OILERS RE 12-13: FORWARDS

I am projecting a strong improvement for Oiler forwards, mostly at the top end and among the gifted kids. The forwards might be playoff worthy. Injuries will factor in and we shouldn’t forget about the veterans, but new coach Ralph Krueger has four young guns and plenty of youthful support behind them.

The top 6F is young, extremely skilled and should drive this train in 12-13:

2012-13 PROJECTIONS

NAME G A PTS Diff
TAYLOR HALL 66 35 32 67 +14
RYAN NUGENT HOPKINS 72 22 45 67 +15
JORDAN EBERLE 75 24 31 55 -21
SAM GAGNER  74 19 36 55 +8
ALES HEMSKY 60 14 40 54 +18
NAIL YAKUPOV 77 18 26 44 Nil
RYAN SMYTH 69 16 22 38 -8
SHAWN HORCOFF 75 11 16 27 -7
MAGNUS PAAJARVI  67 5 16 21 +13
RYAN JONES 70 12 8 20 -13
ERIC BELANGER 77, 10 10 20 +4
TEEMU HARTIKAINEN 50 9 8 17 +12
BEN EAGER 50 4 4 8 -5
LENNART PETRELL 53 4 4 8 -1
DARCY HORDICHUK  44 0 2 2 -1
TOTAL 203 300 503

 2011-12 TOTALS

NAME G A PTS
JORDAN EBERLE 78 34 42 76
TAYLOR HALL 61 27 26 53
RYAN NUGENT-HOPKINS 62 18 34 52
SAM GAGNER 75 18 29 47
RYAN SMYTH 82 19 27 46
ALES HEMSKY 69 10 26 36
SHAWN HORCOFF 81 13 21 34
RYAN JONES 79 17 16 33
ERIC BELANGER 78 4 12 16
BEN EAGER 63 8 5 13
LENNART PETRELL 60 4 5 9
MAGNUS PAAJARVI 41 2 6 8
ANTON LANDER 56 2 4 6
TEEMU HARTIKAINEN 17 2 3 5
LINUS OMARK 14 3 0 3
DARCY HORDICHUK  43 1 2 3
TOTAL 182 258 440

The Oilers score about 10% more overall in my 2012-13 model, so as you can see most of those totals were gobbled up by the forwards. I have the two #1 overalls who have NHL experience taking over the offense and then Eberle, Gagner and Hemsky following them. Rookie Yakupov and veteran Ryan Smyth will also get their share of prime minutes, and I have the offense ending around three players who will battle each other for playing time all year long (barring trade): Ryan Jones, Teemu Hartikainen and Magnus Paajarvi.

Horcoff and Belanger spend another season in their own zone and the trio of Eager, Petrell and Hordichuk spend time in the lineup and in the pressbox.

I think we could talk about playoffs based on the forwards. I keep hoping for a ‘mentor’ forward to be added but maybe Horcoff is that player and Gagner can begin to take on some of the ZS chores.

Up next: the blue.

written by

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

107 Responses to "OILERS RE 12-13: FORWARDS"

  1. Halfwise says:

    So is the basic debate around the forwards “Too young, small and fragile, except for the old and overpriced” versus “So gifted that they will score as if they were sharing 90 minutes not 60?”

    I love the young talent and have eternal hope that the vets will contribute better than last year. But the season is long (wear and tear) and the games are short, in terms of finding minutes for the kids while playing them safely against men.

  2. Lowetide says:

    Halfwise: Very true, but the gifted forwards are money once the puck gets to them. The problem is–and has been–getting them the puck. The veteran F’s Edmonton needs to count on–Horcoff, Belanger, Smyth–are older men and only Smyth is coming off a good season.

    I don’t think the Oilers have a choice based on the roster, they will have to run a designated hard match line of Horcoff-Smyth-Jones–at least to begin the season.

  3. sliderule says:

    If Yakupov doesn’t get at least 25goals and 55pts we are not going to be in the running for a playoff spot.
    I am hopefull Halls shoulder won’t slow him down too much but I expect a slow start from him

  4. oilswell says:

    I guess we’ll see Krueger’s strategy, but I could imagine where ZS is less of a tool for him than matchups due to him mixing older kids onto harder opposition. As an example, would Hall+Horcoff+Hemsky harder match against the top opposition get things done at both ends of the rink? That could make managing the rest of the roster a lot easier.

  5. raventalon40 says:

    What is the number in the far left column next to their names?

  6. fuzzy muppet says:

    raventalon40,

    Games played

  7. rickithebear says:

    oilswell,

    look at WOWY.
    History says Hall-Horc-XXX against 1st/2nd.

  8. DirtyOil says:

    Lowetide,

    First time poster, LONG time reader. I know line combinations have been beaten to death, but Lowetide, what do you think of the following lineup:

    Yakupov-Hopkins-Eberle
    Hall-Gagner-Hemsky
    Smyth-Horcoff-Jones
    Eager-Belanger-Petrell

    PB: Hordichuk
    First Callups: Paajarvi, Harti, Lander

    Or do you think Paajarvi/Harti will get the push above guys like Petrell/Eager? In fact, who gets the first push, Paajarvi or Harti?

    If I was coach I would be giving Belanger and Petrell 1st PK unit since their 5×4 and 5×5 time will be limited at best (~9 mins?) as they should be. 2nd PK unit I would have a rotating cast between Horcoff, Smyth, Jones, Hall, and Eberle, just to keep the opposition honest.

    Can we talk about PP too or am I stealing your thunder for future articles….??

  9. hunter1909 says:

    LMAO @ Jordan Eberle with 55 points in 75 games.

    Bad Lowetide, trolling his own blogbase.

  10. Jesse says:

    Lowetide,

    I was reading an article over at NHLNumbers about NHLE and noticed that the conversion rate for the WHL (0.30) was equivalent to the OHL and a slight bit higher than the QMJHL (0.28); but when I watched the Memorial Cup, the Oil Kings seemed way out of their league compared to the other teams. Do you think this was just due to the Kings’ grueling schedule leading up to the Memorial Cup? Or do you think they actually were outmatched? If the latter, what does that say about the NHLE ratings, if anything?

  11. Jesse says:

    Or am I simply misunderstanding NHLE?

  12. Jordan says:

    hunter1909,

    HA!

    You’re full of it Hunter. I think that suggesting Eberle “only” has 55 points in 75 games next year is perfectly reasonable – As a line in the sand for what would be a an okay year for the “junior NHLer” (if we’re using college terminology). I also hope that he does better than that (and know he certainly could) but suggesting that as a base line for expectations does make sense, based on Sh% if nothing else. Not saying he won’t keep a higher than average Sh% throughout his carreer, but he was really shooting the lights out with his 18.9% last year – that’s not sustainable.

    As for the rest of them, they fit the narrative that LT has built quite well – except perhaps for Hall breaking out. I’m not sold he will after his surgery this year. I’d wager on him being below a PPG this season, even if he does miss the start of the season.

    Was listening to crew on Team 1260 this morning – can you imagine what the F depth would look like with Doan in the mix? I mean, I can’t beleive it happens, but man… that would sure add a whole lot of what we lack in the top 6. I’d love to see him on RNH’s wing and push Ebs and Hemmner down – what a gong show of a forwards crew that would be!

  13. DirtyOil says:

    Jordan,

    I think if Hall breaks out this year with ~70 points, he will drag his 5×5 linemates with him.

    Last year it seems fairly obvious Eberle was driving the offensive bus year-round, this next year it might be difficult to pinpoint who exactly is driving the bus and who is riding the bus.

  14. jonrmcleod says:

    LT, is it reasonable that the Oilers will only use 15 forwards in 2012-13? You have that group playing 979 games. That leaves 5 games unaccounted for if the Oilers use 12 forwards each game. Who plays those 5 games?

  15. DSF says:

    Jordan:
    hunter1909,

    HA!

    You’re full of it Hunter.I think that suggesting Eberle “only” has 55 points in 75 games next year is perfectly reasonable – As a line in the sand for what would be a an okay year for the “junior NHLer” (if we’re using college terminology).I also hope that he does better than that (and know he certainly could) but suggesting that as a base line for expectations does make sense, based on Sh% if nothing else.Not saying he won’t keep a higher than average Sh% throughout his carreer, but he was really shooting the lights out with his 18.9% last year – that’s not sustainable.

    As for the rest of them, they fit the narrative that LT has built quite well – except perhaps for Hall breaking out.I’m not sold he will after his surgery this year.I’d wager on him being below a PPG this season, even if he does miss the start of the season.

    Was listening to crew on Team 1260 this morning – can you imagine what the F depth would look like with Doan in the mix?I mean, I can’t beleive it happens, but man… that would sure add a whole lot of what we lack in the top 6.I’d love to see him on RNH’s wing and push Ebs and Hemmner down – what a gong show of a forwards crew that would be!

    Doan has already told both Calgary and Edmonton, thanks but no thanks.

  16. vishcosity says:

    Yesterday Vor wrote:

    “As I said in the last thread, the evidence is overwhelming that what we see is very often wrong. Statistics can help tell us if that is what is happening. Can we use the math to enforce our delusions? Sure. Can we use the math to enrich our understanding? Sure. The choice is ours. What is happening here, and on many sports blogs, is fans are saying our eyes trump the math rather than challenging how the math is being used.

    “If I am right about there being a problem with how we use SP data it doesn’t invalidate the math. It simply points out yet another statistic we need to learn how to generate and evaluate. Meanwhile, our eyes will continue to deceive us.”

    When math doesn’t reflect what my eyes see, there are two possibilities: either the math is whack, or my hockey IQ sucks. Generally at that point I go hunting for junk science. Sometimes I find it, sometimes its probably me.

    Yesterday Ricki wrote:

    In this they present an elevation view of the shot. by location in 5 areas. to give a further influence of the shot. i would break it up into
    1. over left shoulder
    2. Head
    3. over right shoulder
    4. chest area
    5. shoiulder-blocker arm swing
    6. Shoulder -glove arm swing
    7. block side non-arms swing area above pads.
    8. glove side non-arm swing area above pads.
    9. left pad side
    10. five hols
    11. right pad side.
    There are 74,000 shots in the season.
    I would like to know the general success rate for
    A even back hand 13ft on the rebound above the goalies pads under the blocker arm.
    the one factor that would be missing is that shot taken as the goalie is
    1. square to the position.
    2. sliding across on pads.
    3. diving across.

    One of the simplest statistical analytical models involves two electrons in a finite 2D box. The wave equation for any motion is really just a sine wave for each electron. Super simple. Can predict every force in play. Add things like truculence or poise and suddenly the math gets way more complex. For me I dare say its flat out impossible to consider all the electronic forces in a single shuffle in a single strand of DNA. Already at this point I need to bring in estimates and neglect a significant number of variables. And that is for one rearrangement of one strand of one protein, to model the molecules in a hockey game is way beyond my comprehension.

    So to me, until all the variables are considered, any formula is going to be neglecting potentially pertinent data. Enter Discretion.

    So discretion comes into which variables to consider, and the relative weight to give each.

    Enter Lennard Petrell.

    The Vollman graph, iirc, showed Petrell left and north of the axis, in his first NHL season. But the corsi and boxcar numbers suggest he’s not going to contribute to any offense really at all. And while we all agree there are many hats, its unclear to me that while statistics are clearly missing potentially relevant variables, that we can say math can out perform my eyes when it comes to assessing players and value to a cap pressed team. Or, instead of my eyes continuing to deceive, possibly my eyes are finding evidence that the mathematical models can’t yet replace saw him good.

    Including some of the metrics that ricki describes above could well resolve the discrepancy between the two schools. But how to assemble the data?

  17. vishcosity says:

    DSF: Doan has already told both Calgary and Edmonton, thanks but no thanks.

    Source?

  18. OilClog says:

    These kids are going to score.. Eberle will near ppg status again this season, so will a few others. It’s more about keeping it out of our own net.

  19. DBO says:

    Been wondering about Hall at centre.If he could pull a Messier and move to the middle, allowing us to move Gagner for a better top 4 defender (Bogosian maybe), it makes our lineup quite interesting.

    Yakupov-Nuge-Eberle
    Harti-Hall-Hemsky
    Smyth-Horc-MPS
    Eager-Belanger-Jones

    Or if you move A RW like Hemsky

    Hall-Gagner-Eberle
    Harti-Nuge-Yak
    Smyth-Horc-MPS
    Eager-Belanger-Jones

    Not sure which is better, bottom lineup younger and places way more tough zone starts on Ebs and Hall. I have been against moving gagner, but I think I like the first lineup better, especially since the Hall line and the Horc line can play toughs and still shelter Eberle and Nuge, giving them easier zone starts and secondary matchups.

  20. DSF says:

    vishcosity:
    DSF: Doan has already told both Calgary and Edmonton, thanks but no thanks.

    Source?

    Read it online about a week ago…can’t remember where I saw it.

  21. jake70 says:

    Cheers to the canadian women soccer ladies, nice work….now on to the Americans.

  22. vishcosity says:

    commonfan14: http://tvasports.ca/tvasports/hockey/archives/2012/07/20120724-151919.html

    That kind of language doesn’t fly.

    The internet. Where some random reporter claims to be in contact with an agent who apparently said something (which may actually be intended as leverage) which is immediately retweeted on 60 different sites. Back in the old days only lawyers would try and sell that kind of evidence as fact. So I figure either you’re showing your stripes or those logical leaps have spread beyond the british accreditation regency. As for me, I assume the latter and blame it directly on LA Law.

  23. rickithebear says:

    VOR: When talking Shot

    We are dealing with a pitched ball:
    1. point of release (O zone location)
    2. type of pitch
    3. Distance of pitch
    4. velocity of pitch
    5. path of pitch Shot release to Goal. Tip, Block, clear path screened
    6. size of strike zone (goalie position)

    Situation is also dictated by who is
    1. pitching (shooting)
    2. hitting (goalie)
    The results for baseball can vary. In Hockey it is binary. Goal/No Goal.

    Hockey play is Dynamic
    But we have defined dynamic influence factors
    1. zone start.
    2. Qual teamates (wowy)
    3. Qual Comp (who you defend against)
    4. Phase of Pocession

    Now we discuss corsi and other values.
    The analysis of the dynamic aspects of the game leads to one point.
    The shot.
    I believe in the comcept of Corsi.
    But hear is the Key.
    Corsi takes us to the point of puck release.
    Then thats it .
    end of time.

    CORSI TELLS US FUCK ALL BEYOND THE FINAL CONTACT OF PUCK ON THE STICK.

    From this point there is the more critical factors that dictate result Goal/No Goal. We have the ability to establish the probability

    Now alot of people here stop at Corsi and say this is the best reference. Really, Really!
    I thought goals won games.
    So people who obviously understand the importance of the goal. look at GF, GA, +/- Fairly accurate result measure.

    Basically in baseball terms the dymamics leading to the shot is all the peripherals like the pitcher scratching his ass up to realease of Pitch (corsi)
    Then in the hockey world some look at the GA, GF, +/- results.
    In baseball terms ignoring the pitch from release to hitter and looking at result hit, strike, Ball, Balk, WP. ETC (results)

    Baseball does not ignore the dynamics of the pitch or the action of the hitter.
    So why do we? We do not.
    We are forced to give a generic stat for goal scoring SH% (WHIP) and goaltending SV% (batting average)

    Baseball anouncers convey the situation for each pitch.
    The hockey announcers convey the shot and its factors both live and thru replay.

    Hockey just needs to record the Data.
    Create shot situational splits just as baseball does with pitching and hitting.

    VOR. the other day you said Allan ryder made an error by saying the shot had effect. Allan ryder did the most important analysis of the game to date.
    It is the best way to provide value from shot release to goal net. it needs to be added to.

    Now i have my favorites here: Woodguy, Coach, Steve Smith etc. and this last period i had you at the top.

    A lot of you guys just do not get it,
    Shot is the process of getting Goals.
    Creating Situational factor based shot results allows us to Give real value to Corsi.
    We will be a small sample game because of shot count per year basically in around
    60 shots/game with a variance of 26 to 34/ team 74,000/season

    I hear sample size, Sample size.
    think of Baseball:
    Baseball since 2000 290 pitches per game so 705,000 pitches. we get equal data count every 9.5 seasons for everyone of thiers.

    Do not be mad that Allan ryder and myself want to fill in hockeys version of Pitcher to Catcher

  24. TigerUnderGlass says:

    Am I wrong in my understanding that Corsi is intended as a proxy for time of possession? If the purpose of your data is to determine who is carrying possession then how exactly is shot quality relevant? You call the data useless because you want to use it in ways for which it was not intended.

    I don’t see how a claim can be made that possession data is useless either. You’ve been using baseball, so I’ll go elsewhere too. Football tracks possession. Why? Possession doesn’t equal points on the board does it? Yet it’s been proven that one of the most effective ways to win a football game is to dominate the clock. Obviously a hockey team dominating the possession battle has a leg up on winning the game, even if it doesn’t ensure a win.

  25. DSF says:

    TigerUnderGlass:
    Am I wrong in my understanding that Corsi is intended as a proxy for time of possession? If the purpose of your data is to determine who is carrying possession then how exactly is shot quality relevant? You call the data useless because you want to use it in ways for which it was not intended.

    I don’t see how a claim can be made that possession data is useless either. You’ve been using baseball, so I’ll go elsewhere too.Football tracks possession. Why? Possession doesn’t equal points on the board does it? Yet it’s been proven that one of the most effective ways to win a football game is to dominate the clock. Obviously a hockey team dominating the possession battle has a leg up on winning the game, even if it doesn’t ensure a win.

    Of course, there is a simple way to track possession…it’s called a stopwatch.

  26. Bar_Qu says:

    vishcosity,

    TigerUnderGlass,

    Thanks gents. Perfect.

    Ricki, this is what the confusion was about. You are:

    a) trying to set up a statistical model that has nearly unlimited variations

    II) arguing the wrong value of the stat

    I appreciate how these two post-ers clarified the issue. I can now go back to skipping Ricki’s blizzards of data in happy oblivion.

  27. vishcosity says:

    rickithebear:

    Baseball anouncers convey the situation for each pitch.
    The hockey announcers convey the shot and its factors both live and thru replay.

    Hockey just needs to record the Data.
    Create shot situational splits just as baseball does with pitching and hitting.

    ^This.

    I think that is what I think is missing. When the numbers don’t line up with what I see, then I think the math is whack.

    If the NHL could record that kind of data, and make it public, then there is a chance that the math comes closer to the saw him good analysis. I think these are critical considerations, without which, it all sounds like OHL stats that miss TOI.

    Until the math guys get access to critical data, the numbers may or may not reflect what is really going on.

  28. Woodguy says:

    rickithebear,

    CORSI TELLS US FUCK ALL BEYOND THE FINAL CONTACT OF PUCK ON THE STICK.

    No.

    TUG is on the right path. Shot differential (corsi) is used as a proxy for possesion in the ozone, and has been shown to have a very high correlation with winning.

    A better correlation than goal scoring actually.

    All the Shot Differential data that sites like BTN use comes from the NHL’s shot data.

    The NHL does track where every shot comes from as well.

    DSF can watch every game and use a stop watch and in the end he’ll find the Shot Differential ratios will be almost identical to the posseison differential via time in ozone.

    So when we talk about Corsi (Shot differential when that player is on the ice) in evaluating a player, we are talking about which players drive puck possesion in the ozone, which drives winning hockey games.

    If you are interested in this stuff go to nhlnumbers.com and read a bunch of stuff and make sure to read all the links in the articles as they usually refer to the previous work done to show various correlations of shot differential data.

  29. TigerUnderGlass says:

    DSF,

    Of course the data used to calculate corsi is readily available while the other is not. Why do you think a proxy was developed in the first place?

    Edit: or ignore this and read Woodguy’s comment above instead.

  30. leadfarmer says:

    DSF,

    I have also read it in a couple places, can’t remember where.

    Also, my problem with stats is it gives numbers without giving any underlying situational information. For example faceoffs. The difference between a poor faceoff taker and a decent one is 5% and a good one is another 5%(45% for poor, 50% for decent, 55% for good). That is only 1 in 20 faceoffs difference between the groupings. Player A could be winning all of his own zone and offensive faceoffs (the important ones) and loosing all the neutral zone faceoffs, while player B could be winning all the neutral zone faceoffs (the easy ones) and loosing all the important faceoffs. Therefor neutral zone faceoffs should be removed from the equation so the statistic is relevant

    Another example is shot percentage. Shot percentage tells you how many shots went on goal but not the quality of the shots. For example 2 goalies, 1 has great defense and faces 20 perimeter shots while the second goalie faces multiple odd man rushes. and faces 20 shots. If they both let in 2 goals they have the exacts same save percentage. Over a season there is some correction but not complete correction. Thats why I think shots should have a difficulty factor attached to them to help calculate save percentage as a percentage of the difficulty of shots faced. Something like the folling (only an example and no where near complete.
    3.0 for a 2 man breakaway
    2.8 for a Penalty shot
    2.5 for a breakaway.
    2.2 for a 2-1
    1.9 for a rebound in the slot on PP. 1.7 otherwise
    1.6 for a rebound in the slot on PP. 1.5 otherwise
    1.4 shot in the slot
    1.0 shot from within the blue line
    0.7 shot from side wall
    0.4 shot from outside blue line.

    If you attach these factors and also use them to determine the difficulty of shots that the goalie lets in as a replacement for GAA. you can actually make sense of who is a good goalie and who hides behind a good team defense.

  31. leadfarmer says:

    I forgot the Paajarvi fudge factor for shots: 0.1.

    J/k

  32. leadfarmer says:

    Before I get lynched, I don’t mind corsi as a stat.

  33. DSF says:

    TigerUnderGlass:
    DSF,

    Of course the data used to calculate corsi is readily available while the other is not. Why do you think a proxy was developed in the first place?

    Edit: or ignore this and read Woodguy’s comment above instead.

    It would be incredibly simple for a couple of trained chimps to start and stop a stop watch when the guys in black have it and the guys in white have it.

    They’ve been doing it forever in football and, although hockey is faster and possession changes are more frequent, using Corsi to track possession is like using a chainsaw to crack a walnut.

    Then, extrapolating Corsi as a method of evaluating individual players is just laughable.

    Paajarvi has decent Corsi but can’t put the puck in the ocean as one relevant example.

    Shot totals, especially when you include missed and blocked shots, with no reference to quality or result is just like being in a rocking chair….gives you something to do but doesn’t get you anywhere.

    Rickibear has this exactly right.

  34. Lowetide says:

    I just want to thank all of the posters who come on this site and give their insight. After a long day, it’s terrific to have a beer and read through the comments. I think WG makes a world of sense up there, and love that he’s talking about shot differential/Corsi.

    We need to make these things easier to understand for the layman. Because once the tumblers fall in the right order, the layman can move the conversation forward.

    And that’s the good stuff.

  35. Woodguy says:

    DSF,

    hockey is faster and possession changes are more frequent, using Corsi to track possession is like using a chainsaw to crack a walnut.
    Then, extrapolating Corsi as a method of evaluating individual players is just laughable.

    Are you going to back up that statement with data, or just go with “I think that’s right, so it must be right”.

  36. VOR says:

    I object to absolute statements made using math or statistics. Especially when the underlying logic seems clearly flawed. However, I hold absolute statements and lousy logic in the same low regard when they are propagated by the seen him good school.

    I understand it is nearly impossible for most people to accept that the evidence of their eyes is almost always flawed. It is even harder for most of us to accept that things are not simple. We want to believe our eyes and deploy the same simple heuristic trees in all situations. Often this is an incredibly effective strategy. However, when it fails us the result can be catastrophe.

    Statistics can be misused and abused in a myriad of ways. Even various mathematical tools can be misapplied. The thing is that when these metrics, tools, and techniques are used properly they can incredibly enrich our understanding of the world. That enrichment is not possible when we pick as our default position that our eyes always trump our math. It is exactly when the math is counter intuitive that we need to take it most seriously.

  37. "Steve Smith" says:

    VOR: It is exactly when the math is counter intuitive that we need to take it most seriously.

    Fuckin’ A.

  38. Lowetide says:

    I think we’re getting there, folks. These numbers have to be taken in context, so we probably need more than one Vollman sledgehammer:

    1. The one we have now.
    2. The one we have now but with straight Corsi so we can see if someone is being punished for playing on an exceptional team (this thought is based on a conversation I had with WG)..

    I remember when I got the first plus minus issue of the Hockey News, maybe 1975 summer. All of the California Golden Seals looked like death and the Flyers looked like world beaters. I’d like a CorsiRel to give me the best players within the team metric, and then a raw Corsi that tells me actual performance.

    Is that fair?

  39. DSF says:

    Woodguy:
    DSF,

    hockey is faster and possession changes are more frequent, using Corsi to track possession is like using a chainsaw to crack a walnut.
    Then, extrapolating Corsi as a method of evaluating individual players is just laughable.

    Are you going to back up that statement with data, or just go with “I think that’s right, so it must be right”.

    Back up what?

    That chimps can be trained to run a stop watch?

    Perhaps a little hyperbole.

  40. godot10 says:

    Exactly why was Renney fired if the Oilers are going to run with the line of Smyth-Horcoff-Jones again? That was one of Renney’s really dumb ideas.

    I will be extremely disappointed in Krueger is so uninspired in his vision.

    Jones is a 4th liner and PK’er. He is not one of the Oilers six best wingers.

  41. DSF says:

    VOR:
    I object to absolute statements made using math or statistics. Especially when the underlying logic seems clearly flawed. However, I hold absolute statements and lousy logic in the same low regard when they are propagated by the seen him good school.

    I understand it is nearly impossible for most people to accept that the evidence of their eyes is almost always flawed. It is even harder for most of us to accept that things are not simple. We want to believe our eyes and deploy the same simple heuristic trees in all situations. Often this is an incredibly effective strategy. However, when it fails us the result can be catastrophe.

    Statistics can be misused and abused in a myriad of ways. Even various mathematical tools can be misapplied. The thing is that when these metrics, tools, and techniques are used properly they can incredibly enrich our understanding of the world. That enrichment is not possible when we pick as our default position that our eyes always trump our math. It is exactly when the math is counter intuitive that we need to take it most seriously.

    You really need to spend a few hours looking at Chaos Theory and how just minor variables like a stick that doesn’t fit or a sore ankle or a disgruntled third liner can affect outcomes.

    When the math is counter intuitive it’s generally because it’s wrong.

  42. godot10 says:

    Krueger should go on the offensive and make the other coaches make tough decisions about their matchups.

    3 scoring lines, say with each anchored by a #1 overall. Force an extreme mismatch in your favor somewhere.

    I’m certainly hoping that creating mismatches by moving Hall around is what Krueger had in mind in that post-hiring interview. You are a scaredy cat coach if you run out Smyth-Horcoff-Jones.

  43. sliderule says:

    As an engineer I really like reading all the debate about all the stats and relevance of each set.

    As I am relativity new to the debates and analysis I feel that I am still learning even though I am probably much older than most of you.

    I think that there is much to be learned from corsi which is a time of possession stat but Corsi people should not dismiss plus minus and boxcars.

    The truly skilled players are not governed by all the stats we spout and that’s the true magic of our game

  44. Kris11 says:

    Boxcars adjusted for context like SH%, PDO, TOI, and Zonestart are very useful measures, especially in large sample sizes. You just have to keep in mind that boxcars vary a lot due to puck luck and context. That’s the whole idea behind so-called “advanced” stats; the boxcars can lie really badly because they don’t show the massively important effect of teammates, quality of opposition, and luck on goals for, goals against, and thu, plus/minus.

    Nobody thinks RelCorsi (or fenwick or whatever) is the whole story for a player. Its just a large part of the story in that it tells you how well that player helps his team get the puck going in the right direction in terms of scoring chances and shots. And the great thing is that corsi is predictive even with much smaller sample sizes than you need for the boxcars. (There are si many more shots than goals in a given game that luck determines how well you do by shots far less than it does how well you do on the score board. Thus, shot data can tell you how well a player is doing reliably with a far smaller data ste than goals.)

    Corsi tracks possession time and chances. In the NHL, quality of chances end up being pretty evenly distributed over large enough samples, so that’s not an issue.

  45. Kris11 says:

    DSF’s point about time of possession is the paradigm of all stat-skeptic responses.

    He says maybe somehow shots don’t track possession or vice versa. And maybe quality of chances matter. He then says that this shows these stats are useless.

    But he didn’t actually do the work of looking at -say 10 taped games of data- and recording shots, chances, and time of possession to look for a correlation.

    BTW, that’s what the stats guys would do. They would actually work to advance the conversation. Not DSF. DSF says the data would show that such and such, but he is too lazy to go get the data.

    Amazing. Guys like Tyler and Vic work for hours to show us something really cool about hockey. Guys like DSF come along and piss on it. Then the pissers say “Oh, well I could show that you’re wrong by looking at X, Y, and Z, but I’m too lazy.” And if somebody got the data, they wouldn’t even care,

    Why are they comsidered part of this conversation?

  46. Kris11 says:

    I mean if you want us to listen to you, and you think that you can show that the stats are misleading by showing such and such. Then do it. Otherwise you’re just listing possible weird ways that something might be wrong which any lazy fool can do.

    If Corsi is flawed and you can show it by recording game data with a watch, do it and show us.

  47. rickithebear says:

    Corsi is about Possession?

    No Corsi is about Shots directed.

    Like in football and soccer the possession bar would be a reflection of time of possession. In rugby they have a 1st phase, 2nd phase 3rd phase, 4th phase and the success rates.

    Pocession in Hockey:

    1. retrieval type
    2. time of pocession
    3. type of pocession: board, zone entry, counter, defensive.
    4. Phase of pocession.
    5. location of pocession.
    It is pretty easy game by game.

    that would be a start to analysing pocession.

  48. melancholyculkin says:

    In the NBA there are teams that have installed missile tracking cameras in their arena that allow them to precisely track every movement of the ball during the game. These teams now have mountains of useful data at their disposal for every possible situation in a game.

    The great limitation when it comes to hockey analysis is the data that is made available. I’d bet that the NHL game sheets miss over half the events that occur during a game, and incorrectly record half the events that they do track. The fact that people like Vic Ferrari and Desjardins and Eric T and others have been able to make so much out of such bad data truly is a testament to their abilities.

    If some type of tracking system were to be put in place like many NBA teams have, and the data is made public (there’s the catch), then we’d be cooking with gas and Corsi would likely become a dinosaur.

  49. melancholyculkin says:

    rickithebear,

    The thing that makes Corsi so useful is that it does (to an extent) capture things like puck retrieval and puck battles won and passes made and other things.

    You retrieve the puck, you gain possession and create a shot attempt. Corsi records that.

    You lose a puck battle, the other team gains possession and creates a shot attempt. Corsi records that.

    You make a brilliant outlet pass that leads to a scoring chance. Corsi records that.

    For such a simple measure the amount of hockey that Corsi captures really is remarkable.

  50. Cactus says:

    vishcosity: That kind of language doesn’t fly.

    The internet.Where some random reporter claims to be in contact with an agent who apparently said something (which may actually be intended as leverage) which is immediately retweeted on 60 different sites.Back in the old days only lawyers would try and sell that kind of evidence as fact.So I figure either you’re showing your stripes or those logical leaps have spread beyond the british accreditation regency.As for me, I assume the latter and blame it directly on LA Law.

    It makes no sense for Doan to consider Edmonton or Calgary, so even if that article isn’t credible, we shouldn’t be holding out any hope that he will appear. At this stage of his career, Doan will be motivated by one of two goals (assuming he doesn’t work out something in Phoenix):

    1) The best chance to win a Stanley Cup
    2) The largest possible contract.

    The Oilers cannot provide either of those next year and the Flames certainly can’t do the first and probably not the second.

  51. rickithebear says:

    Kris11: But he didn’t actually do the work of looking at -say 10 taped games of data- and recording shots, chances, and time of possession to look for a correlation.

    As you will remember i asked if anyone wanted to take a portion of shots the 74,000 shots and look at past history.

    Did not see you offer!

    FireTruuuuuuuuuuuuuuuuuuck!
    The concept of Corsi is not flawed.
    it is a point in time. when a puck is directed at the net.
    Meaning it was somewhere on the ice.
    Key word being somewhere.
    Heck one goal crease to the next from one goalie to the next. that occurs. A rebound shot in front of the net. both equal in corsi.
    And highly reflective of being equal pocession.
    I mean come on!
    How do you say that with straight face!

    In football (europe) the build up of pocession ball path is displayed to the point of shot. that would be reflective of pocession. listing who holds what were.

    It bothers me that all these technologies exist. i mean the game is waves of types of attack pocession. defensive postion. these waves of attack and defence paths could diplayed like radar maps on the weather channel. up to the point of shot, giveaway or retieval.

    A shot results in a Analysis of factors that give an expected result value that can be measured against actual.
    Giveway establishes opprnents next wave of pocession.

    Retrieval after a shot or dump in becomes the next phase.

    Corsi – Somwhere on the Ice at some time.

  52. "Steve Smith" says:

    rickithebear:FireTruuuuuuuuuuuuuuuuuuck!

    You are rapidly ascending to unintentional self-parody.

  53. VOR says:

    So DSF,

    Why should I spend time studying chaos theory? Are you suggesting it doesn’t work? That I don’t understand it? What is your point? How do you think it applies to hockey?

    I am really quite confused. Do you not read my posts?

  54. rickithebear says:

    melancholyculkin: For such a simple measure the amount of hockey that Corsi captures really is remarkable.

    That is why i say it is useful. to the point of shot.
    But shots directed per pocession says more.
    computer generated puck path is were we need to end up.

    A study was done on the success rate of scoring relative to dribble in the NBA the score rate was best with ball movement with no dribble.

    I want to be able to say that a breakaway shot created by a turnover is more succesful than a shot fromcenter ice on a line change.

    Kris11!
    Corsi says they are the same.
    Why don’t you look at the needed games of data to prove that.
    Cause assuming it is reflective of pocession is a huge error.
    Right VOR!

  55. DSF says:

    VOR:
    So DSF,

    Why should I spend time studying chaos theory? Are you suggesting it doesn’t work? That I don’t understand it? What is your point? How do you think it applies to hockey?

    I am really quite confused. Do you not read my posts?

    What doesn’t work?

    Chaos theory?

    By definition, it “doesn’t work”.

    Yes, I do read your posts and, yes, you are confused.

    There are so many variables in 60 seconds of hockey, there are so many variables that affect individual performance, basing any conclusion at all on who “drives possession” is just a silly notion.

    Hint: look at results.

    See Patrick O’Sullivan for reference.

  56. "Steve Smith" says:

    DSF,

    I had no idea that chaos theory made modelling impossible. Thank you for enlightening me. Please make sure to pass the word on to every scientist (social or hard) whose area of study is on a scale beyond what fits in a petri dish.

  57. rickithebear says:

    “Steve Smith”: You are rapidly ascending to unintentional self-parody.

    Unintentional! No.

  58. DSF says:

    “Steve Smith”:
    DSF,

    I had no idea that chaos theory made modelling impossible.Thank you for enlightening me.Please make sure to pass the word on to every scientist (social or hard) whose area of study is on a scale beyond what fits in a petri dish.

    Chaos theory doesn’t make modelling impossible.

    It’s just makes it incredibly imprecise.

    Imagine an animal that changes sexes in the absence of the other sex in order to procreate.

    Impossible!

  59. VOR says:

    Ricki thinks the more scattered the data the more affect and DSF believes Chaos Theory somehow prevents mathematical modeling. Now clearly neither has a background in math but I find their ignorance an appalling indictment of our educational system.

    Can anybody explain what DSF thinks I am wrong about?

  60. "Steve Smith" says:

    VOR: Can anybody explain what DSF thinks I am wrong about?

    Because you believe that it is possible to construct predictive and descriptive models from large amounts of data, you are necessarily saying that it is possible to devise a formula that will allow you to predict with absolute certainty exactly what will happen in any game of hockey.

    Alternatively, you are expressing support for statistics that are popular among people who disagree with DSF, therefore you must be wrong.

  61. Lowetide says:

    What is pocession?

  62. "Steve Smith" says:

    Lowetide,

    FIRETRUCK!

  63. jimbones100 says:

    Ha Ha Ha!! Love it. Now I feel satisfied all is right with the world and can go to bed.

    “Steve Smith”:
    Lowetide,

    FIRETRUCK!

  64. VOR says:

    Steve,

    Thank you. I thought he was objecting to my suggesting that when mathematical modeling and analysis contradict our beliefs about what we are observing we should pay attention. Then he brought up chaos theory confusing me. Chaos theory is counter intuitive and far from common sense. Yet it has changed the world and is thus a great example of the point I was trying to make.

  65. vishcosity says:

    I just has my computer crash on a super long post and now the beer has made me forget most of what I wrote. Just know that it was pretty good.

    I remember something about Rick Nash’s 2008 goal in Phoenix, I think I suggested that the lead up to the shot would need a difficulty multiplier of 10, but his tap in finish would get a 1 (metric being, if I could do it, it isn’t very hard).

    I think it also had something about weighting variables through multipliers determined by veterans of the saw him good school. If Ken Holland thinks MPS is unwilling to go to the hard areas, then my tendency would be to bend the math until it matches what Holland says.

    I just covered that in about 10% of the text. Maybe I should have my post deleted more often.

    VOR:
    Ricki thinks the more scattered the data the more affect and DSF believes Chaos Theory somehow prevents mathematical modeling. Now clearly neither has a background in math but I find their ignorance an appalling indictment of our educational system.

    Can anybody explain what DSF thinks I am wrong about?

    I’ll take a shot: DSF is unhappy with your protocol of coming to conclusions based on the math rather than trying to make the math fit a preconceived conclusion. (I think its kind of a lawyer thing, not necessarily about you.)

  66. "Steve Smith" says:

    VOR: I thought he was objecting to my suggesting that when mathematical modeling and analysis contradict our beliefs about what we are observing we should pay attention.

    He is, because chaos theory means that mathematical modelling doesn’t work.

  67. VOR says:

    The problem DSF, is that chaos theory makes mathematical modeling more precise. There are also many areas of human enquiry where it would alter a very robust model only slightly. I have no idea what selective hermaphrodites might have to do with chaos theory or hockey.

  68. Bar_Qu says:

    VOR:
    Ricki thinks the more scattered the data the more affect and DSF believes Chaos Theory somehow prevents mathematical modeling. Now clearly neither has a background in math but I find their ignorance an appalling indictment of our education system.

    My apologise for this. Even though they are only ten I will ensure my students receive a full dose of chaos theory and advanced statistics during their regularly scheduled LA time. They won’t know how to spell ‘possession’ but they will beat the tar outta math.
    ;-)

  69. Kris11 says:

    DSF,

    Is the extent of your knowledge of “chaos theory” what Jeff Daniels says in the movie “Jurassic Park” or did you actually read the book?

    —-

    Ricki,

    Tablecloooooooooth!!!!!!!

    Cool, radiant, nightmare.

  70. "Steve Smith" says:

    Kris11:
    DSF,

    Is the extent of your knowledge of “chaos theory” what Jeff Daniels says in the movie “Jurassic Park” or did you actually read the book?

    I do not think that “Daniels” means what you think that it means.

  71. Kris11 says:

    Not Jeff Daniels. The other guy. Whatever.

  72. Kris11 says:

    Firetruuuuuucck

  73. Bar_Qu says:

    “Steve Smith”: I do not think that “Daniels” means what you think that it means.

    Gold

  74. Kris11 says:

    Jeff Daniels was one of the Velociraptors (uncredited.) Its the performance of a lifetime. Really believable. Put on a ton of weight.

  75. Gerta Rauss says:

    Bar_Qu: Gold

    Blum

  76. vishcosity says:

    “Steve Smith”: He is, because chaos theory means that mathematical modelling doesn’t work.

    Before 1992, iirc, generally people did realized that mathematical models were limited to the variables they consider. Similar situation for cohort studies in nutrition, they even knew this in Framingham in 1972.

    I think the chaos theory was memorialized with the butterfly flapping its wings in china can create a tidal wave in Hawaii, or something. And while that is definitely a criticism of ignored variables in meteorology, I think the bigger picture of that bit of pop culture was that the probability is also really low, and us figuring out what was the original source is even lower. So it is true that knuckle balls from behind centre do sometimes go in on a goalie, the probability of it having any chance is pretty limited. Which gets back to the whole bit Ricki brought up, namely, shot chances and scoring chances are all well and good, but if they’re not qualified by the quality of the shot, they can’t really be very reliable.

    I absolutely believe modeling does work because I can calculate the revolutions of krypton at STP. Or, I could in college, I think. And certainly I’m with Vor on the desire to have a model so advanced that it bests my eyes regarding what is really happening out there. And while we’re never going to measure every variable in a hockey game, we can select the statistics differently and I believe create something much better than corsi or box cars could ever be. Including the metrics that Ricki described just today would be, imo, miles ahead of where we’re at with math and hockey right now.

  77. VOR says:

    Vishcosity,

    I really admire Ken Holland. However, the research shows that experts are even more likely than the lay public to let their prejudices blind them.

    I have been pondering a little experiment that goes to the heart of how I believe corsi should be used. I am in the middle of a large project where I am looking at whether corsi can be used to spot young players like Jamie Benn before they break out. I have no idea what I will find. Then I thought what if we went back and examined the Pit Martin for Phil Esposito trade. We all know I am a bit obsessed with hockey history so it seemed like a great little project. Would the corsi of the throw ins, who by box cars were all duds, have predicted the future of Ken Hodge and Fred Stanfield?

    Is there anyway to know? Could we reconstruct some of the advanced stats from historical records? I don’t know yet but I can reconstruct a few games.

    I am rambling, Mike’s Hard Lemonade in my case. What I am trying to say is I am not sure exactly what corsi tells us about puck possession, I can argue it either way. What I suspect is in young players corsi as an independent variable is predictive of future productiveness on the ice as measured by box cars and plus minus.

    I know that is a very bold statement and will need some serious proof. It would however unite two warring camps if I was right.

  78. "Steve Smith" says:

    I love this place.

  79. Gerta Rauss says:

    “Steve Smith”:
    I love this place.

    Fart

  80. Gerta Rauss says:

    I really need training camp to start. Or a trade. Or something.

  81. VOR says:

    I a cruelly, and I probably should have made this clear earlier think their is some legitimacy to Ricki’s idea. It is his certainty that is driving me nuts (that and some concerns about the quality of LA in the schools he attended but English might not be his first language. When I have had too many Mike’s it isn’t mine either.) I think to date research on shot quality is flawed by the belief that if a shot goes in it was a high quality shot. As a former goaltender I can tell you that even when sober I sometimes turned a flutterby into a canon.

    I also thinking that my problem with DSF may be that his understanding of chaos theory might be derived from popular culture and literature and that he may not understand where the theory and application has gone in the real world. The power of iteration was only one of many insights in chaos theory. mathematical modelling is deeply enriched by these insights.

    Nothing in chaos theory precludes modeling. All models are dealing in probabilities and mathematicians knew that before chaos theory. They also understood that if you miss a variable or misunderstand the relationship between them your model is screwed. Though I think Vishcosity already said that.

  82. VOR says:

    DSF,

    Please define incredibly imprecise.

  83. oilswell says:

    rickithebear,

    I have no idea what that means (I know what the WOWY part means…not the rest).

  84. "Steve Smith" says:

    VOR:
    DSF,

    Please define incredibly imprecise.

    The definition is hard to pinpoint.

  85. vishcosity says:

    VOR:
    What I suspect is in young players corsi as an independent variable is predictive of future productiveness on the ice as measured by box cars and plus minus.

    Are you set on corsi, like fur sur, that one? I mean, while we can’t even get TOI for the CHL there isn’t much hope on improvement maybe, but if you’re going to reconstruct games, I think it’d be super stellar if you were to look at each shot and try and find a metric to assess the quality of each one. Or, since I have nothing to do until Tuesday, if there are links to OHL games or whatever is your cross sectional group, I’ll do my ten game bit and try and use whatever metric you develop for any group of games you want. If its just about corsi though, then I don’t know. Until the shot quality is somehow considered, it seems arcane to count them all the same.

    Modulating the shot quality variable by some arbitrary scale will be most likely done best by a mind like Ken Holland. Is maybe what I was aiming to say. Or, if everyone agrees that shot number 17 in game #104 is awarded an 8, then the relative weight of that shot quality variable, only as it regards the initial development of the formula, the weight of that variable is something that very important and connected old white men all agree to say.

    So lets say KH et al say that MPS is a 7.6, then I think its the job of the metric development crew to find a way to have the player stat for MPS work out to 7.6. At least to get it running.

    So for instance, lets say we’re aiming to classify a player with an offensive rating and a defensive rating, each on the scale of 1 – 100. RNH last year may be a 85/43 while Horc gets 34/75.

    Or any other system metric with units. It drives me nuts when there are no units. it has to have units.

    Moving on.

    If the measure is measured by the following considerations:
    A = Goals/60
    B = Assists/60
    C = Shot total x average shot quality (.1 – .9), (or sum(SQ1…SQn), I think that’s the same)
    D = time of pocession
    E = poise quotient
    F = some other metric that tomorrow I will be ashamed I missed

    Then the offensive marker could be composed by:
    40%A + 20%B + 15%C + 10%D + 7%E + 5%F + 3%G
    or whatever percentage combination works to give Magnus a 76. And actually, instead of asking KH to assess each player, we could probably start with yahoo fantasy hockey rankings or something equally publicly traded, just to get the formula working kinda close to what is being seen good.

    Goaltending would be pretty interesting. Imagine finding a system that could well differentiate between a good goalie on a bad team and a bad goalie on a good team.

    Okay. And this is where I’m really going. Landing the plane.

    Each player could have some kind of value number, maybe one each for offense, defense, and poise, and when each player is entered into a second formula, when the totals from 2010 – 11 and prior are entered and the system predicts something relatively close to 2011 – 2012, then we know the model is starting to work. The task that I see is finding a way to modulate the first formula to make the second system’s result better. The tweek factor. Where the art happens.

    If that actually worked, we could plug Getzlaf in for Sammy and Petry and see if in the model, the team then would project to improve in the standings. Does the team actually project better without Eager?

    I guess it could be done now with corsi, or boxcars, but I think we can do it better if we could include even just some kind of measure of shot quality.

  86. delooper says:

    As an actual mathematician i’d like to request we tone down the “chaos theory this and that”. For one thing “chaos theory” isn’t really a subject, more the title of popular books on mathematics associated to global properties of deterministic dynamical systems. I don’t see an argument here you can’t make intelligent reliable predictions about hockey using advanced stats. The point of advanced stats is to make better predictions than we could make without them, not predicting the thought rushing through Eberle’s head when he scores his 500th goal. The former “chaos theory” says nothing about, the latter maybe a bit depending on your model of the brain. There’s a lot of interesting insights in chaos theory but practically speaking it tends to tell you more characteristic phenomena of dynamical systems. There is some new technology coming down the (academic) pipe called “topological statistics” which no auther has found a flashy name for yet but this is a subject soon to have a far deeper impact on humanity than chaos theory. Google “darpa robomule” to find out more!

  87. vishcosity says:

    DSF: Chaos theory doesn’t make modelling impossible.

    It’s just makes it incredibly imprecise.

    Imagine an animal that changes sexes in the absence of the other sex in order to procreate.

    Impossible!

    Well I’m sitting by a pool in San Diego, all the ladies went to bed, and my beer isn’t gone. With nothing to do until Tuesday, I’m going to troll the troll because no one has responded for what seems like forever.

    If a species is able to switch sexes, actually, not impossible, but entirely predictable because nature seems to abhor a vacuum. so with the right formula considering the probability of all insects being and staying the same sex, its probably more probable that some switch. Equilibrium.

    Chaos. Post modern math. It was heralded by the NYT crowd yet in the halls of the chemistry department, I think everyone felt it was really just cold fusion. As it turned out, Schroedinger had it right the whole way. Granted, Einstein’s theories on EM don’t yet jive, there is a wrench in the gears, and while at one point I thought I could prove Heisenberg was wrong, really its generally all pop psychology and a golden attempt to sell books. Like making electricity from magnets or catalysts that allow hydrogen production to be a net gain, while chaos theory wasn’t really a theory at all, I think its cute that dsf is the one who brings it up as some kind of proof.

  88. vishcosity says:

    Yay! Someone is alive!

    So I googled the modern version of the Mexican, but I honestly have no idea how that relates to topography, unless each step the automon takes is somehow based on GPS elevation. Which would be stupid in a crevass. Unless you’re suggesting that a means to carry contraban into Bisbee without being detected by all those inconvenient satellite heat seekers is going to affect humanity, may I remind you that the BPS shuts down by Fort Huachuca on what appear to be random schedules and the pack horse may not affect the price of things actually that much.

    It is possible that I do not think robomule is intended for what you think robomule is intended to do.

    And further its also possible that Mexican beer should cost way more than it actually does, which may be the same point as above.

  89. jb says:

    I too think shot quality has to be factored in. Doesn’t the stat favor the speedy grinder type? while basically dismissing the so call “sniper”?

    Maybe I’m off base relating this to soccer, but I’ve always played competitively and it seems to apply. Say your going into a game against a superior team.. you already know they’re going to dominate possession, and chances are they’ll generate far more shots than you… But that doesn’t mean they’re going to score a goal on you. Your teams going to do everything possible to keep the shots to the outside and not let them penetrate the high scoring areas. So Corsi is going to tell you that team A’s flank out-possessed and out-shot team B’s… so what? It didn’t tell you the 10 shots team A got were floaters from harmless areas. Maybe team B’s one shot was a breakaway they managed to capitalize on.
    1 high quality chance is far greater than 10 harmless floaters every day of the week.

    Wouldn’t Corsi give a “+1″ to the grinder who dumps the puck in on the goaltender from center ice while changing? Yet a player in on a breakaway who misses the top corner by an inch was irrelevant?

    No doubt out-possessing and out-shooting the opponent is a good thing, but I fail to see how it could ever predict results without shot quality factored in.

    Ricki appears to be on the right track to me.

  90. OilLeak says:

    Discussion on the blog is becoming a wee bit convoluted, I may need to take some higher level mathematics classes just follow day-to-day comments. :P

  91. mike.c33 says:

    vishcosity,

    I think the good folks at Electronic Arts already do all that for us. #joke

  92. mike.c33 says:

    vishcosity: Are you set on corsi, like fur sur, that one?I mean, while we can’t even get TOI for the CHL there isn’t much hope on improvement maybe, but if you’re going to reconstruct games, I think it’d be super stellar if you were to look at each shot and try and find a metric to assess the quality of each one.Or, since I have nothing to do until Tuesday, if there are links to OHL games or whatever is your cross sectional group, I’ll do my ten game bit and try and use whatever metric you develop for any group of games you want.If its just about corsi though, then I don’t know.Until the shot quality is somehow considered, it seems arcane to count them all the same.

    Modulating the shot quality variable by some arbitrary scale will be most likely done best by a mind like Ken Holland.Is maybe what I was aiming to say.Or, if everyone agrees that shot number 17 in game #104 is awarded an 8, then the relative weight of that shot quality variable, only as it regards the initial development of the formula, the weight of that variable is something that very important and connected old white men all agree to say.

    So lets say KH et al say that MPS is a 7.6, then I think its the job of the metric development crew to find a way to have the player stat for MPS work out to 7.6.At least to get it running.

    So for instance, lets say we’re aiming to classify a player with an offensive rating and a defensive rating, each on the scale of 1 – 100.RNH last year may be a 85/43 while Horc gets 34/75.

    Or any other system metric with units.It drives me nuts when there are no units.it has to have units.

    Moving on.

    If the measure is measured by the following considerations:
    A = Goals/60
    B = Assists/60
    C = Shot total x average shot quality (.1 – .9), (or sum(SQ1…SQn), I think that’s the same)
    D = time of pocession
    E = poise quotient
    F = some other metric that tomorrow I will be ashamed I missed

    Then the offensive marker could be composed by:
    40%A + 20%B + 15%C + 10%D + 7%E + 5%F + 3%G
    or whatever percentage combination works to give Magnus a 76.And actually, instead of asking KH to assess each player, we could probably start with yahoo fantasy hockey rankings or something equally publicly traded, just to get the formula working kinda close to what is being seen good.

    Goaltending would be pretty interesting.Imagine finding a system that could well differentiate between a good goalie on a bad team and a bad goalie on a good team.

    Okay.And this is where I’m really going.Landing the plane.

    Each player could have some kind of value number, maybe one each for offense, defense, and poise, and when each player is entered into a second formula, when the totals from 2010 – 11 and prior are entered and the system predicts something relatively close to 2011 – 2012, then we know the model is starting to work.The task that I see is finding a way to modulate the first formula to make the second system’s result better.The tweek factor.Where the art happens.

    If that actually worked, we could plug Getzlaf in for Sammy and Petry and see if in the model, the team then would project to improve in the standings.Does the team actually project better without Eager?

    I guess it could be done now with corsi, or boxcars, but I think we can do it better if we could include even just some kind of measure of shot quality.

    I think the good folks at Electronic Arts already do all that for us. #joke

  93. franksterra says:

    vishcosity,

    Stats-wise I am, as they say, out of my league Donny, but as a geographer I’d say you might be confusing topology and topography.

    In hockeyterms topography might be best applied to the contours of Hall’s forehead post-Potter, or the ice conditions at the end of the 3rd in May.

  94. Woodguy says:

    If you are interested in a very good explanation of shot differential (corsi) being the basis for o-zone possession, please read this:

    http://nhlnumbers.com/2012/5/11/the-theory-and-nature-of-current-advanced-hockey-analysis

  95. Woodguy says:

    vishcosity,

    Read through Arctic Ice Hockey posts.

    Here’s a couple that refer to what you are getting at:

    http://www.arcticicehockey.com/2011/3/30/2075223/the-significance-of-tagging-a-shot-as-a-scoring-chance

    http://www.arcticicehockey.com/2010/5/28/1490918/more-on-goal-scoring-and-true

    Here’s a list of all the shot quality articles that have been on Gabe’s site:

    http://www.arcticicehockey.com/section/shot-quality

    Good reading there.

  96. sliderule says:

    My knowledge of shot quality is limited to Blair Macdonald.

    46 goals playing with Grtezky then tlet go in his prime and never had more than 19

    I think the shot quality provided by the great one was the only reason he ever scored more than 20.

  97. VOR says:

    Woodguy, thanks for the great links, fascinating.

    Delooper, I am just curious about what sort of mathematician you are? I sensed a bit of bias in your comment about Chaos Theory not being a real field in mathematics but rather pop culture. I’d gladly, off this blog, debate that point with you.

  98. DSF says:

    VOR:
    Woodguy, thanks for the great links, fascinating.

    Delooper, I am just curious about what sort of mathematician you are? I sensed a bit of bias in your comment about Chaos Theory not being a real field in mathematics but rather pop culture. I’d gladly, off this blog, debate that point with you.

    Yeah, I was particularly amused by the notion that some goals aren’t counted as scoring chances.

  99. rickithebear says:

    Woodguy: If you are interested in a very good explanation of shot differential (corsi) being the basis for o-zone possession, please read this:

    No shit sherlock!
    we can also establish that
    goals are a reflection of Offensive pocession.
    Blocks are areflection of defensive play.
    No shit sherlock.

    Shots are not a reflection of goal success. 6545G on 73179 shots.

    http://www.kuklaskorner.com/index.php/psh/comments/team_corsi_numbers/

    Goals given and goals got is a reflection of goal success.

    So i will hang my hat on a statistics system that groups paths of attack versus defensive resitence to point of shot taken. then analyze shot path from blade release to goalie and net.

    Pocession paths set up like drive charts in football. Cause the goal of any sport is to identify team and player trend to attack weakness.

    But yes corsi tells us that players are inclined to try and shot from the offensive zone. LOL.

  100. rickithebear says:

    Steve:

    Corsi: The measure of achieving a puck directed towards the net.
    Shot path and variables cause the goals.
    Non causa Pro causa

  101. Lowetide says:

    DSF: Yeah, I was particularly amused by the notion that some goals aren’t counted as scoring chances.

    You’ll find that to be universal imo. sometimes the puck goes in the net without a scoring chance being recorded and rewarding one would be inconsistent with the point of the study.

  102. rickithebear says:

    Lowetide: You’ll find that to be universal imo. sometimes the puck goes in the net without a scoring chance being recorded and rewarding one would be inconsistent with the point of the study.

    Scoring chance is a progression of Corsi to include a SUBJECTIVE review of the shot.
    Saw that shot good.
    A step ahead of Corsi yes.
    Situational analysis no.

    Never understood the Vitrol directed at Staples for the aproach.
    Esspecially when Corsi does not show an ability to predict goals.

  103. delooper says:

    Most of my work is in what’s called Geometric Topology. This is a field quite closely-bound to dynamical systems with a wide overlap — Hubbard (of mandelbrot set fame) writes good books whose topics are purely in geometric topology for instance. I don’t think its controversial (at least among mathematicians) to say that “chaos theory” is more of a buzzword for relating to laymen rather than the name of a proper field of mathematics. I suppose you could find a mathematician that has stong feelings to the contrary but i don’t know that person.

  104. VOR says:

    Delooper,

    I was still working as a physiologist in the eighties when chaos theory reinvigorated the field. In 2009 I was at a conference where the subject was using topological statistics to map brain function and model several medical biomechanics problems. Nearly everybody there was using the term chaos theory in talking about ongoing work in predicting epileptic seizures and heart attacks. I can’t swear any of the speakers or participants was a mathematician. The ones I knew personally were physicists, engineers, biologists, medical researchers, etc. That same year I read a paper by I think it was Charlotte Wrendl tracing the history of chaos theory and arguing for topological mixing as a unifying way of thinking about all the aspects of chaos. She lost me early on but seemed to end up claiming differences in initial conditions are irrelevant. So whether or not chaos theory is a formal school of mathematics or not we biologists, economists, psychologists, neuro-scientists, engineers, computer scientists, and medical researchers are still using the term to refer to some of the inter-disciplinary work we do together.

    When I said that the insights that arose from chaos theory enriched modeling I was thinking of things like models of schizophrenia, futures trading, weather forecasting, cardiac pathology, oncology, biological mathematics, etc. I can draw a pretty straight line from Farmer to the current work in predicting epileptic seizures and clearly from Mandelbrot to modern robo-trading on stock markets.

    I think the difference in our positions can be summed up by your referring to Hubbard as of
    Mandelbrot set fame. To me he is the father of biological mathematics for his work on eye movement patterns in schizophrenia. I can’t make sense of most of what you mathematicians are talking about when you start into manifolds and topological mixing but I get the gist that it should allow us to handle very large and complex data sets. This may revolutionize biology and medicine (and probably a lot of other fields). However, chaos theory, whether it is simply a handy way of lumping a lot of different ideas together or a unique field of inquiry, has already done that.

  105. Spydyr says:

    You know the only stat that matter….wins and loses.The rest don’t mean a hill of beans.Window dressing.Just win baby.

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

© Copyright - Lowetide.ca