NHL goalies better vs high shot volumes

Discussion in 'By The Numbers' started by Mathletic, Mar 26, 2014.

  1. SladeWilson23 Black Lives Matter

    Joined:
    Nov 3, 2014
    Messages:
    26,242
    Likes Received:
    2,602
    Trophy Points:
    156
    SB Cash:
    $ 100,000
    Gender:
    Male
    Location:
    New Jersey
    I'd love to hear your response.

    That's fair, but even if we did include those games and extrapolate them out to full 50+ minute games, it wouldn't tip the scales that much in the other direction.

    But we have to keep each game individual because every game is a unique event. There's a ton of varying factors that affect goalie performance game by game.

    Not all teams are equal, and not all goalies are equal. It has to be team to itself, and goalie to himself comparisons. Game to game data is really the only way to do it because aggregate data does not take into account all the variables.
     
  2. SladeWilson23 Black Lives Matter

    Joined:
    Nov 3, 2014
    Messages:
    26,242
    Likes Received:
    2,602
    Trophy Points:
    156
    SB Cash:
    $ 100,000
    Gender:
    Male
    Location:
    New Jersey
    Here's the splits for when TEAMS allow 30 or more shots versus allowing 29 or fewer shots in the past 5 years.

    0-29 SHOTS: .895 SV%
    30+ SHOTS: .917 SV%

    Here are the gamelogs from 2013 through 2018.


    I would like to just clarify something about this table.
     
    Last edited: Jan 5, 2019
  3. morehockeystats Unusual hockey stats

    Joined:
    Dec 13, 2016
    Messages:
    487
    Likes Received:
    158
    Trophy Points:
    56
    SB Cash:
    $ 100,000
    Occupation:
    sysadmin
    Location:
    Columbus
    Home Page:
    Damien Dahrk likes this.
  4. Filthy Dangles .

    Joined:
    Oct 23, 2014
    Messages:
    22,026
    Likes Received:
    26,162
    Trophy Points:
    186
    SB Cash:
    $ 100,000
    Location:
    Greenland
    As I understand it, there's conflicting views and evidence on how these things are connected. Was wondering what this Board has to say about it.

    Also, I remember a thread that had some good info, think it was on the History Board, and it pertained to binning games based on how many shots the goalie faced which showed that SV% rose with Shots against; but posters broke down how there were problems with that analysis and what not. Wasn't able to find it though, if someone could, that might be helpful.
     
    Bear of Bad News likes this.
  5. Bear of Bad News HFBoards Escape Goat

    Joined:
    Sep 27, 2005
    Messages:
    7,648
    Likes Received:
    5,833
    Trophy Points:
    201
    SB Cash:
    $ 100,000
    Location:
    Windsor
    Awards:
    Many things in this thread (now merged). Tables appear to have been boofed. Enjoy!
     
    Filthy Dangles likes this.
  6. Hockey Outsider Registered User

    Joined:
    Jan 16, 2005
    Messages:
    6,762
    Likes Received:
    5,315
    Trophy Points:
    216
    SB Cash:
    $ 100,000
    I'll have to see if I can dig up the threads (they may have been pre-2017 data migration), but the evidence is quite straightforward.

    At the individual game level, it appears that there's a positive relationship between shots faced and save percentage. In other words, goalies who face more shots in a single game are more likely to have a higher save percentage.

    At the season or career level, there's essentially zero relationship between shots faced and save percentage.

    This sounds confusing because how can something be true at the game level, and not at the seasonal level? There are two explanations:

    1. There's a wide range in shots faced at the individual game level. It's rare, but not really exceptional, to see a goalie face <20 shots, or >35 shots, in a single game. (From 2006 to 2020, in the regular season, there have been 171 instances of a goalie facing at least 50 shots in a game; and 278 instances of a goalie facing no more than 15 shots in a game - with a minimum of 50 minutes played so the data isn't getting skewed from goalies who are pulled early). You simply don't get that range in shots at the seasonal level, where usually the spread from most to fewest shots faced per game is perhaps 10 shots. So maybe this phenomenon would hold true if one goalie averaged 18 shots per night over the course of a season, and another averaged 43; but that simply doesn't happen at the NHL level. The phenomenon, if it really does exist, is ultimately irrelevant because the spread in shots faced (at the season and career level) is so small.

    2. There are likely game-specific effects that, by definition, don't impact an entire season. It's generally accepted that losing teams will try to shoot the puck more, and often these are less dangerous shots (ie throw the puck on net from the point). So it might look like the goalie is doing better when he's facing more shots - but the real reason is he's facing more, easier shots which inflate his save percentage, but that has nothing to do with his actual level of performance.

    My conclusion, which I think is supported very clearly by the data, is you shouldn't draw conclusions from a single game's save percentage (especially when you're comparing two goalies who faced a significantly different number of shots). But the impact of shot volume at the season and career level is essentially zero.

    So to answer your question - there's no conflict in the data that's been presented. The issue is some people don't understand that what's true at the individual game level doesn't hold true at the season/career level.

    I'm tagging @Doctor No as I believe he looked into this as well.
     
    Last edited: Aug 25, 2020
    Doctor No and Filthy Dangles like this.
  7. Doctor No Registered User

    Joined:
    Oct 26, 2005
    Messages:
    8,792
    Likes Received:
    2,989
    Trophy Points:
    211
    SB Cash:
    $ 100,000
    Home Page:
    My preconceived notion is that (1) this phenomenon exists on an in-game level, (2) it's primarily due to score effects, but it also has to do with (3) shot counting biases and perhaps (4) team strength.

    I have more to share later, but here's a quick graph that I put together from my database - this is all regular-season data, National Hockey League, 1999 through present day. I have some older seasons entered as well but those are not included here.

    upload_2020-8-25_12-42-24.png

    Caveat: my data has typos; I'm sure of it. I'm trying to eradicate them from my database (actually developing the above graph was a nice clean-up exercise); my conjecture is that at the level of granularity presented in this exhibit, the typos won't be material. For some of the stuff I'm working on (like save percentage based on score at the start of the period and shots faced) I think that it does matter, so more to come.

    If score effects (roughly defined as "trailing teams trade shot quality for shot volume") this should manifest in the data as a "steeper" save percentage curve when looking at shots faced in the period. We do see this in the data (compare the gray line - third period - to the other two lines).

    It's also interesting (but bears out with logic) that second period save percentages are basically always lower than first/third period save percentages - it's easier to "get caught" in the second period and give up a bad chance.

    Anyhow, more studying is warranted (and in progress).
     
    Last edited: Aug 25, 2020
  8. Doctor No Registered User

    Joined:
    Oct 26, 2005
    Messages:
    8,792
    Likes Received:
    2,989
    Trophy Points:
    211
    SB Cash:
    $ 100,000
    Home Page:
    Couple of other things as I think of them - these are shots faced by the team in the period (not a single goaltender) because I didn't want to tease out how many shots two different goaltenders faced in a period, and I think it's purer here.

    The blue line (first period) is probably the purest result relative to score effects, since those would have far less of an impact.

    The first "17" in the chart is "16" - it's a typo that doesn't affect the graphs.
     
    Last edited: Aug 25, 2020
    Filthy Dangles likes this.
  9. morehockeystats Unusual hockey stats

    Joined:
    Dec 13, 2016
    Messages:
    487
    Likes Received:
    158
    Trophy Points:
    56
    SB Cash:
    $ 100,000
    Occupation:
    sysadmin
    Location:
    Columbus
    Home Page:
    It's interesting to plot # of shots/fenwick attempts vs xg correlation.
     
  10. Windy River Registered User

    Joined:
    Jan 31, 2013
    Messages:
    1,408
    Likes Received:
    438
    Trophy Points:
    94
    SB Cash:
    $ 100,000
    IMO the most fatal flaw of goaltender statistics is that it tends to ignore the most important event that takes place - the ‘weak’ goal. In general terms anything other than a weak goal means the goalie is performing at a competent level. Most stats today focus on saves and goals and ratios thereof.

    The problem is looking at saves tells very little- a shot from center ice is treated the same as a breakaway when the puck touches the goalie.

    Likewise looking at goals is same flaw as shots.. we don’t know the quality!

    Now if there is a way to score the quality of the goal (ie quality of scoring chance preceding goal) then we could start to apply that to the goalie. Overall numbers (goals against/saves) can basically be ignored, instead just scoring how effective the goaltender is for a given ‘quality’ of scoring chance, then merging that with the typical breakdown of scoring chance types against a given opponent to predict game outcomes or league average to measure overall effectiveness (ranking).
     
    Mike Farkas likes this.
  11. Mike Farkas Grace Personified

    Joined:
    Jun 28, 2006
    Messages:
    11,025
    Likes Received:
    4,324
    Trophy Points:
    231
    SB Cash:
    $ 100,000
    Location:
    PA
    Home Page:
    I'm just posting in the same minute so that we know this isn't my burner account...
     
  12. Windy River Registered User

    Joined:
    Jan 31, 2013
    Messages:
    1,408
    Likes Received:
    438
    Trophy Points:
    94
    SB Cash:
    $ 100,000
    ?? I take it I’m saying many of same things you’ve been preaching here? If so... that’s great!!
     
  13. Mike Farkas Grace Personified

    Joined:
    Jun 28, 2006
    Messages:
    11,025
    Likes Received:
    4,324
    Trophy Points:
    231
    SB Cash:
    $ 100,000
    Location:
    PA
    Home Page:
    Yes, absolutely. I wonder if some of them aren't in this very thread...
     
  14. ted2019 Know your History

    Joined:
    Oct 3, 2008
    Messages:
    4,720
    Likes Received:
    1,294
    Trophy Points:
    173
    SB Cash:
    $ 134,471
    Gender:
    Male
    Location:
    pittsgrove nj
    I didn't realize you were Tony DeAngelo!
     
    Mike Farkas likes this.
  15. morehockeystats Unusual hockey stats

    Joined:
    Dec 13, 2016
    Messages:
    487
    Likes Received:
    158
    Trophy Points:
    56
    SB Cash:
    $ 100,000
    Occupation:
    sysadmin
    Location:
    Columbus
    Home Page:
    There are several statisticians around, including myself, that classify and evaluate each unblocked shot in terms of probability of becoming a goal. Myself, I filter out EN and penalty shots.
     

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"