Canucks Analytics

Discussion in 'Vancouver Canucks' started by Bertuzzzi44, Jan 27, 2019.

  1. Melvin

    Melvin Yes.

    Joined:
    Sep 29, 2017
    Messages:
    9,204
    Likes Received:
    11,686
    Trophy Points:
    105
    Gender:
    Male
    Occupation:
    Data Nerd
    Location:
    Houston, TX
    Like I said it's more of a pet peeve than anything. If you change inputs to a formula but don't change the formula, you can't now declare that your formula is "advanced." the formula is not advanced and has not advanced, we just expanded the type of events to put into it and then gave it a stupid name and declared that its advanced.

    Anyway I mostly get irritated because someone will say "that guys has good advanced stats" or "that guy has good analytics" and all they mean is "that guy has good shot attempt +/-." Analytics are a million times broader than that.

    Like you would never hear someone in baseball say a player has "good analytics" instead of just saying he has a high WAR. But this happens in hockey all the time. Saying a player has "good analytics" is literally gibberish. One highly flawed, noisy metric is not analytics.
     
    Last edited: Feb 9, 2019
    Hindustan Smyl likes this.
  2. PuckMunchkin

    PuckMunchkin Registered User

    Joined:
    Dec 13, 2006
    Messages:
    5,808
    Likes Received:
    2,280
    Trophy Points:
    156
    Location:
    Lapland
    What about the rest of the data that gets collected. You seem to act like shots is the only thing being tracked..?
     
  3. Melvin

    Melvin Yes.

    Joined:
    Sep 29, 2017
    Messages:
    9,204
    Likes Received:
    11,686
    Trophy Points:
    105
    Gender:
    Male
    Occupation:
    Data Nerd
    Location:
    Houston, TX
    Look at the OP. None of the "advanced stats" are anything more than +/- applied to shots and expressed as a % instead of a delta, and given fancy sounding names like "Fenwick."

    Sorry, I just find it dumb and annoying. And we've already spent far more time on what jve already told you is a pet peeve.
     
    Hindustan Smyl likes this.
  4. Melvin

    Melvin Yes.

    Joined:
    Sep 29, 2017
    Messages:
    9,204
    Likes Received:
    11,686
    Trophy Points:
    105
    Gender:
    Male
    Occupation:
    Data Nerd
    Location:
    Houston, TX
    I fixed my issues with those "impact" metrics. It's very mathy and boring so I won't go into it, but it had to do with how I was doing the weighted averages.

    Here is the new data, using Shot Attempts, AKA Corsi.

    As a quick tl;dr: This is the average effect that each player has on his teammate's SAT+/-, by comparing "with" to "without." A low score means that the player tends to bring down their teammates relative to when the teammates are not playing with the player.

    Canucks:

    EvaluatedPlayerAverageImpact (SAT)
    ERIK GUDBRANSON-38.5
    BRANDON SUTTER-23.51
    CHRISTOPHER TANEV-23.48
    JAY BEAGLE-16.39
    TYLER MOTTE-15.82
    BEN HUTTON-12.73
    ALEXANDER EDLER-8.89
    SVEN BAERTSCHI-5.95
    ADAM GAUDETTE0.09
    BO HORVAT1.24
    BRENDAN LEIPSIC1.52
    JAKE VIRTANEN2.98
    LOUI ERIKSSON3.54
    MARKUS GRANLUND3.62
    NIKOLAY GOLDOBIN3.77
    TIM SCHALLER4.04
    DERRICK POULIOT10.6
    MICHAEL DEL ZOTTO11.07
    ALEX BIEGA12.37
    BROCK BOESER13.25
    ANTOINE ROUSSEL18.44
    ELIAS PETTERSSON19.16
    JOSH LEIVO21.34
    TROY STECHER24.95

    League:

    EvaluatedPlayerTeamAverageImpact
    MARC-EDOUARD VLASICS.J-46.47
    JUSTIN BRAUNS.J-45.46
    BRETT HOWDENNYR-42.6
    ERIK GUDBRANSONVAN-38.5
    NEAL PIONKNYR-37.39
    RON HAINSEYTOR-36.57
    PATRIK LAINEWPG-35.16
    TOM PYATTOTT-34.74
    NICK BONINONSH-34.66
    CODY EAKINVGK-34.33

    That top 2 is certainly interesting; both players have had much better results apart from one another as they have had together:

    upload_2019-2-9_18-20-26.png


    upload_2019-2-9_18-21-1.png

    Pretty interesting. I don't follow the Sharks so not sure if these results are just a fluke or if they have really struggle together, or if there is anything else going on.

    Top 10:

    EvaluatedPlayerTeamAverageImpact
    MARK STONEOTT67.16
    ARTEMI PANARINCBJ64.63
    SIDNEY CROSBYPIT59.3
    DYLAN LARKINDET53.88
    SEAN COUTURIERPHI49.99
    BRENDAN GALLAGHERMTL48.95
    TOMAS TATARMTL47.11
    JAKE GUENTZELPIT43.92
    NATHAN MACKINNONCOL43.19
    RYAN JOHANSENNSH42.94
     
    PuckMunchkin, Peter10 and Dana Murzyn like this.
  5. Ronning On Empty

    Ronning On Empty Formerly BleachClean

    Joined:
    Aug 9, 2006
    Messages:
    23,850
    Likes Received:
    3,140
    Trophy Points:
    186
    Stecher being the best surprises me.

    Also, Larkin at 53.88 and Horvat at 1.24... wow. I need to change my vote in that poll.
     
  6. F A N

    F A N Registered User

    Joined:
    Aug 12, 2005
    Messages:
    8,515
    Likes Received:
    1,130
    Trophy Points:
    139
    Nice post. Vlasic has certainly declined. Laine is the real surprise for me, even though his struggles this season isn't a secret.
     
  7. Hindustan Smyl

    Hindustan Smyl 2022/2023 = Canucks being a contender

    Joined:
    Jun 21, 2014
    Messages:
    7,636
    Likes Received:
    1,473
    Trophy Points:
    119
    Gender:
    Male
    Occupation:
    Manager + Lecturer
    Location:
    Beijing
    Interesting lists.

    One question - how much of the above is related to on ice deployment?
     
  8. Hindustan Smyl

    Hindustan Smyl 2022/2023 = Canucks being a contender

    Joined:
    Jun 21, 2014
    Messages:
    7,636
    Likes Received:
    1,473
    Trophy Points:
    119
    Gender:
    Male
    Occupation:
    Manager + Lecturer
    Location:
    Beijing
    Guys like Vlasic, Sutter, Beagle, and Tanev are defensive-oriented players however and are used accordingly as a result.

    That’s why I would like to know more about this statistic......and how it combines with the on ice deployment and role of said players.
     
  9. Melvin

    Melvin Yes.

    Joined:
    Sep 29, 2017
    Messages:
    9,204
    Likes Received:
    11,686
    Trophy Points:
    105
    Gender:
    Male
    Occupation:
    Data Nerd
    Location:
    Houston, TX
    Alright one more analytics post for me! I am burning myself out researching like a million different things at once, lol. I think I've had a buildup of projects I've wanted to do the last 5 years that I've just finally had the time to do and now I'm hitting them all. So apologies for the spam.

    Upthread I posted data on shot probabilities based on team-strength, shot type and distance-from net. the idea was to quantify the top shooters by making the most out of their shots.

    Now, the next step is to apply the same methodology for goalies. Here is the same graph, with goals expressed as a % of shots-on-goal this time:

    upload_2019-2-10_0-51-6.png

    Since a lot of deflections miss the net, it makes sense that the curve when using SOG as denominator instead of SAT is closer to the regular curve.

    Based on this, you can calculate the overall % chance of a shot being stopped and then sum it up for all goalies to get their expected goals allowed vs actual:

    (Note my data from tonight's games is still updating so tonight's games are not included. I think Markstrom will be #1 after it's done. :D)

    GoalieExpected GoalsGoals+/-Sv%Exp. Sv%Sv% +/-
    JOHN GIBSON146123230.9140.8980.016
    ROBIN LEHNER7656200.9320.9070.025
    THOMAS GREISS7659170.9260.9040.022
    BEN BISHOP9176150.9240.9090.015
    LAURENT BROSSOIT4127140.9400.9100.030
    JIMMY HOWARD10896120.9140.9040.011
    COLLIN DELIA4634120.9230.8960.027
    FREDERIK ANDERSEN10594110.9230.9140.009
    ANDREI VASILEVSKIY8877110.9270.9170.010
    JACK CAMPBELL4434100.9330.9130.020

    John Gibson is arguably having a very underrated season because of the horrible play of his team. Based on this methodology he has done a better job stopping shots based on the types of shots he's faced and the volume than any other goalie. In fact, he was #1 last year too. His .898 expected save% is among the lowest in the NHL as he has simply had to face a lot of difficult shots.

    Collin Delia, who stopped 40-of-43 against the Canucks the other night has also had fantastic results thus far in just 12 games to already be +12.

    ...trailers...

    GoalieExpected GoalsGoals+/-Sv%Exp. Sv%Sv% +/-
    ALEX STALOCK2736-90.8910.918-0.027
    ROBERTO LUONGO6575-100.8970.911-0.014
    CORY SCHNEIDER2030-100.8670.914-0.046
    ANTTI NIEMI4354-110.8930.914-0.021
    MICHAL NEUVIRTH1526-110.8590.917-0.058
    MIKE SMITH6678-120.8890.905-0.017
    CRAIG ANDERSON105117-120.9040.913-0.010
    PHILIPP GRUBAUER5769-120.8890.908-0.019
    CAM TALBOT7690-140.8950.911-0.016
    KEITH KINKAID93111-180.8960.913-0.017

    Ooof some familiar names there for Canucks fans. Anderson was a disaster for Ottawa as the play of Nilsson has been demonstrating lately. Based on the shots faced he should have had a closer-to-average save%.

    Speaking of Nilsson:


    GoalieExpected GoalsGoals+/-Sv%Exp. Sv%Sv% +/-
    ANDERS NILSSONVAN3437-30.8950.903-0.007
    ANDERS NILSSONOTT241950.9340.9160.018

    The guy definitely seemed to have a raw deal in Vancouver, as you can see from the .903 expected save% vs .916 in Ottawa. It isn't a lot of data and not including tonight but supports what @Jyrki21 was saying about it being as much a change in context as a change in performance as he's seemingly had an easier time in Ottawa thus far.
     
    Jyrki21, MadaCanuckle and Dana Murzyn like this.
  10. MadaCanuckle

    MadaCanuckle Registered User

    Joined:
    Jun 25, 2012
    Messages:
    1,861
    Likes Received:
    590
    Trophy Points:
    109
    Gender:
    Male
    Location:
    Lisboa
    Melvin, i would live to see the data on Marky.
     
  11. Melvin

    Melvin Yes.

    Joined:
    Sep 29, 2017
    Messages:
    9,204
    Likes Received:
    11,686
    Trophy Points:
    105
    Gender:
    Male
    Occupation:
    Data Nerd
    Location:
    Houston, TX
    Here's everyone since 2014-15.

    SeasonGoalieTeamExpected GoalsGoalsGoals Saved Above AverageSv%ExpectedSv%Sv% +/-
    20142015EDDIE LACKVAN11994250.9220.9010.021
    20152016RYAN MILLERVAN159136230.9170.9030.014
    20162017RYAN MILLERVAN163150130.9140.9060.008
    20152016JACOB MARKSTROMVAN938490.9150.9060.009
    20162017JACOB MARKSTROMVAN716290.9110.8980.013
    20182019JACOB MARKSTROMVAN12711980.9100.9040.006
    20172018JACOB MARKSTROMVAN16015370.9120.9090.003
    20142015RYAN MILLERVAN11210750.9110.9060.005
    20162017RICHARD BACHMANVAN161330.9200.9000.020
    20152016RICHARD BACHMANVAN3300.9030.914-0.011
    20182019THATCHER DEMKOVAN3300.9230.9150.008
    20142015JACOB MARKSTROMVAN34-10.8790.916-0.037
    20172018THATCHER DEMKOVAN34-10.8670.903-0.036
    20182019ANDERS NILSSONVAN3537-20.8950.902-0.007
    20182019RICHARD BACHMANVAN26-40.7930.925-0.132
    20172018ANDERS NILSSONVAN7984-50.9010.907-0.006
     
    racerjoe and MadaCanuckle like this.
  12. MadaCanuckle

    MadaCanuckle Registered User

    Joined:
    Jun 25, 2012
    Messages:
    1,861
    Likes Received:
    590
    Trophy Points:
    109
    Gender:
    Male
    Location:
    Lisboa
    Excellent. It seems our goaltending has been a strength for our team. I would say a lot about the Miller signing but I don't want to derail this thread.
     
  13. zcaptain

    zcaptain Registered User

    Joined:
    Apr 4, 2012
    Messages:
    672
    Likes Received:
    128
    Trophy Points:
    61
    Numbers are interesting...….

    What I would like to see is a analytic for

    player production X team-mate production X offensive starts X defensive starts X Opposition quality*

    *Opposition Quality being same calculation
     
  14. vanuck

    vanuck But Gillis...!!!

    Joined:
    Dec 28, 2009
    Messages:
    14,718
    Likes Received:
    710
    Trophy Points:
    139
    Thought it was time to bring this back...

    What do people make of WAR? I.e How accurate and useful is it as a stat if you were to look back at previous predictions? Is it more useful at a team level than individual? As someone who has always been a supporter of using analytics as an evaluative tool, WAR is relatively unfamiliar to me and it would be great if we talked about it more. Even GAR...

    This article from Canucks Army for instance is using an example of WAR to estimate how well the team will do this season, though it's obviously a little optimistic given that there's essentially zero chance of those 4 lines/3 pairings staying together all year:

    https://canucksarmy.com/2019/08/03/getting-the-most-points-out-of-a-canucks-lineup/
     
  15. PuckMunchkin

    PuckMunchkin Registered User

    Joined:
    Dec 13, 2006
    Messages:
    5,808
    Likes Received:
    2,280
    Trophy Points:
    156
    Location:
    Lapland
    WAR is the one new stat I feel most uncomfortable with. Feels like you would need to control for SO MANY things for it to mean anything.
    And I do mean "feels like" as opposed to it necessarily being the case, since I honestly dont completely understand the way it is calculated...
     
  16. Norade

    Norade Registered User

    Joined:
    Nov 1, 2011
    Messages:
    1,296
    Likes Received:
    359
    Trophy Points:
    94
    Location:
    Vancouver, BC
    Is the exact formula and weighting system even publically available? I went looking for it couldn't find any concrete data. Without that, honestly, WAR just seems like a black box where you have to trust that the person who made it put together correctly.
     
    PuckMunchkin likes this.
  17. RonningMorrisonBooth

    RonningMorrisonBooth Support the boys!

    Joined:
    Jun 21, 2017
    Messages:
    2,750
    Likes Received:
    969
    Trophy Points:
    94
    How much stock can we put in a metric that would suggest that Derek Pouliot is so much better than Chris Tanev that the gap is wider than the gap is between Tanev and Gudbranson?
     
  18. Melvin

    Melvin Yes.

    Joined:
    Sep 29, 2017
    Messages:
    9,204
    Likes Received:
    11,686
    Trophy Points:
    105
    Gender:
    Male
    Occupation:
    Data Nerd
    Location:
    Houston, TX
    Took me a little but longer than it should have but I have the WOWY tool up and running for the new season. Also think I've fixed some performance issues. Let me know if you find any kinks.
     
    vanuck and ccjon like this.
  19. Norade

    Norade Registered User

    Joined:
    Nov 1, 2011
    Messages:
    1,296
    Likes Received:
    359
    Trophy Points:
    94
    Location:
    Vancouver, BC
    It seems to be down at the moment.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice
monitoring_string = "358c248ada348a047a4b9bb27a146148"