HOH Top 40 Stanley Cup Playoff Performers of All Time

Discussion in 'The History of Hockey' started by quoipourquoi, Apr 25, 2017.

View Users: View Users
  1. MXD

    MXD Dead again

    Joined:
    Oct 27, 2005
    Messages:
    39,938
    Likes Received:
    3,144
    Trophy Points:
    186
    Occupation:
    Dead
    It blows my mind how someone would, at a point where we could be considered in extreme nitpicking territory, refuse to consider the more "complete" of the data, for whatever reasons that are not spelled out.
     
  2. daver

    daver Registered User

    Joined:
    Apr 4, 2003
    Messages:
    15,731
    Likes Received:
    795
    Trophy Points:
    214
    Home Page:
    If adjusting points based on opposition RS defensive strength is in "extreme nitpicking territory" then by all means let's have a discussion. I get the impression that Quoi is using his method as a primary metric in player comparisons.
     
  3. quoipourquoi

    quoipourquoi Goaltender

    Joined:
    Jan 26, 2009
    Messages:
    7,802
    Likes Received:
    916
    Trophy Points:
    119
    Location:
    Hockeytown, MI
    How would shooting on the 1999 Dallas Stars (168 GA) be the same as shooting on the 1999 Pittsburgh Penguins (225 GA)? Performance vs. Peers treats it the same. VsX both treats it the same and ignores GP. Adjusting strictly to league average (216 GA) treats it the same. Adjusting to rank isn't even a thing, and you would need to set arbitrary levels where the best (1st) and worst (21-30) team should be relative to the average to generate a final number. But adjusting the team GA to an 82-GP environment and then adjusting every player to a universal environment? Pretty straightforward stuff, and the closest way to compare any player to any player.

    If you need to make additional adjustment for line usage, TOI, EN scoring, X player is injured, etc, I'm not going to eat your lunch for that, but it starts with the understanding that even in the same playoff year, some teams are harder to score on than others.

    And I know you get that, because you've told me twice about Crosby scoring 1.4 points-per-game against Washington and Columbus.
     
  4. Canadiens1958

    Canadiens1958 Registered User

    Joined:
    Nov 30, 2007
    Messages:
    17,698
    Likes Received:
    1,194
    Trophy Points:
    179
    Gender:
    Male
    Occupation:
    Retired
    Location:
    Lake Memphremagog, QC.
    True

    Interesting comment.

    Canadiens historically(Cecil Hart thru Serge Savard via coaches and GMs) only cared about the actual goals against since goals against are the only statistic that will actually beat a team. Seasonal team bonuses were on the line for winning the old Vezina or for having the league lowest GA.

    The other measures that get ignored in the bolded, especially the fantasy measures such as VsX, are whether a team rolls 3 or 4 lines, 2 or 3 defenceman pairings, TOI - especially extra shifting and similar deployments of star players, PP structures - important since teams either used a regular forward line or a mix of forwards from the lines or a specialist. Blending from two lines for a PP unit has a ripple effect since if a coach blends from his top two lines then it is almost guaranteed that the third line plays the post PP shift.
     
  5. daver

    daver Registered User

    Joined:
    Apr 4, 2003
    Messages:
    15,731
    Likes Received:
    795
    Trophy Points:
    214
    Home Page:
    First bolded part - Where did I say they were the same? I pointed out that the Pens faced slightly worse defensive teams (based on GAA placings) on average than the AVs did.

    Second bolded part - Of course they are teams that are harder to score on but RS GAA doesn't seem to have a direct correlation to that. See the Preds this past year, see the 2016 Sharks (#11 in RS GAA, and #2 in the playoffs), see Tampa in 2015 (#11 in RS GAA, #1 in the playoffs), see the Kings in 2014 (#1 in RS GAA, 5th among the eight QFs).
     
    Last edited: Jul 18, 2017
  6. daver

    daver Registered User

    Joined:
    Apr 4, 2003
    Messages:
    15,731
    Likes Received:
    795
    Trophy Points:
    214
    Home Page:
    So if I understand you correctly you feel Mario's adjusted 1.37 PPG for his five best playoff runs is the best reflection of him as a playoff performer (for those five those runs at least).

    As opposed to being all very nitpicky kind of stuff.
     
  7. daver

    daver Registered User

    Joined:
    Apr 4, 2003
    Messages:
    15,731
    Likes Received:
    795
    Trophy Points:
    214
    Home Page:
    Why wouldn't I point that out to you after you have made it clear that you place emphasis on how players fare against teams with better GAAs?

    I pointed out the 1.4 figure to hopefully put to bed any narrative that Crosby struggles against the best defensive teams.
     
  8. Black Gold Extractor

    Black Gold Extractor Registered User

    Joined:
    May 4, 2010
    Messages:
    570
    Likes Received:
    40
    Trophy Points:
    51
    Using quoipourquoi's data here, I've calculated that Gretzky's best five years (1985, 1983, 1988, 1984, and 1993) result in 1.51 adjusted points/game. 1.51 and 1.37 look low only because we're accustomed to seeing their raw numbers.

    The strength of the performance-versus-peers method is that we can account for many variables (primarily deployment), but the key assumption is that all top players play in a similar environment. While this is a decent assumption in the regular season (each team faces a wide range of teams of varying strengths), this is no longer the case in the post-season, where you face anywhere from one to four teams.

    Using standings as a proxy for defensive strength is about as valid as using +/- as a proxy for defensive strength because standing points correlate fairly well with goal differential as opposed to any single side of the coin. For example, while typically an average team is more likely to have average offense and average defense rather than excellent offense and terrible defense (1988-89 Penguins come to mind) or vice versa, it's certainly possible. There's no need to use the proxies (i.e. standing points or +/-) when the actual data is available (i.e. team and player GA respectively).
     
  9. daver

    daver Registered User

    Joined:
    Apr 4, 2003
    Messages:
    15,731
    Likes Received:
    795
    Trophy Points:
    214
    Home Page:
    They look low to me because they are not in line with how they look vs. a performance vs. peers analysis; an analysis whose only fault is it presumes each peer played against the same competition. I believe this presumption would be evened out the more games that someone plays. I find that method a lot more palatable than one where a team is presumed to have played to their RS level of defense despite many examples of teams performing well below or above their RS level.

    We are probably looking at the extreme of league GAs from the high flying 80s/early '90s to the depths of the DPE and can appreciate that a GAs in 1999 can be, on a relative basis, a lot closer to GAs from 1990. I would tend to want to temper Wayne and Mario's #'s a bit given how much tighter scoring is these days but I don't think that is a position that would be well accepted if we start talking about RS #'s.
     
  10. quoipourquoi

    quoipourquoi Goaltender

    Joined:
    Jan 26, 2009
    Messages:
    7,802
    Likes Received:
    916
    Trophy Points:
    119
    Location:
    Hockeytown, MI
    The numbers we ran during the project were more precise because we calculated adjusted points within each series instead of taking the total points in a run and adjusting for total environment. Takes longer, but it's improved. ~1.51 probably wouldn't be too far off, BGE.

    Also the 200 GA environment I adjust to is intentionally low. The ~1.51 or 1.37 numbers only mean anything when being compared to another number in the same low 200 GA environment. Saying Mario Lemieux's is 1.37, daver, is meaningless unless his 1.37 is being held against someone else's number (which unless it is Wayne Gretzky, will be worse). I could have just as easily put it in a 150 GA or 250 GA or 300 GA environment and assigned a different number.

    And to address the other point that teams don't perform exactly the same in the playoffs that they do in the regular season... of course. But a balanced 82 GA sample is better than treating Jennings winners and high-GA teams the same just because GA-82 might not perfectly represent how a team plays for 4-28 GP in Spring.

    And what's the alternative? Fawning over how the 2017 St. Louis Blues scored 11 GA in 6 GP in the 2nd Round because the new calculation for Nashville is a 0.75 GAA?
     
  11. quoipourquoi

    quoipourquoi Goaltender

    Joined:
    Jan 26, 2009
    Messages:
    7,802
    Likes Received:
    916
    Trophy Points:
    119
    Location:
    Hockeytown, MI
    And that's the whole point of why I ran the adjustment... players in the same playoff could face wildly different competition.
     
  12. TheDevilMadeMe

    TheDevilMadeMe Registered User

    Joined:
    Aug 28, 2006
    Messages:
    47,096
    Likes Received:
    849
    Trophy Points:
    169
    Location:
    Brooklyn
    Awards:
    I'm not really comfortable adjusting playoff numbers based on regular season scoring, because we know that some of the best offensive teams in the 80s and early 90s didn't bother playing defense in the regular season, before tightening up in the playoffs. Especially in the 1980s, when 16 of 21 teams made the playoffs.
     
  13. MadLuke

    MadLuke Registered User

    Joined:
    Jan 18, 2011
    Messages:
    2,791
    Likes Received:
    20
    Trophy Points:
    66
    Looking at the GA in playoff vs regular season for all the teams in the combined season between 1980 and 1984

    Correlation between regular and playoff season GA is quite good: 0.68
    Better than the correlation for goal for at 0.6

    The team with the 3 worst offensive to make the playoff that time, Canucks, Pens, Red wings were 3 of those that improved.

    Team by improvement in absolute goal againt in the playoff

    Red Wings (low sample size): 0.99 less GA by games
    Canadiens: 0.7
    Nordiques: 0.52
    Oilers: .48
    Canucks: .47
    Islanders: .41
    Penguins: .38
    Blues: .25

    It almost perfectly balance out with those who accorded more goal instead of less too.
     
  14. 10Ducky10

    10Ducky10 Registered User

    Joined:
    Jul 5, 2015
    Messages:
    5,593
    Likes Received:
    1,083
    Trophy Points:
    128
    Dryden played in 8 playoffs and won 6 cups.
     
  15. lancer247

    lancer247 Registered User

    Joined:
    Jan 16, 2007
    Messages:
    3,828
    Likes Received:
    83
    Trophy Points:
    101
    The list needs more Bernie Parent and Bobby Clarke.
     
  16. ImporterExporter

    ImporterExporter I troll harder than Poppy

    Joined:
    Jun 18, 2013
    Messages:
    10,769
    Likes Received:
    1,147
    Trophy Points:
    139
    Location:
    Oblivion Express
    Why? Parent did have 2 elite playoff performances, but outside of that, nothing at all. Nobody should make a top 40 all time list with only 2 notable runs to their name.

    Clarke was a solid playoff guy, but not good enough to make the top 40. He was outshined by Rick MacLeish in both 74 and 75 (their Cup winning years) and then again the following year by Reggie Leach in a losing effort.
     
  17. daver

    daver Registered User

    Joined:
    Apr 4, 2003
    Messages:
    15,731
    Likes Received:
    795
    Trophy Points:
    214
    Home Page:
    Only on paper.

    Why not, at the very least, include some consideration for playoff GAs which, at the very least, would account for the lower overall scoring that usually occurs in the playoffs? This, at the very least, would temper some of the huge variances in GAs between teams from seasons where scoring was close to 4.00 and seasons where scoring was close to 2.5.

    This whole project is about playoff performances and recognizing players that performed in the most important games of all. We are supposed to ignore a player's RS resume completely. Incorporating these RS GAs seems counter to that mindset.

    As I have been mentioning all along, if any consideration for RS GAAs is deemed to be worthy, it should be based on GA rating in respective seasons. IMO, this is more in the spirit of comparing players' respective Art Ross placings vs. a straight up comparison of raw point totals and PPGs. You use a 200 GA as your base which makes no sense as Wayne and Mario that they didn't even have the chance to play a team with a 200 GA even though they did play the best defensive teams at some point.

    This exercise introduces an unnecessary hypothetical scenario; one that asks "what if Wayne or Mario played against, not only better RS defensive teams, relative to another era, but also better defensive teams in general as the scoring environment had dramatically changed?" First of all, this assumes that the Cup winning Oilers and Pens would not be any better or worse, relative to the league, if they played in the late '90s, that their opponents would be the exact same as the AVs faced, and that they would get hot and cold in the exact same way as the Sakic or Forsberg did.

    According to your method, Malkin's 11 points in 5 games against the #2 GA team this year is given significantly more value than Crosby's 7 points in 6 games against the #15 GA team even though the Jackets' and Preds' playoff GAs were dramatically different. Perhaps this is an extreme example, one that gets evened out the more games that are considered, but that's my feeling about your method; that the more games considered for each player, the more a difference in the defensive opposition gets eliminated.

    And even if Wayne faced considerably worse defensive teams, relative to his era, why isn't that a reflection of the Oiler RS success of which Wayne played a huge part in?

    I don't have an issue at all with accounting for the different scoring environments between eras but your method makes way too many assumptions.

    EDIT I do appreciate that playing in a 21 team league vs. a 28 or 30 team league is something that can possibly be accounted for but am wondering how to capture that without giving an unfair advantage to a group of players over another. My instinct is that a performance vs. peers method would account for this as, for example, Wayne's peers like Bossy or Trottier would have played under the same type of conditions and faced worse defensive teams than their other peers of the era. The highest scorers over a number of seasons, you would think, would have faced a similar level of defensive competition. I am sure Yzerman and Federov would have faced a similar level of defensive opposition as the Sakic/Forsberg did from 96 to 2004 if you averaged it out.
     
    Last edited: Jul 19, 2017
  18. Canadiens1958

    Canadiens1958 Registered User

    Joined:
    Nov 30, 2007
    Messages:
    17,698
    Likes Received:
    1,194
    Trophy Points:
    179
    Gender:
    Male
    Occupation:
    Retired
    Location:
    Lake Memphremagog, QC.
    1974 and 1975

    But he was not playing against Rick Macleish in 1974 and 1975. Defensively those playoff years, Clarke faced Jean Ratelle(2 points in 7 games), Phil Esposito(3 points in 6 games), Gilbert Perreault(2 points in 6 games). Three HHOF centers who Clarke limited to 7 points in 19 games total. Hard to do better.

    Clarke had some problems in the playoffs notably against Darryl Sittler, who had a losing series for the ages in 1977 against the Flyers as well against Henri Richard in 1973.
     
  19. Canadiens1958

    Canadiens1958 Registered User

    Joined:
    Nov 30, 2007
    Messages:
    17,698
    Likes Received:
    1,194
    Trophy Points:
    179
    Gender:
    Male
    Occupation:
    Retired
    Location:
    Lake Memphremagog, QC.
    Realities

    Realities are such that the NHL RS play of teams and players set the playoff match-ups within the divisions and conferences before the SC Finals.

    As for facing tougher defences, taking the start of the Oilers dynasty:

    1983-84 Top 8 defensive teams based on GA were in the East.
    1984-85 Top 6 defensive teams "
    1985-86 Top 8 defensive teams "
    1986-87 Top 3 defensive teams "
    1987-88 Top 5 defensive teams "

    Eastern skaters faced stronger defensive teams more often, given that the schedule was not balanced during these seasons.

    The stronger defensive play carried over into the playoffs.

    1984 Playoffs
    Against west 14 Games with 73 GF, east, 5 Games with 21 GF.
    1985 Playoffs
    Against west 13 Games with 77 GF, east, 5 Games with 21 GF.
    1987 Playoffs
    Against west 14 Games, with 65 GF, east, 7 Games with 22 GF.
    1988 Playoffs
    Against west 14 Games, with 66 GF, east, 4 games with 18 GF.
    Note power failure results not included.

    Championship Oilers scored less against the defensively stronger east in the playoffs.
     
  20. Black Gold Extractor

    Black Gold Extractor Registered User

    Joined:
    May 4, 2010
    Messages:
    570
    Likes Received:
    40
    Trophy Points:
    51
    The regular-season GA is to determine a player's opponents' strength. It has nothing to do with the player's regular-season resume.

    If it were up to me, I'd use 220 GA over 82 games as a better representation of the current era's average. That's easy enough to do: just add 10% to everyone's adjusted points/game. This yields Gretzky's best 5 post-seasons at 1.66 adjusted points/game, Lemieux at 1.51, etc. But as quoipourquoi said, as long as everyone is adjusted to the same hypothetical environment, you just need to compare their relative numbers.

    Everyone opponent is adjusted to the extremely tough hypothetical 200 GA-in-82-games team. If it happens that some players faced tough teams more frequently than others, I'm not sure why that shouldn't be accounted for.

    I'd argue that the extreme difference in the Jackets' GAA comes from the fact that they played 100% of their playoff games against the defending (and eventual) Stanley Cup champions while they played everyone (good and bad) in the regular season.

    As for the Predators, Rinne got hot. Malkin got hot too. Small sample sizes are a pain. As you say, though, it evens out over larger samples. Five post-seasons are probably a decent size for most accomplished players.

    Because in the regular season, each team plays everyone else, so everything averages out (more or less, given that no schedule can be perfectly balanced).

    As for the playoffs, what Canadiens1958 said above. Due to how things shake out, playoff schedules can end up being very unbalanced.

    With regard to Gretzky versus Bossy/Trottier, I would say that the fact that Gretzky had to face the hardened defense of the Islanders as opposed to Bossy/Trottier, who had to face the mediocre defense of the Oilers, is an immediate advantage to Bossy/Trottier in head-to-head match-ups in terms of scoring points. Of course, conversely, Gretzky faced weaker defensive teams getting to the Finals than Bossy/Trottier, so why not try to normalize everyone's opponents? The mathematical model isn't perfect, but it's better than handwaving or raw unadjusted numbers.
     
  21. BraveCanadian

    BraveCanadian Registered User

    Joined:
    Jun 30, 2010
    Messages:
    11,977
    Likes Received:
    51
    Trophy Points:
    101
    I'd tend to agree.

    The playoffs are a different beast because teams can gameplan more specifically against one another as well.
     
  22. MXD

    MXD Dead again

    Joined:
    Oct 27, 2005
    Messages:
    39,938
    Likes Received:
    3,144
    Trophy Points:
    186
    Occupation:
    Dead
    The thing is -- it's probably more accurate than to consider every team being the exact same defensively. And if we're to do that, there are very good reasons to believe that, sometimes, there is a difference between the 5th best defensive team in a given regular season and the 5th best defensive team in another regular season.

    Whether coming up with a hard formula is the right way to do it, I don't have a clue.

    EDIT : And there's the "matching up against one team" side of things that BraveCanadian refers to, as well.
     
  23. MadLuke

    MadLuke Registered User

    Joined:
    Jan 18, 2011
    Messages:
    2,791
    Likes Received:
    20
    Trophy Points:
    66
    But regular season goal against seem to be a better predictor of playoff goal against result than goal for in regular season for playoff goal scored.

    And we would not hesitate as much to use goal for in regular season to rank team offense potential in playoff I would think, the correlation between regular season defense and playoff seem good enough to me to have more value than league average.
     
  24. BraveCanadian

    BraveCanadian Registered User

    Joined:
    Jun 30, 2010
    Messages:
    11,977
    Likes Received:
    51
    Trophy Points:
    101
    Due to sample size that would intuitively make sense.

    However, the playoffs are not the regular season and due to the small number of games, single blowouts etc. will change the average much more, injuries will have more effect, and repeatedly facing the same opponent would make a difference in theory.

    I think that both GF and GA would suffer from the same problems in the change from regular season and playoffs.

    The regular season might be a better indicator overall because of the sample but I think playoff GF and GA would vary much more in any individual series.

    Of course then you have to question how much weight you want to give those blowouts or whatever that are so out of line with their regular season performance. Could just be a fluke.
     
  25. Black Gold Extractor

    Black Gold Extractor Registered User

    Joined:
    May 4, 2010
    Messages:
    570
    Likes Received:
    40
    Trophy Points:
    51
    Here are some adjustments broken down per series using quoipourquoi's method (min. 2 playoff series played):


    Best 5 post-seasons combined:


    Top 4 post-seasons after excluding their respective single best post-seasons:

     

Share This Page

monitoring_string = "358c248ada348a047a4b9bb27a146148"