Likes Received:
    2,092
    Trophy Points:
    111
    Are you saying that you can predict a team's results 62% of the time, but 38% you are going to be lead astray by unevenly distributed variance ("luck")? Has this been tested?

    I tend to get hung up on value and how it's derived. For example, saying shot quality drives 10% of the output. There's a lot of wiggle room on what a quality shot is, no?
     
Thread Status:
Not open for further replies.

Share This Page

monitoring_string = "358c248ada348a047a4b9bb27a146148"