Oilers at Sharks, 73/07-08

That’s Gilles Meloche in goal, Seals defender Paul Shmyr (played later for the Oilers) wearing #3 and tying up Habs #24 (Chuck Arnason) maybe 1972.

I love looking at these old photos. Looking at the crowd makes me think it may have been cool that night (a few winter coats) and knowing the gap between the two teams in terms of quality at that time in history I’d bet Meloche was busier than the Hab goalie.

Also, notice the skates Shmyr is wearing and the equipment of these players. It’s like a photo of Babe Ruth pitching in terms of time and place.

Tonight’s game features a bit of a gap in terms of quality but both clubs are hot as Hades right now. The Oilers are 8-2 in their last 10 and the Sharks are perfect in the same span.

The Oilers are making a nice little run here but the schedule beginning tonight is a bit of a sphincter tightener if you’re hoping the Ducks draft in double digits. The good thing is the roster has a ton of kids who have no idea they’re supposed to lose.

written by

The author didn‘t add any Information to his profile yet.

302 Responses to "Oilers at Sharks, 73/07-08"

« Older Comments
  1. Slipper says:

    >Numbers without context… thanks slipper<

    Oilman have you checked out the fantastic site called hockey’s future?

    Over there we can make all sorts of poetic comparisons between hockey players based on things we eyeballed. Fuck comparing production and all that useless nonsense.

    I wanna talk about hockey’s INTANGIBLES and what makes us smart hockey watchers. I can tell you who fucked up seconds before a goal and I think Cogliano is like a young Mike Modano. I’ve gots me a sharp eye for hawky!

    Cogs is so defensively sound!

    Does Stortini outscore opponents, or create more powerplays than he gives up?

    Who fucking cares!?!?

    He’s like a young Darren McCarty. Maybe in looks only, but fucking brilliant none the less!

  2. Cam says:

    And here I thought it was St. Patrick’s day. I guess I missed the ‘treat your peers like assholes’ day announcement.

  3. Coach pb9617 says:

    You use the term ‘peers’ loosely.

  4. mc79hockey says:

    I know that blogger sometimes converts my mc79hockey login to “tyler”. Is that Cam RiversQ? This is all very confusing.

  5. Cam says:

    It appears that I (kinger) and RiversQ share the same first name.

    I’m too lazy to log out of my personal profile into my blog one.

  6. Bank Shot says:

    Well they should be. There are three of them on the ice compared to two defensemen. ;p

    Alot of the things defensemen do to control the flow of the game aren’t recorded in the game stats, such as making a good breakout pass, standing the opposition up at the blueline, or crushing some weasel in the corner to break up the cycle.

    That makes it pretty hard to compare their contributions to those made by the forwards.

  7. Lowetide says:

    Well I don’t know about anyone else but this was a beauty thread today. I may have to post Stan Weir in the next GDT if everyone is this interested.

  8. Bruce says:


    Bruce, in 30 words or less, all I was saying is that I believe forwards are more responsible for generating shots. If that doesn’t compute with you, so be it.

    Now get busy and count the words in my post.

    40, counting the cheap shot at the end. :)

    Bookmark it.

    Done, with thanks. I’ve had trouble with the interface to that site for some reason.

    It’s interesting to note how severely those Fenwick numbers have changed. Vic called them “Corsi numbers” in his IOF post of that title on October 31, but I think the list he provided there was actually Fenwick numbers as he mentioned excluding blocked shots. Are these definitions really that recent?

    Anyway on Oct. 31 — around the times first impressions were hardening into opinions :~) — Hemsky led the Oilers at +9, Gilbert was hard on his heels at +5, all other players were minus while at the bottom of the list was Grebeshkov, Greene and Stoll around -40 and Stortini already at -103, in very limited and clearly very unsuccessful minutes.

    Today Hemsky still leads the pack among regulars at a modest but positive (+3). Gilbert is now third from the bottom at -165. (Cogliano -186, Staios -185). Meanwhile Stortini has actually recovered up to -78, meaning his Fenwick number has been +25 since Nov. 1. Luckily any fool could have predicted that, so it’s no big deal.

    He’s like a young Darren McCarty. Maybe in looks only, but fucking brilliant none the less!

    Call me old school, but I like to do comparables, recognizing the inherent limitations to the method. Even the Sedins aren’t identical. Er, wait a minute …

    But in the context of team-building, what are you looking for? If you’ll pardon the unacceptable labelling, “a Darren McCarty” is distinct from “a Cam Neely” or “a Kirk Maltby”, but all are prototypes of the type of player you may want. The Oilers have been looking for a Chris Pronger type for all but one of the past 15 years.

    Is there any basis for comparison? Well maybe you can tell me Darren McCarty’s Fenwick number from his second year in the NHL and we can compare notes. What’s that you say — you can’t? No Fenwick numbers before the lockout? Guess I’ll have to trust my tried-and-true old-fashioned “comparables” method (eyeballs, traditional stats). It’s not perfect, and it’s not fanboy stuff, it’s just a projection. What role might this player fill on an idealized roster? How far from that ideal is he? Etc. It’s more measuring stick than true comparison.

  9. heed says:

    if anyone cares…

    i am going to start looking at all the various “stats” (traditional nhl as well as the new “fangled” variety) to see which ones correlate to wins and which ones correlate to nothing. i am going to start with the traditional nhl figures since the lockout (GF, GA, PK%…) and compare them to total points. once the season is over, i will move onto such things as corsi, fenwick, etc. i want to see if any of these numbers actually are related to winning. wish me luck.

  10. Slipper says:

    I like Lowetide’s comparisons that are based just on production values from identical season’s age-wise in a player’s development seasons.

    The problem with stylistic comparisons is that one minute a guy’s talking about the similarities between Stortini and MaCarty’s game, and then the next minutes someone says Grebeshkov looks a little like Coffey and Pronger.

    For some sadistic reason I still skim over fanboards and just today I read the Cogliano reminds people of Butch Goring, Langenbrunner, Jere Lehtinen and Scott Gomez (only slicker and better defensively). If being a dick prevents that sort of drivel from seeping over here, it’s worthwhile.

    But hey what the hell do I know… Jan Hejda plays in effing Columbus and the guys on TSN are calling Holmstrom the new garbage goal man!!1!

  11. Slipper says:

    Good luck Heed.

    You should look at the correlation between goal differential and winning.

    Just throwing that out there.

  12. heed says:

    slipper,

    i intend to look at every single one. i am hoping to find the strongest correlation between winning and the number of players on the roster whose first name starts with “j”. that will really get the blogosphere going.

  13. Dennis says:

    Coach: I can see the merit in both points but moreso in 2 than 1.

    18′s as brittle as fuck and we don’t need him anymore to ice a better-than-average PK so I don’t know why we’d bother to keep him.

    Btw, to take this back to something that was discussed way up thread, when’s the last time someone as young as 16 posted a PPG rate like he did in ~130 combined games between -06 and 7, sucked in the next season and then finally bounced back?

    I ask this because I saw him past someone in the early shifts of the third period last night and I don’t think the concussion has played a big a role in his regression as others do.

  14. Bruce says:

    Bruce, in 30 words or less, all I was saying is that I believe forwards are more responsible for generating shots.

    Corsi numbers by position:

    -403 Garon
    -157 Roli
    -560 = Goalies (/1 = -560)

    -261 Staios
    -216 Gilbert
    -202 Smid
    -124 Pitkanen
    -87 Greene
    -87 Grebeshkov
    -75 Tarnstrom
    -55 Rourke
    -53 Souray
    -7 Young
    -2 Peckham
    +7 Roy
    -1162 = Defencemen (/2 = -581)

    -232 Coglinao
    -184 Stoll
    -167 Pisani
    -146 Gagner
    -142 Reasoner
    -136 Moreau
    -108 Stortini
    -98 Brodziak
    -87 Torres
    -70 Sanderson
    -68 Nilsson
    -50 Horcoff
    -40 Penner
    -23 Hemsky
    -13 Pouliot
    -10 GlenX
    -5 Thoresen
    0 JFJ
    +1 Reddox
    +5 Schremp !

    -1573 = forwards (/3 = -524)

    So not exactly a 3:2:1 ratio but trending that way. Some ES shots would be taken 4v4 so ratios might not be exact and forwards should be a little light. I suppose it’s not impossible that there may be small errors in the collection method too.

    These are just the net Fenwicks but I imagine if you added up all the individual columns you would get a similar distribution.

    I’m trying not to be thick, Slipper, but I don’t see the bias against defencemen or evidence of the forwards driving play from this. Tell me what you’re seeing.

  15. Slipper says:

    One last thing and it’s directed at the oilman types.

    At the very least, I assume that the Bruces and Bankshots and Coachpb3241s are arguing in good faith.

    On the other hand, a guy like oilman confronts anyone presenting a stastical evaluation of a player or the game in general, with the tired and time worn tradition of posing a million hypothetical suppositions. “Waht if a player is doing all the right things over and over but everyone else on the ice sucks and they get scored on 100 times, blah, blah…”.

    When he has asked for context and it’s raised to him that there are a plethora of options available that can provide a decent picture of what’s happening during a player’s regular shift, he counters with the old “there aren’t numbers that can tell him who ‘yaps at other players blah blah blah’ and the net result it has on the game” argument.

    If you’re looking for an expression to calculate what a player ate for lunch on a fucking game day you’re obviously being a stubborn douche bag who will never be satisfied with the answers people around here put effort into providing for the rest of us. That’s not arguing in good faith. That’s just crowning yourself king of the douche bags, so don’t shit on my head for being blunt about it.

    One day even an old goalie like yourself may come to realize that every save you ever made, even the one that hit you dead on the crest, were being aimed by the shooter at the open spaces of the net and that goaltenders are credited with alot of saves that were never going to hit the net in the first place.

  16. Tyler says:

    Am I missing something? I would think that, as a rule, the ratio has to be 3:2:1, or thereabouts. The question is who’s responsible for the number – ie., whether the forwards, defencemen or goalies are going to make the biggest change in the shooting numbers as a result of their skill. You won’t be able to find it with that sort of comparison.

  17. Coach pb9617 says:

    If being a dick prevents that sort of drivel from seeping over here, it’s worthwhile.

    Yes, you keep the discussion on the straight and narrow. Without you, all hockey discussion would break down. We all thank the Prince of Wales that he sent you here to moderate.

  18. Slipper says:

    Bruce, I feel that an average defender is at the mercy of the forwards for the ammount of shots directed at the opposing net. I’m not saying that they don’t have any control, but that with forwards ahead of them forechecking and cycling and bringing the puck to the net (doing the things defensemen aren’t commonly responsible for), that they’re slaves to the ability ahead of them. Forwards (again, this rallies against everything we’ve come to know about the puck moving defender) have more to do with getting the puck out of their own end of the rink than defensemen have to do with keeping it in the attacking zone. Maybe this is just caused by pure out numbering (3F to 2D), but on the other hand, the position dictates that atleast one D is often not involved in the attacking zone play.

    This is just a working idea at the moment, and I can’t confirm it just yet. Although I think taking the elite defensemen in the league and comparing their numbers with and without certain forwards would be a start. Something tells me Lidstrom with and without Zetterberg, or Hejda with and without Nash and Zherdev would show a huge discrepancy.

    I think this discrepancy is exacerbated on a bad team like the Oilers where so many more of the forwards are terrible at taking/keeping the puck to the right end of the rink.

  19. Coach pb9617 says:

    18′s as brittle as fuck and we don’t need him anymore to ice a better-than-average PK so I don’t know why we’d bother to keep him.

    Again, I don’t disagree, but I don’t see him being waived and I don’t see them especially waiving the active captain.

  20. Slipper says:

    Have you tried http://www.edmontonoilers.com coach? It might be right up your alley.

    Just recently I read that Gagner can be likened to Doug Gilmour, mostly because of their “crouching style”. Does that sound up to your speed?

  21. Bruce says:

    Am I missing something?

    I was aking the same thing myself, Tyler, with my original challenge of this statement:

    By my eye defensemen seem to get the short end of the stick by Corsi’s standard, routinely doing worse than forwards, and it’s more evident on the poor teams.

    I would think that, as a rule, the ratio has to be 3:2:1, or thereabouts.

    Well right, that was my point, but when I got slapped down in 30 words plus 10 after the whistle, I thought I better do a background check to make sure I was on sound footing. Of course it should be 3:2:1, and it is.

    The question is who’s responsible for the number – ie., whether the forwards, defencemen or goalies are going to make the biggest change in the shooting numbers as a result of their skill. You won’t be able to find it with that sort of comparison.

    No you won’t. I’m (still) questioning what Slipper sees in those Corsi numbers that gives defencemen “the short end of the stick”. I’m not deliberately being thick, I just can’t help it. :)

  22. Coach pb9617 says:

    Have you tried http://www.edmontonoilers.com coach? It might be right up your alley.

    Your ad hominems really keep everyone here on the straight and narrow. Keep it up!

  23. Oilman says:

    “I wanna talk about hockey’s INTANGIBLES and what makes us smart hockey watchers. I can tell you who fucked up seconds before a goal and I think Cogliano is like a young Mike Modano. I’ve gots me a sharp eye for hawky!”

    Slipper….when they made you they broke the mold….that or they started using it to make giant dildos….either way, as far as dickheads go, you’re one of a kind.

  24. Slipper says:

    On a 3:2:1 basis?

    How does a goalie affect the number shots directed at him?

  25. Slipper says:

    Bruce, I wasn’t suggesting that it’s unevenly distributed, if that’s your contention. I just don’t believe that the actual responsibilities and roles of the positions were created equal.

  26. Oilman says:

    How does a goalie affect the number shots directed at him?

    Slipper…let’s pretend you don’t hate me for a second and see if this sounds reasonable? Is it not likely that a goalie who is playing well causes the shooter to often try to get into a better position before taking the shot, which often results in no shot taken? Or how about when a goalie is pulled and the new one comes in cold…doesn’t that often cause the other team to just try and shoot from anywhere to “test him early”….it’s another of those things that probably aren’t measurable…you know, if Luongo needs to take a shit between periods and Sanford has to replace him, no one will ever know if the opposing team took more shots of lower quality on Sanford as they would have on the diarrhea free Luongo?….or am I still being a douche for asking the question?

  27. Coach pb9617 says:

    No you won’t. I’m (still) questioning what Slipper sees in those Corsi numbers that gives defencemen “the short end of the stick”. I’m not deliberately being thick, I just can’t help it.

    “I just don’t believe that the actual responsibilities and roles of the positions were created equal.”

    The same subjective POV that he so desperately hates.

  28. Slipper says:

    As way of example, I’ll site Pronger and the Ducks.

    http://www.behindthenet.ca/2007/basic_5_on_5.php?sort=9&mingp=20&mintoi=5&team=ANA

    On the Ducks, Pronger’s Corsi/60 rating is second last only to Sean O’Donnell. He’s obviously not the the second worst defender on the Ducks, but he is on the ice for a signifigantly greater quantity of faceoffs in his own end than the other Ducks D.

    http://timeonice.com/tfaceoffs.php?team=ANA

    All the forwards that are listed as having signifigantly more defensive zone draws taken than offensive zone draws taken (R.Neidermyer, Moen, Pahllson) also have signifigantly lesser Corsi numbers.

    Pronger spends the most time with the worst Corsi producing forwards on the team, and begins with them most often in the wrong end of the dink, and thus he’s the seocnd worst defensemen on his team by this standard.

  29. Bruce says:

    Bruce, I wasn’t suggesting that it’s unevenly distributed, if that’s your contention. I just don’t believe that the actual responsibilities and roles of the positions were created equal.

    Yeah, of course. Funny, though, when the same distinctions get made in attempting to determine responsibilty for goals against (as opposed to shots) — namely the Errors stat — that gets dismissed with derisive statements about “I can tell you who fucked up seconds before a goal . But when a similar distinction is requested for who drives shots on goal, it’s: “ I feel that an average defender is at the mercy of the forwards for the ammount of shots directed at the opposing net. … This is just a working idea at the moment, and I can’t confirm it just yet.

    I’m not saying there’s nothing to it or that your interpretation is wrong, Slipper, just that I’m going to need something a little more convincing than a “working idea”. The Corsi metric of crediting and debiting everybody on the ice doesn’t account for any positional biases as to who’s repsonsible. So you’ll need a different methodology to find them.

  30. mc79hockey says:

    Slipper…let’s pretend you don’t hate me for a second and see if this sounds reasonable? Is it not likely that a goalie who is playing well causes the shooter to often try to get into a better position before taking the shot, which often results in no shot taken? Or how about when a goalie is pulled and the new one comes in cold…doesn’t that often cause the other team to just try and shoot from anywhere to “test him early”….it’s another of those things that probably aren’t measurable.

    Would it kill some of you people to write in proper fucking English? I mean, as a basic courtesy to others.

    As for your point, I’m amazed at the constructs some of you people develop. Watch a game some time and watch how fast it is. Ask yourself if it really seems plausible that players are passing up shots to try and get better ones because of how good the goalie is. Personally, I don’t think that there’s time for that and I don’t think that most players think that way…when they’re in a position from which they’d generally shoot, they shoot the fucking puck. No magic.

  31. mc79hockey says:

    I’m not saying there’s nothing to it or that your interpretation is wrong, Slipper, just that I’m going to need something a little more convincing than a “working idea”. The Corsi metric of crediting and debiting everybody on the ice doesn’t account for any positional biases as to who’s repsonsible. So you’ll need a different methodology to find them.

    I’ve actually gone through and taken a look at this, using ES GF/GA rates. I found that the forwards had a bigger spread than the defencemen, and that it wasn’t particularly that small of a spread either. There were way more forwards at the extremes than there were defencemen. I tend to agree with slipper.

  32. mc79hockey says:

    (My point being, I guess, that there’s a bit of background to this idea from back in the day.)

  33. Slipper says:

    Oilman, if you’re talking strategy wise, I don’t disagree with you. Game strategy is certainly affected by personell, and by many other situations. Some argue that Babcock teams are shot happy. I’ve seen this with my own eyes and the Corsi number bares this out aswell.

    As far as physically affecting the shot ammounts, outside of rebounds and some puck handling, I think they’re hooped. At the very least, they have signifigantly less control than the forwards, who could presumably keep the puck at the other end of the rink all game long, if their ability allowed it.

  34. Slipper says:

    Yeah, the “fucked up before a goal comment” might seem like a spear aimed directly at you and David Mainstream, but the history of people shitting on players who fucked up leading up to a goal (and conversely annointing those players they see regularly on the TSN highlights package as world beaters) surpasses your time on the OilerNet. That and people discounting the value of the statheads work in it’s entirety, because it doesn’t account for all the inane variables their imaginations can come up with.

  35. heed says:

    Good luck Heed.

    You should look at the correlation between goal differential and winning.

    Just throwing that out there.

    i just took a look at this comment and realized that you were likely being sarcastic. i guess i must be one of the ignorant masses that doesn’t deserve to post on the site (point taken and i will only come here to read the posts). i just actually want to take a look at that various stats using linear regression as a starting to point to see if there is any “surprises” that pop out. that’s the thing about numbers, they can surprise you once in a while. conventional wisdom does not always win out when you really dig into them. once i figure out what stats correlate to winning games, i am going to see if i can’t figure out what combination of those factors would maximize the number of points. could be an exercise in futility but it might also literally make me laugh my bag off. who knows?

    this ignorant ass is out for the night.

  36. Oilman says:

    Would it kill some of you people to write in proper fucking English? I mean, as a basic courtesy to others.

    Sorry to have offended you so, Old Chap…..I say, you wouldn’t have a copy of Grammatical Theory about?

    As for your point, I’m amazed at the constructs some of you people develop. Watch a game some time and watch how fast it is. Ask yourself if it really seems plausible that players are passing up shots to try and get better ones because of how good the goalie is. Personally, I don’t think that there’s time for that and I don’t think that most players think that way…when they’re in a position from which they’d generally shoot, they shoot the fucking puck. No magic.

    This just fucking kills me….especially from someone as edumacated as yourself…have you never watched Hemsky play? Do you think he’s ever looked for a better shooting position? Have you never heard that players ofter begin to try and aim their shots a little to finely when the goalie seems to be stopping everything?…Any fool with an LLB can see that.

    And really, is the “watch a game sometime” line…..get over yourselves.

  37. Slipper says:

    It wasn’t.

    I’d look forward to what you came up with, Heed.

    I know Mudcrutch did alot of work on goal differential and winning in the past, and I was only suggesting that as a good place to start.

  38. heed says:

    It wasn’t.

    I’d look forward to what you came up with, Heed.

    I know Mudcrutch did alot of work on goal differential and winning in the past, and I was only suggesting that as a good place to start.

    sorry about that. i think all the “holier than thou” posts in this thread got to me. not by you in particular, just in general.

  39. Lowetide says:

    I’m pissed that the posts after 200 go on a different page. How do I fix that?

  40. Oilman says:

    Slipper…not only strategy wise like Babcock’s “throw everything at the net” game, but confidence wise as well….I don’t dismiss all the stats guys because I want to be an ass….but you can’t tell me that a goalie “in the zone” doesn’t get into a shooters head. Look at Hasek in Nagano as a good example…players didn’t know what to do to beat him…Shanahan in the shootout was so confused by the time he got to the net he didn’t know what to do with the puck. Crawford leaves Gretzky on the bench because he thinks the only thing that can beat him is a perfect shot, so he sends in a Dman who can hit 4 targets out of 4 in the all-star skills contest. He had that whole bench psyched out – right down to the coaches.

    seriously, the game isn’t played on paper. You guys have a great set of tools to analyze what has happened, but it’s not so great a tool to predict what will happen (just my opinion) – because a lot of what does happen either isn’t measured or can’t be…..

  41. grease trap says:

    Wow, I hop in and there’s a bitchy catfight!

    raowrr!

    Okay ladies, you’ve had your fun. off to the powder room now…

    ;)

  42. Tyler says:

    have you never watched Hemsky play? Do you think he’s ever looked for a better shooting position?

    Fine keep up with ellipsis bit. Could you at least try to get my point. Go back and read what I wrote: “Ask yourself if it really seems plausible that players are passing up shots to try and get better ones because of how good the goalie is.”

    Hemsky does his dipsy doodle shit whether it’s Luongo in net or the pre-resurrection Conklin. In order for your point to have some validity, Hemsky has to start doing more or less dipsy doodling based on the goalie.

    Have you never heard that players ofter begin to try and aim their shots a little to finely when the goalie seems to be stopping everything?…Any fool with an LLB can see that.

    As everyone knows, LLB’s are for people who aren’t good enough to get a JD. As to your point, I like how you’ve somehow referred to a piece of a folk wisdom, spouted by the same type of people who don’t understand how the long change works, and then suggested that I should have noticed its effects. It’s as if it’s true just because someone said so.

  43. Oilman says:

    Players that have done so have said so….I really begin to question your understanding if you can’t follow the simple concept that players actually think about things while playing, or have you just decided to disagree with everything I say? Honestly, the players aren’t robots you know.

    I’m not saying that a player will shoot differently just because he’s playing against a certain goalie (although, some players have said they’ll change a certain favorite move or shot because the goalie catches on the right side ala Garon (and myself) vs. the other 90% of the goalies out there….but if you constantly get stopped with the glove, you may try to force one to the blocker side, and end up not getting the shot off at all…it’s pretty simple stuff – not magic.

  44. Oilman says:

    BTW, “the ellipsis thing” …why does it bother you so much?

  45. Slipper says:

    OM, Nagano’s a bad example to draw circumstancial evidence to pursue this point of yours. You’re talking about the Dominator, who was ten years younger in ’98 and in the middle of his prime dominating years. We’re talking about the genetic freak who had muscle memory and reflexes that are unsurpassed. His save percentage is enough to explain why he lead his country to a gold medal in Nagano without entering the fuzzy realm that requires psychoanlysis to comprehend.

    Remember, my point was that goalies can’t control the shots directed at them. I’m not arguing that goalies can’t control outcomes by stopping shots. They just have little physical control over the shots they face over the course of the game.

  46. Slipper says:

    Oilman, if you were really confident in your point you’d look at lefty and righty goaltenders and see if there were any trends in where opponents shot at them. Even if you found one it still wouldn’t have proven your original argument though: that goaltenders have more than limited physical control over the ammount of shots directed at the net.

  47. Slipper says:

    As for the example of a goalie coming in cold having an effect on the coach’s gameplan decisions, while the change in the decision process may be true, the team must still execute that decision in the physical world.

    Once the attacking team has “decided” to be less picky with their shot location choice, or in simpler terms, become shot happy, it’s still out of the goaltender’s physical control whether the attacking can properly execute said gameplan.

    We can still play all the minutes in the other end of the rink.

  48. Oilman says:

    slipper…muscle memory? That’s the first time I’ve ever heard that term associated with goaltending – or hockey in general for that matter…interesting. I guess body position in goaltending is muscle memory, but I never really associated the two – like you do when considering the golf or tennis swing. Odd, considering I’ve played goal for so long…anyway – I thought Hasek was a pretty good example – probably because it’s the most obvious. I think most of these “head games” happen within a game and a specific example, other than the obvious one, is honestly hard to come up with, without it just being anecdotal.

  49. Oilman says:

    Slipper – I don’t think I ever said the goaltender can control the amount of shots…if I could, I’d be playing at a much higher level than I do. But they can have an effect on the number of shots. Like I said, it’s just not one of those things that can be measured (whether or not a shooter overthinks his shot to the point of not getting the shot off).

    At one point in the San Jose game last night, Pitkanen walked in with plenty of time to shoot and hesitated for way too long on his backswing. He was in a hell of a shooting area and got his pocket picked – never got the shot off. Whether he thought he couldn’t beat Nabokov with a wrister from that spot, or he felt he had more time than he did, we’ll never know – he was in a prime shooting area but the result was a takeaway for SJ instead of a shot for Pitkanen. He had 2 or 3 other chances from about the same spot prior to that…I think that resulted in him taking too much time to try and make the prefect shot the next time.

  50. Oilman says:

    Oilman, if you were really confident in your point you’d look at lefty and righty goaltenders and see if there were any trends in where opponents shot at them.

    I wouldn’t know where to begin, and I wouldn’t want to. Again, I’m just going by anecdotal evidence. The most recent that I heard was from Cheechoo in a shootout vs. the Oilers where he said it was awkward to come in on a goalie with his glove on the wrong hand. Back to your muscle memory thing….most of these guys practice their shootout moves against guys who catch left. If you’re a right handed shooter who is used to coming down the right side and going low blocker, that makes a difference.

  51. Paulus says:

    I don’t understand where this enmity came from. Let’s just get along, shall we?

    This might be the Guinness speaking (I kinda have an Irish accent right now, very weird), but I have a suggestion: if anecdotes and memory are your vessel, put up a sail and go with it. Just don’t knock the amount of work poured into stats and their analysis.

    Statistics may imply generalization, but they also point out trends; not perfect, but adequate to suggest that which is to come. I’ve hated charts and their ilk since grade 7.

    Me, I don’t have the patience of Job that our resident stats hombres have.

  52. Bank Shot says:

    “I’m pissed that the posts after 200 go on a different page. How do I fix that?”

    Do you think it’s ever come up in the history of the internet?

    I’ve never seen a blog thread this long before.

  53. Oilman says:

    me either – no patience that is.

    Look, I don’t knock the work…I think some is valuable and some isn’t (goalposts, for example). These guys put in more time crunching numbers in their offtime than I’d ever care to – and good on them….but when the numbers guys say the forum used to be good when there was only them, or that I should learn to watch a game – well, I feel like going down the hall and punching one of our accountants – just in case he might be the one.

  54. Paulus says:

    Aye, good Oilman, and hence the dire plea for calm.

    You gotta know where folks come from, though. The frank diatribes may be a little much, but frustration rarely affords the prevalence of cooler heads.

  55. danny says:

    It’s about humility and common goals. A few key players on each side will filter the noise, embrace humility and eventually make strides forward for everyone else. I’ve seen it happen here.

    Theres just too little humility and a whole lotta noise going on.

    The good money is always on the people that can figure out whats most likely going to happen, the smart money is kept in your pocket because chances are something different will happen.

    12 months from now, some people are going to be right on things such as Jan Hejda, and people are going to be wrong on things such as MAB vs Grebeshkov.

    Vic might call the Hejda thing something a moron could see, stats would never have told the story of Grebs development.

    Humility people.

  56. Bruce says:

    Goalies have a difficult-to-measure impact on shots against, largely through puck retrieval and movement, rebound control, and housecleaning of stray pucks around the crease. I would imagine they have a much smaller input on shots at the positive end, largely through ability (or otherwise) to get the puck moving in the right direction. I would think the effect of the goalie getting into the shooters’ heads would be pretty small; some players might try to make that extra pass to beat the hot goalie, while others will simply fire more shots his way on the “blind squirrel” theory.

    Hard to think even a great puck control goalie like Marty Brodeur, whose teams always seem to have low shots against, having the kind of impact as, say, Nicklas Lidstrom on Corsi numbers. But he would surely have a non-zero positive effect.
    ***

    PS: There was one thread just over 200 posts during the Edmonton at Vancouver barnburner a few weeks back. Same thing happened then, a second page, confused the shit out of me for a while.

  57. Doogie says:

    I don’t think I ever said the goaltender can control the amount of shots…if I could, I’d be playing at a much higher level than I do. But they can have an effect on the number of shots. Like I said, it’s just not one of those things that can be measured (whether or not a shooter overthinks his shot to the point of not getting the shot off).

    That first sentence contradicts itself.

    The second, I think, misses the point. Shots directed at the net includes missed shots, and from what I’ve seen, you seem to be arguing that guys will try to “pick the corners” a little too much against a hot goalie, resulting in a missed shot…which still counts in Corsi and Fenwick +/-. If I’ve misunderstood you, please correct me.

  58. Oilman says:

    Paulus….I read your post with an Irish accent! Weird!

    I know all about frustration….and I should apologize for getting sucked into (or starting – I’m not sure anymore) a pissing match that had no chance of going anywhere.

    that said – I’ll still take the word of a player who doesn’t know a spreadsheet from a bedsheet describing how things effect his game, over a smart lawyer (see, I said smart), saying that the game is too fast for those things to matter.

  59. Oilman says:

    doogie…no, I mean that the player takes too much time and doesn’t get the shot off….maybe it doesn’t happen enough to matter, but I think it happens….in that sense, the first sentence doesn’t contradict itself.

  60. Alice says:

    A lot of time spent on “who’s responsible for the Corsi number” question. Why? To suggest that – out of context – Pronger’s number is 2nd worst, therefore the stat is meaningless, misses the point entirely. The useful comp in this case is not Pronger vs his teammates, or the absolute number itself, but – say – Pronger G20-40 vs Pronger G40-60. A movement in the number might point out something, maybe other teams have adjusted to something he’s doing, favoring a side, whatever. The stat Might point to something happening that can be fixed. So the stat is good – it can aid in decision making.
    Hard to figure what the number crunchers could have made of Lupul in the off-season though, do you believe his Ana stats or his Edm ones? Obviously someone read some Context into his Edm numbers, and picked [guessed] the other.
    So to argue over whether numbers are ‘accurate’ or not is retarded. The numbers are the numbers, we use them at our own discretion.

  61. Vic Ferrari says:

    I’ve posted this on a thread at IOF that’s over a week old, and will go further into it when time permits. Mostly I just hope that other people do the same. If you have basic Excel skills, all the data, and even some of the mathematical tools are available online at behindthenet.ca and timeonice.com.

    ..

    Using NHL game 600 as the midpoint of the season, and looking at how the the Corsi numbers from the first half to the second half for the 751 players.

    The correlation for Fenwick and Corsi for the individual players are overwhelming (.634 and .661 respectively). So the odds of that happening by pure coincidence:

    4,503,599,627,370,490 – to – 1

    There are big numbers and there are big numbers. This is a really big fucking number.

    The ZoneNumber correlation will be similar, possibly even a touch bigger, though I haven’t checked.

    .

    On the other hand, the correlation of “EVshooting% while a player was on the ice” is -.015 … indicating that there is nothing there at all. In terms of odds:

    About 3 – to – 1 of it being pure chance.

    So the confidence interval is far too small for a reasonable person to draw any relationship from it at all. EVshooting% simply doesn’t show repeatibility.

    Now if you filter out the goalies (who obviously have no effect on EVshooting%) it will get quite a bit stronger I’m sure.

    Still, the width of the gap between the two, in terms of repeatability, it should be measured in light years.

  62. Vic Ferrari says:

    Bruce:

    I’m not running you here. You make a genuine effort to support your arguments, and you notice a lot of things in a game that really did happen, and really did matter. Which should be more common than it is. I don’t know what games a lot of these people are watching, but it’s like they’ve been channeled to them through Glenn Healy’s subconscious.

    It’s a forest and trees issue Bruce. When talking about Corsi numbers, or even shots-on-goal and shots-against while a player was on the ice:

    You’re going to start thinking about specific shots and blocked shots. Dozens of plays that you saw and then saw again in slow motion. And of course not all shots are created equal. But that’s not the intent of the thing.

    Forest guy will think: Well, you don’t get shots directed at net unless you’ve got the puck in the right end of the rink. And he’s not going to be surprised that much by the way that it shakes out.

    So teams that end a lot of their shifts in the offensive end of the rink probably have a lot of offensive possession, and teams that have big Fenwick or Corsi numbers probably have the puck in the offensive end of the rink a lot.

    Some coaches are more shot happy (Babcock, Maurice) than others (Quenneville, Wilson). Still, the correlation between these two is very strong. The correlation between either one, and scoring more goals than you allow at evens is again very strong. Factor in goaltending, which is huge of course, and it just gets stronger. (because you can win a lot of games that you are outplayed in if you have a great goalie in your net)

    None of this is about persuading people to believe in magic. The exact opposite in fact. To me hockey is a simple game. And the breaks happen when they happen.

  63. Oilman says:

    Vic,

    I have no problem admitting that I have a problem seeing past the trees to get a good view of the forest in the statistics department. To me, there are a lot of subtle things that happen during a game that change its course, but it seems they are filtered out with the numbers. Are they? Or do they just not matter in the grand scheme of things? Where is the line drawn or does there even have to be a line?

  64. Vic Ferrari says:

    correlation of goal differential to points in the standings.

    .899 (Which is crazy strong) chances of that being coincidence alone are about 137 TRILLION to 1, which is a better way of presenting it than saying the confidence level is 99.99999999%. In any case would hope to god that everybody already realized that.

    The main point here is that I timed myself and it took me 35 seconds. Cut and paste http://www.timeonice.com/standings.html into Excel, delete the header rows for the western conference. Make a GF minus GA column (COLUMN ‘N’). Run a Pearson correlation (=CORREL(F2:F31,N2:N31)) and voila! .889

    Enter that in http://www.timeonice.com/coincidence.html
    and voila: 136,945,801,477 to 1.

    It really isn’t that hard to add value, to substantiate ones opinions. It takes a lot longer to write it down in a comment like this.

  65. Alice says:

    Vic,
    I think it’s helpful to point out how these numbers can be used by a GM to manage his roster, or a coach to manage his bench. Obviously, we all want to manage to a higher goal differential, win more games. How to get there is less obvious.
    For oilman: as GM, comp your corsi numbers across the league, across your division, across the span of the season. Are you high/low or average? Are you rising or falling?
    As coach looking at a 5-game losing skid. Corsi on track? Maybe you’ve just run into some bad luck finishing (the breaks happen when they happen, per Vic), which you might have thought by the seat of your pants, but now the numbers back it up. On the other hand, if your skid coincides with falling corsi’s then you probably need to take some action with your lines.

    As I said previous, the absolute number is less important than the vector, so don’t sweat the number in isolation.

  66. Bruce says:

    So teams that end a lot of their shifts in the offensive end of the rink probably have a lot of offensive possession, and teams that have big Fenwick or Corsi numbers probably have the puck in the offensive end of the rink a lot. … the correlation between these two is very strong. The correlation between either one, and scoring more goals than you allow at evens is again very strong.

    Well the comparison is imperfect cuz obviously special teams goals would have an impact here, but looking at the “Outshooting/Outshot by” stats at NHL.com shows there is a correlation but not a real strong one, between outshooting your opponent and winning the game. In games where one team outshot the other (1049 of 1094 to date), the “outshooters” have compiled a .576 points percentage this season, the “outshootees” .531. Surprisingly the gap is greater in three-point games (where the outshooters have “won” 129, their counterparts just 95), while in regulation the gap is just 428-397. Thus in regulation games where one team outshot the other, the outshooters have won 51.8% of the time. Hardly strikes me as one of those impressive-looking googolplex sized numbers you’ve been rolling out there, Vic. :)

    Factor in goaltending, which is huge of course, and it just gets stronger. (because you can win a lot of games that you are outplayed in if you have a great goalie in your net)

    Well how exactly are you going to factor in goaltending? Since shooters take shots and goalies stop them they are working to opposite purposes obviously. In fact the goalie will most negatively affect his team’s Corsi numbers by stopping the puck, cuz then play stays in his end, by and large, for more potantial shots. Whereas if he just lets it in the linesman will carry it out to centre for him every time. :)

    I’m being facetious of course, but there are plenty of games where the goalie is everything. I’ll bet the game Craig Anderson stopped 53 that the Islanders all had great Corsi numbers. So what, they lost.

    And I’d be willing to wager that the there’s a negative correlation between shots in the third period and winning the game (which is distinct of course from the winning the third period itself in which those shots are taken). But it’s fair to say that the context of the score clock affects the flow of playand the Corsi numbers that result.

    Finally, I don’t think the Corsi number does justice at all to counter attacking teams, the ones that soak up the pressure in their own end in exchange for fewer, higher-percentage opportunities on the rush. We saw this work well for the Oilers in 2006 playoff run when they were very significantly outshot in the three series that they won. In the extreme cases, there were three games where Oilers skaters’ collectively blocked more shots than they were credited with in the offensive end, and they won two of them and lost the third in double OT. Oilers played Carolina a lot more evenly on the shot clock and we all know what happened there. (I kept detailed track of this throughout those playoffs, have the numbers kicking around if you’re interested.) Bottom line though was that as a team and presumably as individuals, Oilers’ Corsi numbers sucked, and they sucked all the way to Game 7 of the SCF.

  67. Alice says:

    Bruce, I’ve been with you all the time, till here. Enjoy your posts, esp weathering the storm with slipper. Jeez!
    But… if the corsi number doesn’t always correlate to winning, then you throw it out? Use it to help you Manage – that’s all it and the rest are for! The SC run is a good example, actually, Oilers went very far on some very timely scoring (Pisani, Peca, Torres come to mind) and some very hot goaltending. They rode that a long way, but the corsi numbers would [probably] reflect what it felt like by the seat of your pants – it was a pretty dicy run, and all the more exciting because of it. We didn’t roll over anybody that spring.

  68. Tyler says:

    We saw this work well for the Oilers in 2006 playoff run when they were very significantly outshot in the three series that they won.

    Is this right? My recollection is that things got a lot more even after Detroit. Looking at the SJ series, the home team outshot the road team in each game. Anaheim definitely outshot the Oilers but they’re a shoot from anywhere team too.

  69. Bruce says:

    PS: Vic, I’ve done an exhaustive game-by-game breakdown of goal differential, and after factoring in ersatz shootout “goals” have determined that the team that scores the most goals wins the game 100.0000000000% of the time. The correlation is an astonishingly perfect 1:1.

    (Just kidding of course, but I think that rather obvious point was the source of some of the scrummage higher up the page. No duh there’s a relationship between goal differential and winning.)

  70. Vic Ferrari says:

    Bruce:

    You have bet on exactly the wrong horse here.

    The 2006 Oilers were a very strong Corsi team. I think we all saw that, they dominated possession and outchanced the opposition a lot. A combination of dodgy goaltending and some shitty puck luck pushed them down into the 8th seed.

    In terms of Corsi, they were right there with Dallas, Anaheim (oddly) and San Jose. A light year behind Detroit, and of course they were dominated in the series in every category except goaltending and goalposts.

    Once they got past them they were favourites at home for every game through the playoffs. And rightly so. Slight underdogs in every series, though they would have been slight favourites in every series with home ice advantage.

    You can go back and read everything I wrote during the playoff run, Bruce. I’m nothing if not consistent. As I said after DET were out and OTT fell behind BUF in their series … the planets were aligning for the Oilers.

  71. Alice says:

    And I can’t believe I’m falling on the numbers side, because in their raw form I hate them as much as anyone else. But in context they can illuminate and inform – and yes, mislead too. Take with salt, take them for what they’re worth, use with caution, but to say they’re not useful because they’re not purely predictive… well that makes them no worse than any other scheme, does it?

  72. Bruce says:

    Hi Alice, and thanks. (Thank you too Tyler/MC for your kind comment up the page.)

    No of course you don’t throw it out, I actually think the Corsi number is a pretty useful tool, I just don’t accept it as anything near absolute. Different teams, different coaches, different styles of play, teams rely on their goaltending differently, etc. I’m sure Oilers’ Corsi numbers were far superior in regular season ’05-06 when it became a team imperative to carry the play and not let the other guys shoot on Fricklin and Frakkanen, than they were in the playoffs when Roloson and the stalwart defensive crew would routinely turn aside 50 shots a game among them.

    Tyler, I have those 2006 playoff-run stats on a different computer, let me dig them up and I’ll get back to you.

  73. Vic Ferrari says:

    Bruce:

    That correlation between goal differential may be insultingly obvious to you, but it’s a stunner to a huge percentage of Oiler fans. Seriously. Because it ignores clutchness.

    For the Oilers, a three minute check.

    Look at the Oilers for the first 600 games (more or less the midpoint). Figure out the exact midpoint if you’re keen.:
    http://timeonice.com/teamshots.php?team=EDM&first=20001&last=20600
    and since:
    http://timeonice.com/teamshots.php?team=EDM&first=20601&last=21230

    Now if you believe that players are as good as they are, and the ability to get the puck out of your end and keep it in the other end is a repeatable talent … well you’d be right.

    For all the changing of roles on ths squad, and the oddness of guys like Stortini, Grebeshkov, Souray and Nilsson playing tough minutes for stretches here and there. And Penner playing centre, etc. And with a lot of the rookies improving dramatically in this regard (Gagner) or losing their confidence (Gilbert). And so on.

    The repeatability of the Corsi numbers for the individual Oiler players (sans goalies)is terrific in spite of all this. 68 to 1 against coincidence.

    On ice EVshooting% and EVsave% for the skaters, which you are often referring to, by definition, when you look at short stretches of good GF/GA numbers for players whose underlying numbers are staying about the same … no correlation, no repeatability.

    In fact a slight negative correlation in both cases. Meaning that in this case, if a player had a terrible shooting% when he was on the ice for the first half, it would say absolutely nothing about how his second half would go.

    Because shit happens, and the wheel has no memory.

    I love goals as a stat too, Bruce. We all do. But over just a dozen games or so, there is so much noise that the signal is imperceptible. And even over a long stretch, it’ll balance out for most guys, but an honest person has to check the underlying numbers, and the players history, to make sure that it wasn’t just a stretch of bad luck and/or bad circumstance.

    Also, I like Stortini too, just not as much as you. :) He is what he is, but he tries like hell, leaves some marks on the boards every night, and seems to be able to keep goading the opposition into taking stupid penalties. And though I think that Glencross and Brodziak are both far better players, that’s been a pretty effective fourth line recently.

    I do think that they aren’t as hard on the puck as they get credit for around here though. It’s all relative to the rest of this young team which struggles with that, and is probably a lot of games and experience away from being good at it.

  74. Vic Ferrari says:

    alice:

    Completely the opposite of what I’m saying.

    The numbers I’m talking about, which do NOT come out of the blue, we’re looking at them for a reason.

    These have more predictive strength that 99% of the things that are the obsession of hockey fans in general, and Canadian hockey fans in particular. SEE my recent posts above. And I could go on all freaking day if time permitted.

    It is NOT a matter of looking at the stats differently so much as it is a matter of looking at the game differently.

    And some people don’t want to look at the game differently. And that may very well be wise. People who understand less are often happier, in all things in life. And they should find forums where their ideas and observations will be embraced with any expectation of support or evidence beyond the anecdotal.

    I know it’s not all about me, nobody owns the internet, but if I wanted to watch RiversQ, mudcrutch, slipper, showerhead, mike w, etc argue with clouds … I would pay them to exactly that.

  75. Oilman says:

    Vic,

    You never cease to amaze me. Unbeleivable.

  76. Slipper says:

    The “well duh” sentiment pretty much covers it, Bruce. The whole point of trying to place a value to an individual player’s contribution to winning is essentially trying to figure their overall effect on goal differential. Whenever I get a response like “well duh, this just in, Detroit is good…”, well that’s the basic point I’m trying to make: the good teams have good goal differential and, as a whole, good shot differential. Simple relationship. Simple stuff.

    I think Alice was refering to me suggesting to throw the Corsi numbers out, which is kind of absurd. So Alice, thanks for piling on late. Atleast you’ve exemplified excellent comprehension skills. My point about Pronger and the Ducks was to prove the Corsi number useless and to show how irrelevant context is. Spot on.

    Bruce, I think you continously make a good point that teams regularly exceed statistical expectations. I don’t think many people around here disagree with that and it’s probably what drives all the content on luck and randomness. 7 games is a small sample size and anything can happen.

    Yet what’s a greater indicator of future success: 82 games of outshooting your opposition or 82 games with an exceedingly high shooting percentage? I think Vic has proven that the former is more sustainable than the latter.

  77. Vic Ferrari says:

    One more thing before I stop beating the ever loving Jesus out of this horse.

    I wrote something about Zone Number before on IOF. The impetus was to find a way of trying to quantify what Tortorella had been saying about Lecavalier for years, over and over, and often viciously. That his shifts end badly far too often.

    Anyway, I made the mistake of googling for this recently and found myself being absolutely trashed on different message boards.

    Anyhow, to me it had seemed too obvious to bother arguing with people. Apparently I was wrong.

    If you look at the Oilers, player by player, first half of these season compared to the last … it is the most repeatable stat I can find.

    The odds of a correlation being pure coincidence are 1414 to 1.

    Of course there is next to know correlation in standard newspaper +/-. And absolutely no correlation whatsoever in a players ability to score game winning goals.

    This not because I’m magic. This because hockey is a simple game. And the world is round.

  78. Bruce says:

    You have bet on exactly the wrong horse here.

    Yeah, I shoulda been a Carolina fan. First of all, Vic, see my comments above (written before I read yours immediately above it) w.r.t. the regular season and playoff Oilers of ’05-06. The same team, but different teams due largely to a change in the nets.

    I had no idea of the existence of the ‘sphere back then but I was a hockey fan in those days :), and I followed that Oiler playoff run like no other in a statistical sense. There was no NHL.com event summaries in the 1980s when I was REALLY into hockey. :D In 2006 I didn’t have some of the new-fangled stuff but I did have those event summaries, and I worked out the “Team Corsi numbers” by simply adding up shots on goal, missed shots, and opposition blocked shots for each team, each Oiler game. I even worked it out on a per 60 basis at the end of each series. So let me give you those (S + MS + OBS = Attempted shots/60 | G/60):

    Detroit (6.86 games)
    34.7 + 18.1 + 19.3 = 72.1 | 2.48
    Edmonton
    22.8 + 14.0 + 14.1 = 50.9 | 2.77

    San Jose (6.71 games)
    26.1 + 12.5 + 15.7 = 54.3 | 1.79
    Edmonton
    26.5 + 13.0 + 11.3 = 50.8 | 2.83

    Anaheim (5 games)
    36.6 + 13.2 + 15.4 = 65.4 | 2.60
    Edmonton
    24.2 + 7.8 + 9.2 = 41.2 | 3.20

    Carolina (7.06 games)
    23.2 + 12.0 + 14.7 = 50.0 | 2.69
    Edmonton
    28.3 + 15.2 + 16.4 = 59.9 | 2.27

    Now I find that pretty illuminating that there appears to be a negative correlation between Corsi numbers and winning. The Oilers were outshot in all three series they won, quite badly in two of them; but the Oil carried the play to Carolina and lost. The higher-shooting team lost every series, and were outscored in every series.

    Before you get all apoplectic on me, Vic, let me be the first to say this is not only a small sample but a biased sample as it studies only one team. However, I hope you will agree with me that analyzing playoff series where only two teams are involved over a long stretch of games, can sometimes sharpen up the signal-to-noise ratio within certain contexts. It’s like looking at an image taken through a narrow-band filter. Lots is filtered out, but what gets through in that wavelength can be pretty rich. And in this wavelength specific to this team at that time, statistics appear to support the theory that whatever they may have done during the season, in the post-season they succeeded through great goaltending and bend-but-not-break defence with occasional counterattacks or less-than-sustained pressure at the other guys’ end of the rink. (And Vic I’ll bet the zone faceoff stats were similarly lopsided, I didn’t think to keep track of those. But I agree with Alice and with Corsi, the puck was in our end an awful lot.)

    These differences are not small; Oilers were outshot (meaning all shots) in 16 of the 17 games they played in winning the Western Conference, and then they outshot Carolina in 7 out of 7 games in the final. Edm outscored their western opponents 54-42, and were outscored in the final 19-16. So the four series winners attempted more shots than their opponent in precisely one game of 24.

    The one exception was a truly exceptional game, the “Ryan Smyth” game against the Sharks where Smytty got his chiclets knocked out by Pronger, but came back to set up Horcoff in triple OT to get Oil back in the series. In that game the Oil peppered an astonishing 118 shots at Toskala compared to just 71 on Roli. Oilers threw everything but the kitchen sink at the Sharks that night, exceptional enough that the shots for that whole series were a little more balanced than the others. But in Game 4 the Oilers may have outshot the Sharks 35-25 but the Sharks had 36 other shot attempts (inc. 24! blocks) and the Oilers just 18. Sharks therefore had 61 attempted shots to 53, but it was the Oilers shots that were getting through to the net. And it was more Oilers’ shots that wound up in the net. That correlation is stronger; for example, in the 15 (sob!) games won by the Oilers, some 47% (459 of 982) of the opponents’ attempted shots actually made it through to goal; and in the 9 games they lost, the opposition’s shooting accuracy was 54% (301 shots on goal out of 555 attempts). In the offensive zone, 51% of Oilers’ shots got through in their wins, 47% in their losses.

    The Sharks series was particularly illuminating, as the Sharks percentage of shots through to goal followed this progression through two opening wins and then four and out: 58, 58, 48, 41, 47 and 38%. Meanwhile, the percentage of Oiler attempts that reached goal was: 47, 44, 49, 66, 62, and 52%. In the two games where over 60% of their shots made it through, they scored 6 goals in each. So I wonder if the ratio of shots/attempts is somehow reflective of quality of scoring opportunity, or just that the shooters were hitting their spots that night. For sure Oilers became more successful in winning as they defended shots from reaching their goalie and as they got more of their shots through to the opposing goalie; seems reasonable but the gross Corsi numbers will never show it.

    In the 15 wins the Oilers literally doubled their opponents in shooting percentage, 12.7 % to 6.3 %, in their 9 losses the reverse held although not quite so decisively, 6.8% to 10.6%. There is a strong correlation here of course, the winning team will have the higher percentage most nights by virtue of having scored at least one more goal. It’s not always true, e.g. Game 1 in Detroit where the Wings officially outshot the Oil 57-25 (and a whopping 113-49 all told) yet won only 3-2. One would expect a correlation above 90% here unless you start getting into shootout baloney which of course has nothing to do with what happened in the game. But perhaps the more interesting correlation is that in this identified group of games where Oilers got more of their attempted shots on goal, a higher percentage of those shots went into the goal. And the same holds for their opposition. Subtle differences, but enough to suggest that when counting up Corsi numbers, to simply give all three categories of attempted shots a value of 1 might not be very precise.

    Now in the finals the Oilers changed their game, attempting a whopping 70 more shots over the series than Carolina did. Was this reacting to a different opponent? A different style of play in the other conference? A team that played the counterattack game even more patiently than they did? Forgot the game plan during the 37-day layover to allow Carolina to properly rest up? Or, was it the difference in the Oiler nets, where Roli played just a fraction of a game before the Fricklin/Frakkanen combo moved back in? For sure those finals resembled that regular season in terms of flow of play and the disappointing results that frequently ensued.

    I don’t want to put it all on Jussi Rebounds, either, who after a difficult Game 2 responded by allowing just 8 goals in the last 5 games. Roli did have an off-night before Ladd put him out, and Conklin’s brief “relief” appearance wasn’t. But the Oilers team save percentage by series is pretty illuminating: Detroit .929; San Jose .931; Anaheim .929; Carolina .884.

    I’ll stop now and throw this out there while the thread is still alive. Lots more grist for the mill here, this was an extremely interesting case study. Maybe it is the 1 in 234,567,890,123,456,789 exception that proves the rule, and because I’ve only really studied this one exception my interpretation is skewed, but you can see where I’m coming from when I question the value of Corsi numbers as a predictive indicator. In the above case study, the relationship between Corsi numbers and success is strongly inverse. Of course it’s done from a team perspective, not indivduals within that team, but if I can’t establish “good” Corsi numbers = success from the macro viewpoint, in my view it calls the whole metric into question.

  79. PerformanceOil says:

    Long time lurker (great blog BTW Lowetide), was curious about something and felt the urge to make the plunge.

    I’m not sure how valuable it is to say that Corsi numbers show high reproducibility for a player, without addressing whether the Corsi numbers actually mean anything for individual players (note: players, not teams).

    Given that I haven’t seen this number yet (at least I don’t think any of the numbers you gave refer to that correlation, you seem to be focusing more on reproducibility), I’m curious Vic, what is the correlation between a player’s goal differential and their Corsi stat?

    Thanks.

  80. Bruce says:

    Whenever I get a response like “well duh, this just in, Detroit is good…”, well that’s the basic point I’m trying to make: the good teams have good goal differential and, as a whole, good shot differential. Simple relationship. Simple stuff.

    It’s not so simple to say Henrik Zetterberg is better than Ilya Kovalchuk when one plays with Datsyuk and Lidstrom all the time and the other one, uh, doesn’t. And that’s what I was questioning in the “This just in” comment that apparently made it all the way through. :)

    I think Alice was refering to me suggesting to throw the Corsi numbers out, which is kind of absurd. So Alice, thanks for piling on late.

    As I read it, Alice was questioning my post immediately above when she said “why throw out Corsi numbers?” And she wasn’t piling on at all. So happens I agree with her, they’re useful indicators to help interpret a game and manage the team, but they don’t necessarily drive results.

    Anaheim definitely outshot the Oilers but they’re a shoot from anywhere team too.

    A style-of-play decision which will have a bigger effect on Corsi numbers than it does on the scoreboard. One of many factors reducing the correlation between the two.

  81. Slipper says:

    /A lot of time spent on “who’s responsible for the Corsi number” question. Why? To suggest that – out of context – Pronger’s number is 2nd worst, therefore the stat is meaningless, misses the point entirely. The useful comp in this case is not Pronger vs his teammates, or the absolute number itself, but – say – Pronger G20-40 vs Pronger G40-60. A movement in the number might point out something, maybe other teams have adjusted to something he’s doing, favoring a side, whatever. The stat Might point to something happening that can be fixed. So the stat is good – it can aid in decision making./

    This is the post I was refering to, and I stand by my interpretation, but it doesn’t matter as I’m in a less foul mood this afternoon.

    As for Zetterberg with Datysuk in comparison to Kovi with Hossa, or whatever. I’ve made a million and one reference to opposition and linemate quality, and I have to come to the eventual conclusion that it falls on blind eyes. I mean thanks for preaching to the choir.

    I thought my Pronger comparison made it pretty clear that I’m of the belief that all expression aren’t created equal. Similar to strength of oppostion numbers on BTN, there’s more to drawn form comparing things relative to one own teammates than to compare the numbers straight up across the board. Hemsky looks miles ahead of his own teammates by Vic’s timeonice site, but doesn’t seem to hold a candle next to the numbers of Zetterberg. Simple common sense dictates that one is a greater indicator of ability than the other. Saddle #83 up next to Zetterberg and, of course, it’s a total different story.

    Again, nobody has ever claimed that these metrics be taken at face value and without context. Ever.

    The kid’s are going to be slaughtered at even strength was a commonly held belief by people across the sphere going inot the season. I think that being outshot by nearly three hundred over the course of the season confirms that. I don’t need Bruce to be convivned of this. Factor in the lesser light opponents they face and the favourable game situations MacTavish uses them in, and a plethora of other options that add context, and their results are even uglier.

    The fact that they’ve shown the ability to produce points is an obvious positive. I don’t think they’ve yet to show a consistent contribution to winning real hockey games, but I’m of the mind that a team with 17 or 18 regulation wins over the course of a season isn’t very strong.

    As for playoff, 28 games is a miniscule sample size, and teams regularly exceed their statisitical limitation over that short of a period all the time.

  82. Vic Ferrari says:

    PerformanceOil said:

    I’m curious Vic, what is the correlation between a player’s goal differential and their Corsi stat?

    I’ll trust that you did this yourself and saw that the result was infinity. The code at coincidence.html is written in javascript, so you can read it using ‘view source’ and modify it if you really need to know. The chances of it being coincidence are far greater and one-thousand-trillion to one. Beyond that I can’t help you.

    Obviously this correlation is going to be overwhelming, because EV+/- is a stat built directly from EVshots and EVshooting%-while-you-were-on-the-ice and EVsave%-while-you-were-on-the-ice

    So the correlation to all three will be enormous, by definition.

    Corsi has the strongest correlation right now, after 80 games, and the gap will widen more yet before the last game of the season. For reasons that are intuitively obvious.

    But if you looked at short stretches then the more luck-driven numbers (EVshooting% and EVsave% behind you) are going to ring louder than the talent, because one or two bounces can really swing those all over the place.

    To me this all just seems flagrantly obvious. Is anyone following me here?

  83. Slipper says:

    And instead of being a cantankerous prick, I’ll respond to the post of Alice’s I quoted.

    I don’t think I made any effort to prove the stat usless by citing Pronger be second worst on the Ducks by the Corsi metric. I thought, by providing evidence of his linemates and game situation I was providing context to illustrate that other factors can drive down specific player’s results. Some detractors of said metrics might feel that the numbers are meaningless thrown about without context, and thus can’t be relied on. All I’m arguing is that they are extrememly useful, but that nobody is oblivious to the fact they need to be taken in greater context to have meaning derived from them.

    Chris Pronger is the best D on the Ducks, and due to that he’s used for more own zone draws and against the other team’s top players. Carlyle is notorious for employing and old school checking line strategy so when used in this manner Pronger’s teammate are most likely going to be of the Pahlsson, Moen, R.Neidermyer- all who have the worst Corsi numbers on the Ducks- and all these factors play a role in driving down Pronger’s results.

    I’m sorry for trying to suck the context out of the equation.

  84. Oilman says:

    slipper – what’s a good sample size for this type of analysis? (and I have no ulterior motives to asking this question – no need to come at me with guns blazing). 28 games is nearly 35% of a regular season.

  85. PerformanceOil says:

    Thanks Vic,

    I hadn’t done it yet, actually, but while I was waiting I tried to do something quickly.

    You’re clearly using different numbers from me, since I came out with a very poor correlation (R^2 = ~0.2, and that is only after excluding player with <20 games or 10 minutes/game). I used the Corsi numbers and +/- from BtN (which does have the problem of empty net goals I think, but otherwise should be fine, no?), threw it all into excel and made a simple scatter plot.

    Although your data may, in fact, be superior, I’m not sure I’m following why the three would correlate. Two players with the same Corsi values could have very different Sht% and Sv% numbers (and their respective teammates might have very different Sht% numbers as well), resulting in wildly different +/-. Are you answering a question different from the one I’m asking, or have I misunderstood you or the data somehow?

  86. Bruce says:

    To me this all just seems flagrantly obvious. Is anyone following me here?

    I think I’m just in a lower Universe, Vic.

    I agree with Slipper that context is EVERYTHING, and every statistical measurement is an imperfect lens. You can accept one system at most in your belief system that is “perfect” and I choose that to be Wins and Losses and I don’t mind Ties as they appear in the historical records from past centuries but all others are to be handled with suspicion. I measure winning in this hierarchy:

    Stanley Cup wins
    Playoff series wins
    Game wins
    Goals

    … and already the last is diverging from what’s important because who scores next in a 6-1 game likely has nothing at all to do with what might be important at that moment. Or even who wins next when it’s Game 81 and you wrapped up first place last week. But all those moments combined, all those games, and all those wins, build to the Stanley Cup and the level even beyond that, Dynasty.

    So that is the system I correlate other metrics to, and there’s none of them perfect other than the inherently obvious ones. I mean even if you tried to correlate shutouts:wins or ENG:wins or 5-goal-scorers:wins, you’ll find a real big number above >0.99 but it would be less than 1. Only if you choose something definitive, such as my sarcastic “whoever scores the most goals always wins” above, will you find that true correlation. I know you are not saying there is, Vic, but when you start throwing around numbers like “double-digit digits to 1″ it comes across that way.

    Some metrics are stronger than others, just like some games are bigger than others, and that is where context is critical. I agree with pretty much everything Slipper wrote above in that respect, meaning I agree that Corsi number isn’t the be-all and end-all when it comes to measuring, say, CFP. I’ve been trying to say the same thing, albeit in some other language.

    Believe me I commend you guys for the stuff you are pursuing, much of which is in many ways better than what we had. That said, I’m certainly NOT prepared to abandon “traditional” metrics because we have new ones. There’s just a broader spectrum to consider, and that’s fantastic.

    Comparing the metrics against each other is in my view a real good, though never perfect, way to establish that context, to bring things into finer focus. A guy who starts in his own end a lot will see an effect on his minus, no doubt. It might not explain his entire minus, but it’s a consideration. (To include modern metrics, traditional statistics and Brad Richards all in one swell foop.)

    But as you guys may have blazed a trail by questioning conventional wisdom — and good on ya — I reserve the right to question the “new conventional wisdom”. Meaning you. There’s no antipathy intended, quite the opposite, although I have been surprised on occasion by how quickly the sticks can come up around here. Do lawyers, accountants and recreational mathematicians secretly want to be lumberjacks?

    The intent is to advance the process, to not chisel anything in stone (other than as many raw numbers as possible), but to keep examining the method for exceptions and strange loops and non sequiturs. Sometimes it’s easier to establish a correlation by measuring the imperfections.

    It’s also important to establish what the metric actually measures. e.g. the Corsi number might better reflect the territorial play of the game than its outcome. And that’s not a bad thing, it’s providing another shade of context. Just don’t build a glass house around it and try to make it something it’s not. It’s pretty, but ultimately it’s just another colour.

  87. Vic Ferrari says:

    PerformanceOil:

    Huge props for making an effort. If you ever need a place to post anything you’re digging into, contact one of us at IOF. Sometimes good stuff gets lost in comments sections.

    As for corsi correlation, I obviously used the numbers from timeonice.com. Last time I checked Gabe’s corsi numbers were actually EV Fenwick numbers relative to ice time with empty netters included.

    And even though I feel that every time I get mathy I lose credibility as a hockey guy … I’ll get into it.

    Pearson correlation coefficients themselves can be misleading if you’re looking at big sample sizes. Because a smaller number has a more powerful meaning.

    I used a t-distribution here for the sake of consistency (though the sample sizes for most of it are so big that normal or binomial are going to give the same result anyways. But the computer is doing the work, so what the hell.)

    I used s = r * SQRT(n-2) / (1 – r^2).

    Where r is the Pearson correlation and n is the sample size. And s is the the number of deviations under a t-distribution with n-2 degrees of freedom. Then I took the cumulative probability, and calced the odds against randomness. Essentially the polar opposite of p-value. And for commonsensical reasons I looked only at the non-real possibility.

    I would appreciate it if one of you guys would check my scripting there. I couldn’t find any t-dist tables that ran out that far, so I could only checked lower-odds probablities against manual math … didthat once and it worked, so I assumed that I hadn’t mistyped anywhere.

    as an aside: We’re not talking about causation here in the pure sense of sqrt(r) from a Pearson correlation. Because shooting shots for the helluvit won’t make you a better team, but teams who have a lot more shots directed forward than backward are likely to have had the puck more.

    Plus if you add the causation from faceoff zones and Corsi alone, you are over 100%, which doesn’t make sense.

    Alternatively, you can eliminate the goalies manually, and run the correlations for each team individually, then parlay them through. You’ll get even bigger numbers than I’ve posted here.

    But the numbers indicating the odds of pure chance … they are millions of times bigger than they need to be to make the point anyways. In short, whether you’re crushing an ant with a 10 lb sledgehammer or a 5 lb sledgehammer … it’s a convincing crushing of the ant.

  88. Vic Ferrari says:

    Bruce

    Those are such long posts that I’ll have to get back to them later. I know I write too many essays too, but in that sense it’s probably a bit of the “other people’s farts stink … my own farts smell kind of neat!” syndrome.

    Props for throwing up some data. Though at first blush that would seem to support what I’m saying completely, no?

    I did catch the bit about Corsi numbers not having macro predictability or correlation to winning. And this statement of yours is spectacularly wrong.

    And I know that you have the will to dig up info … why not look at the forest?

    take the total EV stats by team: GF, GA, corsi, faceoff-zones, save%.

    The correlations for all, relative to EV goal differential are strong.

    Now just sum up corsi numbers and def-off zone faceoffs for each team, just use those wrong numbers (it’s crude, but it’s not like you’re going to be writing game lines on Monday, this should make the point).

    Also, if you pretend that every team had had the same EVsave% (which is very reasonable IMO, goalies drive that a helluva lot more than teams and coaches) … then it’s even more stunning.

    .

    With that, I hope that, with my standing as the tenth most important blogger on the Oilogosphere, and my belligerent, combative nature, that I’ve motivated at least some of the old guard of the ‘sphere to defend the place.

    If someone says “Yeah, but I watched the game” … and the right response is “So did I, and you’re a dumbass” … then I’m really hoping that gets said.

    I won’t have the time to post around here much in the coming months, I really hope that I come back to the same place that I left, but I don’t think that’s likely. Not that me being around would make much more than a sniff of difference anyways, still …

  89. Slipper says:

    Goodbye Norma Jean
    Though I never knew you at all
    You had the grace to hold yourself
    While those around you crawled
    They crawled out of the woodwork
    And they whispered into your brain
    They set you on the treadmill
    And they made you change your name

    And it seems to me you lived your life
    Like a candle in the wind
    Never knowing who to cling to
    When the rain set in
    And I would have liked to have known you
    But I was just a kid
    Your candle burned out long before
    Your legend ever did

    Loneliness was tough
    The toughest role you ever played
    Hollywood created a superstar
    And pain was the price you paid
    Even when you died
    Oh the press still hounded you
    All the papers had to say
    Was that Marilyn was found in the nude

    Goodbye Norma Jean
    From the young man in the 22nd row
    Who sees you as something as more than sexual
    More than just our Marilyn Monroe

  90. Vic Ferrari says:

    Aw, thanks slipper.

    I thought you’d be angry lately, usually 4 month stretches of taking 20 points over the natural hold for the home team leads directly to overconfidence and Lexus leases from you guys. Nice to see that you’ve kept your perspective through this unfortunate Oiler streak recently. And presumably kept your hedge. Huge props.

    AND: Anyone who uses the word “coach” in their internet handle is probably a spotty 14 year old redheaded boy. Treat them as such.

    Peace out

  91. PerformanceOil says:

    Vic,

    Pretty sure you’re off on a few things.

    First, you’re either using significance wrong, or I’m misunderstanding you big time.

    A high level of significance does not mean the correlation is stronger, the two are completely unrelated. Correlation is asking how much variable Y changes as you change variable X. Significance is asking how reliable the correlation you get from a set of data is (i.e. how likely the result would be given completely random data – the numbers you keep using, or how reproducible the correlation value you get is).

    An example: say you have 2 die 6. You want to know if there is correlation between the two. So you roll both dice, and they both come up one. Then you roll both again, and they both come up two – fluky, no? Correlation between the two dice is perfect – 1.00. However, if you ran those numbers through a significance test, I’m sure they would fail badly, the sample size is much too small to make any meaningful conclusions. As the sample size increased, you would expect the correlation to fall towards zero, since the two dice should be completely independent variables.

    As far as a correlation meaning more with a larger sample size: not true, at least not in the manner you phrased it. A larger sample size means that the correlation value you get is more likely to be accurate. Nothing more or less than that. If you get a low correlation, it doesn’t matter how large your sample size is, it is still a low correlation.

    Just so, the correlation between Corsi and a individual’s +/- is 0.2 (much lower using all the player data actually). That means the variance the two values shares is about 4%. That is very low (actually I am surprised that it is so low, even though I was a bit skeptical of Corsi at the individual level, for the reasons that Bruce named).

    In all, Corsi is an extremely poor predictor of a player’s +/-. This is hard to explain, unless quality of shot (both ways) is a huge factor (and quality of linemates too I’d guess). I’d suggest role to be important as well.

    That said, it is clearly an influencer, though only weakly.

    Thus, in all, one cannot measure the impact of a player by their Corsi number, unless you want to argue that shots for/against drive results more than goals for/against, or that using the +/- and Corsi numbers from BtH are inherently flawed.

    I think measuring scoring chances for/against would be a much better predictor of +/-, but unfortunately that data is subjective and hard to come by.

    In other words, the results Stortini are getting matter a lot more then his Corsi number. Moreover, while the results he is getting may in fact be due to stochastic factors, rather than actual ability, his Corsi number is a very poor predictor of what his ‘true’ ability is.

    Finally, none of the data can predict how good of a player he will be in 1 year, or 10. I would suggest based on his apparent tool-set, the guy has a pretty low ceiling. Then again, he is a rookie (which seems to be often overlooked), he fills a role and I think he has shown he can pass and shoot to some degree of ability. In addition, I think he is quickly becoming part of this teams’ core (right age!) and while I could be dissuaded otherwise, it won’t be by using his Corsi number.

    My eyeballs agree with Bruce’s, and the stats don’t say either of us are wrong.

  92. Bruce says:

    Those are such long posts that I’ll have to get back to them later. I know I write too many essays too, but in that sense it’s probably a bit of the “other people’s farts stink … my own farts smell kind of neat!” syndrome.

    Thanks for reading, Vic.

    Props for throwing up some data. Though at first blush that would seem to support what I’m saying completely, no?

    No.

    I did catch the bit about Corsi numbers not having macro predictability or correlation to winning. And this statement of yours is spectacularly wrong.

    Well it was spectacularly right in the 2006 playoffs, which is the only analysis I have done in serious detail. But obviously I’m not going to convince you that my example is even an exception to your Rule. It’s all just clattering dice, eh, and Oilers got lucky.

  93. Slipper says:

    Performance Oil:

    Aren’t the shots the more controled factor in the equation, though?

    I’m asking since you’ve shown an aptitude with this.

    The SvPct and ShtPct are the random variables, aren’t they? They fluctuate from player to player, goalie to goalie, and from season to season.

    A player has more control over the end of the rink the pucks being played in and the direction the shots are heading than they do over the percentage of shots stopped, at either end. Am I wrong?

    Bruce mentioned above, somewhere along the way, about it being indicative of where the puck is, which is what what I’ve been arguing with him for the last three weeks about.

    But say even if a during an extended streak two goalies are put up phenomenal .940+ %. These things tend to normalize over an extended period of time. Instead of 6 goals for every one hundred shots, 8 or 9 go through, driving toward the league average. If the percentages are out of the control of the players with the curved sticks, yet they percentages are what determine the goal differential, isn’t the greater indicator of extedned perfromance 1)which net they carry the play toward more and 2)the shots (chances) they create and allow?

  94. Bruce says:

    Thus, in all, one cannot measure the impact of a player by their Corsi number, unless you want to argue that shots for/against drive results more than goals for/against

    PO: That’s the way that I read stuff from the Ferrari camp, that shots for/against are what’s real and what goes into the net is just a chance outcome which will gradually normalize to those all-important shots as the goalies ultimately conform to the law of averages. To put it mildly, I don’t agree, but I could just be reading it all wrong. Sometimes Vic’s logic is on a higher plane and I have trouble following along. Especially if I have to smell his farts.

    Bruce mentioned above, somewhere along the way, about it being indicative of where the puck is, which is what what I’ve been arguing with him for the last three weeks about.

    Slipper: It’s hardly a perfect indicator, as someone pointed out Anaheim for example are/were a “shoot it from anywhere” team so might generate more positive Corsi-type results than their actual zone time warrants. But Corsi numbers are probably rough indicators of territorial play. I am more dubious as to how indicative they are of true scoring chances, as they seriously undervalue the sudden strike of the counter attack. The Oilers ’06 playoff example cited above showed that strategy could work fine even with Corsi stacked against them.

    I don’t think we’re that far apart on this, Slipper, it’s just we’re approaching the middle ground from opposite sides. I see shots as a means, not an end. What some teams achieve with quantity, others do with quality. I’ve got a good case study on this as well that I’ll unroll one of these times.

    All I’m arguing is that they are extrememly useful, but that nobody is oblivious to the fact they need to be taken in greater context to have meaning derived from them.

    On this point we entirely agree.

  95. PerformanceOil says:

    Slipper:
    Aren’t the shots the more controled factor in the equation, though?

    Hell if I know. Like I said, I didn’t think Corsi should be the end all be all, but the correlation was way worse than I would have predicted. You can’t score if you don’t shoot and if the other team is drilling your goalie with shots, one has got to find its way through; makes intuitive sense, right?

    Here’s the thing though, Shots, Sv% and Sht% (and TOI of course) all contribute to +/-. If you normalized Sht% and Sv% for everyone, then Corsi would correlate perfectly with +/- (maybe that’s what Vic meant?).

    Problem is, once you start doing that, you are making assumptions that might not be valid.

    Is it really hard to believe that certain players would take low %’age shots from everywhere, whereas other players tend to shoot from higher %’age areas?

    Likewise, are all shots equal to all goalies? We already know that all players aren’t equal in allowing the opposition to direct shots toward the net, but is it hard to believe that some players are better at minimizing high percentage shots against, and maybe they (or their team’s system) don’t give a damn about the low %’age ones?

    Now, if you accept that variability, multiply it by 12 guys on the ice at once. Then, it starts to make sense that the Corsi would correlate so poorly.

    So the question is, should we factor out all those things as being due to chance?

    I really don’t think so. Those things drive results as much as quantity of shot and quality of goaltender does, and I think those things are influenced by player ability, teammate ability, opposition ability, the team’s system and good old puck luck. You can’t just filter it all out and say ‘this is what a player is worth’. What you can do though, is say ‘this is player X’s Corsi number, and if it is higher, that player is somewhat more likely to have a good +/-’ (I’d guess the correlation is probably higher on the very high end of the Corsi scale as well, but it is just a guess). That’s not useless at all, but it isn’t the end of the analysis.

    In the absence of controlled experiments, we are left with this: Stortini is getting positive results on this team, this year, with his linemates, in the role given to him by MacT. Nothing more, nothing less.

    If you watch him and believe he and his line are getting goals they probably don’t deserve, and/or letting through 5 alarm chances that their goalie miraculously saves, then yeah, maybe the results aren’t sustainable.

    Thing is, I watch Stortini, and the sense I get (not scientific at all), is that his line generates more positive than negative, regardless of the shot clock (and not by dumb luck either). It seems when he and his line are on the ice, either they pin the opposition in their end for the whole shift, or the vice versa happens. But I think the quality of chances favour Stortini and co a lot of the time (others may disagree).

    Ideally, you want to find numbers to quantify it all, to make it more objective. Corsi is a step in the right direction, but only a very small step. How do you quantify the ‘aggitator effect’? You can dismiss it as observor bias, and indeed it might be so. Can that be proven though? What about chemistry? Is GBS more than the sum of their parts? Are they just lucky? Who knows? I certainly don’t, though I do believe the three factors I listed are important.

    Trying to ask questions is good. Working to answer them is better. Being certain that you have answered them, probably means you are wrong. Being arrogant about it is stupid.

  96. Oilman says:

    I like this guy;o)

  97. Bruce says:

    I like this guy;o)

    So do I. Well said, PO, I think we’re on the same wavelength here (that’ll be my last spectrum analogy in this thread :).

    Aren’t the shots the more controled factor in the equation, though?
    I’m asking since you’ve shown an aptitude with this.
    The SvPct and ShtPct are the random variables, aren’t they? They fluctuate from player to player, goalie to goalie, and from season to season.

    They’re all random variables, Slipper. Shots may be less random than others, but do not get fooled into thinking they are somehow sacrosanct. They vary, within a range.

    As for shooting percentage being a random variable, it is, but it’s also random within a (wider) range. When Player A’s shooting percentage fluctuates over 7 years between 13 and 18% every year, and his teammate Player B’s jitters between 6 and 11%, do you think they will both normalize towards 9% any time soon? How much do you care that Player B has a slightly better Corsi rating? It takes him more shots to score fewer goals, that’s part of the equation. He fucking better have a good Corsi number. Whereas A might be a Corsi minus and a plus player, because shots taken when he’s on the ice have a significantly better chance of going in the net. I sure know which one I want on my team.

    (Bonus points to the person who identifies A and B.)

    By the same token, if a team’s save percentage is .915 when Player C is on the bench and .860 when he’s out there, you can wait for those nuymbers to normalize out, or you can try and figure out why there are so many ten-bell scoring chances against when C is out there. The information cuts both ways, and to write huge chunks of it off to random chance would be simplistic and frankly, wrong.

    Thing is, I watch Stortini, and the sense I get (not scientific at all), is that his line generates more positive than negative, regardless of the shot clock (and not by dumb luck either).

    Agreed, PO. Sure the sample size is small, but guess what: coaches work with small sample sizes all the time when making their decisions. Since 20 and 46 were first put together, 46 has been +14/-5 in two-and-a-half hours of mostly kick-butt hockey. That’s what I’ve seen, and the scoreboard seems to agree. I don’t imagine MacT expects the outscore rate to be sustainable — I certainly don’t — but when the combo does things like out-Corsi their opposition 14-0 as GlenX incredibly did in the third period last night (in just 3:34!), that metric confirms the other that this chemistry is working and let’s just leave it at that. Yes Alice and Slipper, you can and certainly should use such information as Corsi to help you manage the team, but it’s just a tool, it’s not the Grand Unified Theory or anything.

  98. PerformanceOil says:

    Well said, PO, I think we’re on the same wavelength here

    Thanks, and I agree we think similarly. Moderation in everything, eh?

    Now that I got all this out of my system, I can go back to watching the clock, anxiously waiting for the massive game that is to come. No matter what you think of Stortini’s underlying numbers, how can tomorrow not feel like the biggest, most exciting game in two years?

    22 hours.

  99. Bruce says:

    I wrote:

    Since 20 and 46 were first put together, 46 has been +14/-5 in two-and-a-half hours of mostly kick-butt hockey.

    Small error. For the record, +11/-4.

  100. Vic Ferrari says:

    Performance Oil

    First off, the corsi numbers at behindthenet.ca aren’t corsi numbers at all. Desjardins has an error in his script that reverses the blocked shot counts.

    Pop quiz: Do you know why this is important.

    The result is a mishmash. Which would explain, at least in some part, why your conclusions, and your thinking in general, is so bizarre, and so distant from the simple game of hockey itself. At least to my eye.

    I actually advised you of the things to look for, and where to get the data. You failed to do either.

    Keep an eye on IOF, I’ll take the time to completely rip apart your assertions with uncontrovertable facts. No offense, but your post is just so wrong I really am spoiled for great places to start(see clues above, for Christmas sakes, this is really simple, people). I’m going to be pressed for time this spring and into summer, but I will get back to it for certain.

    In an effort to redeem yourself, you could do what I suggested Bruce do. It’s not hard.

« Older Comments

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

© Copyright - Lowetide.ca