Reasonable Expectations for 08-09 (Cogliano)

There are 6 players from the 2005 Entry Draft who have played more games than Andrew Cogliano. If there was a re-draft for ’05 today, my guess is he’d go just outside the top 10 (Crosby, Jack Johnson, Carey Price, Anze Kopitar, Paul Stastny, it was a very good year) which is terrific value based on where they got him.

Cogliano is tied to Sam Gagner because of their impressive rookie seasons but I don’t think their career paths will be similar. Both should have long careers, but if Cogliano is the new Butch Goring (which is my belief) then Sam Gagner would be the new Marcel Dionne (with hopefully more dimension).

MacTavish used the rookies on the same line for much of the season, but there were subtle differences:

  • EV/game: Gagner(12:46), Cogliano(10:49)
  • PP/game: Gagner(2:48), Cogliano(2:01)
  • PK/game: Gagner(0:05), Cogliano(0:48)

Gagner made better use of PP time (3.88 for Samwise to 2.52 for Cogs) but Cogliano had the scoring edge at EVs (2.28 to 1.96). Cogliano also scored 2-1-3 on the PK. The age difference is only 2 years but that represents a tremendous gap and the kid is probably going to improve at a significantly greater rate than #13.

Cogliano played mid-level competition (tougher than you’d expect a rookie to see) with average to below average linemates and the result was a solid +1. That should give us a clear indication of his Butch Goring future, a quality two-way player who uses his speed effectively in all zones.

However, a deeper look implies he was probably more lucky than good this past season. His Corsi number (-11.7) ranked as the third worst in the division (Trevor Linden of the Canucks was -14.6 and Aaron Voros in Minnesota was -15.6) among forwards who played 50 or more games. I’m having a hard time marrying the two results and welcome any input.

I don’t have Corsi splits (does anyone?) but we did notice a massive improvement for Gagner in the season’s second half and I thought it may also have applied to Cogliano:

  • First 41: 7-12-19 (.463) -6
  • Second 41: 11-15-26 (.634) +7

That’s terrific improvement. If MacT had him out there against tougher than expected opposition one year ago it’s unlikely he’s going to turn back now. That added to those PP minutes being taken away (I’d guess) means his numbers aren’t likely to take a huge step forward this year.

Prediction for 08-09: 80gp, 20-25-45 (.563 per game).

written by

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

90 Responses to "Reasonable Expectations for 08-09 (Cogliano)"

  1. Scott says:

    Wow, that’s the first time that I ever came to this site, hoped for an update within the ten seconds of disappointment at not finding a new blog entry, hit F5 and it actually worked.

    I wonder what other powers my mind has now…

  2. Scott says:

    As for the article, hard to disagree with anything. It was pretty shocking to see such a terrible Corsi number for Linden, though.

    I figure Cogs get more PK time in lieu of some lost PP time, but his stats remain roughly the same (in line with your predictions), which would technically be a slight improvement.

    The biggest improvements I hope for with the kid line is defensive coverage. With MacT as coach, I think it’s a reasonable assumption that it will happen.

  3. Julian says:

    I get the feeling a lot of people are going to be talking about a sophomore slump for Cogliano this season. Didn’t he have something like an 18%SP last year? He’ll probably get more ice time this season, but not enough to make up for a drop in SP%.

  4. Black Dog says:

    This kid is a good player. I think we may see him get a shot on the Pisani line if Brodziak or Pouliot don’t work out but I guess to start it will be a similar role to the end of last year.

    Nice pick by the Oilers. Guy had a nice rookie year for sure.

  5. uni says:

    People keep talking about Cogliano’s shooting percentage and how there can be no sustain there, and this will lead to the inevitable drop-off and slump.

    I think wiser posters than me have pointed out that Cogs gets to high percentage areas to begin with, and that 98 shots over a season isn’t that much and his volume of shots is likely to increase as will his ice time. So even if he falls back to a mean of say 12%, if he fires the puck 50 more times he’s still hitting 18 goals.

    He may very well hit the 2nd year slumps, but it’s not as foregone a conclusion as some people seem to think.

  6. Lowetide says:

    I think it's less likely he'll have a slump as people will perceive it that way. I have him running in place season over season (45 points) and also have Gagner's point total increasing only slightly. Nilsson gets a bump from that line because there's no way his PP total doesn't improve (or at least the odds are extreme it improves).

    I have the 1line doing strong business (I've projected them to score 82 goals versus 64 one year ago) and the 3line will have one "star" offensively (Cole).

    The Oilers aren't going to score 300 goals this season. My bet is that the 1line kicks out the jams this season, grabbing most of the PP time (Horcoff>Gagner until at least Christmas) and that Cogliano misses the net enough to have this next season resemble his first one.

    That's not a bad thing btw. Butch Goring spent much of his second season in the minors.

  7. Jonathan says:

    Aside from his unreal shooting percentage, Cogliano also tied for the second highest 2nd assist/60 number on the team- i.e. his assist total may not reflect his playmaking (which, IMO is good but not great).

    16G – 24A – 40PTS seems pretty reasonable to me.

  8. PerformanceOil says:

    I’m having a hard time marrying the two results and welcome any input.

    You don’t have to marry the two results. Corsi is a poor predictor of performance. In order to truly see if he was lucky, you would have to watch some game film and see if:

    a) The shots against were quality that the goaltender saved when he probably shouldn’t have.

    b) Cogliano’s goals were off shots of the ‘floater’ variety that the opposing goalie should have had.

    I don’t have any recollection of the the type of shot against Cogliano was giving up, but I don’t think many of his goals were weak.

    Personally, I think Cogliano used an elite asset (speed), to break down the opposing coverage and then made the most of the quality chances that were subsequently generated. Hard to say if that will continue (the break-downs and/or the high conversion rate), but I think Cogliano will still generate his fair share of scoring opportunities.

    I think he’s going to be an amazing player. If he plays a similar role to last year, I’d expect better numbers. If he posts the numbers you have predicted, I think it will be in an elevated role (tougher minutes, more PK time).

    I think he was more good than lucky.

  9. Alice says:

    Uni: Thanks for throwing some water on the SH% stuff – that stat gets trotted out more than it deserves. The other guy with the unreal SH% was our cowboy friend Glencross. There seems to be wide agreement that Brodz has some knack for creating chances. Any thought that unreal/unsustainable SH% has plenty to do with what your linemates are doing to set you up? I guess that’s unsustainable if they stick you with some pylon, but why would line chemistry that works have to stop working? Just asking.

  10. Vic Ferrari says:

    Lowetide

    Behindthenet.ca has had a rough time with Corsi numbers for some reason. For a most of the past twelve months they were semi-random, with blocked shots numbers reversed in the script. Now they look to be Fenwick numbers.

    You can tell because Pronger should be in the next clump of ANA players behind the Perry/Getzlaf line (granted a mile behind). If he drops right to the bottom then the blocked shots thing is flipped, if he’s middling then it’s Fenwick. Which makes sense I think, guy never blocked many shots, but did drive possession quite a bit.

    Anyhow, if you want splits on EV stats, use:
    http://www.timeonice.com/playershots.php?team=EDM&first=20800&last=21230

    That gives the individual player stats as well.

    So that link above would be for the Oilers from NHL game 800 (about the time Horcoff got hurt) to the end of the season (game 1230).

    And, by way of example http://www.timeonice.com/playershots.php?team=DET&first=20305&last=20615 would give you the stats for DET in the second quarter of the season.

    You can use exactly the same format for teamfaceoffs.php, to see who was getting the offensive zone draws and vice versa.

    These scripts just use the player numbers,not their names. And there are no icetimes shown either. Still, I hope that helps.

  11. Vic Ferrari says:

    On Cogliano’s shooting percentage, I have the overall shooting percentages for the players dating back to 1987/88 in a spreadsheet someone on OilFans once sent me.

    They don’t give a breakdown of EV, SH, PP … but they do show the empty net goals. And of the 4200ish player seasons of 70+ games played since then, I think that Cog’s 07/08 is probably top 5 in the last 20 years.

    I mean if he could really finish like that, surely to God he would have kicked the ever living crap out of Kariya’s college numbers.

    Still, good player, and he’s got so much room to improve in other areas (defensive zone coverage, as Scott says above, and the play needs to die with him less). I think it’s a given that the he’ll improve a bunch in terms of creating more scoring chances going forward, and cutting down on the number of chances that the bad guys get when he’s out there. Obviously those two things go hand in hand. So that should mask the fact that a bigger chunk of his shots next year will miss or hit the goalie.

    BTW: If you were a GM in that era, and just made the incredibly simple decision to acquire players after bad shooting% years, and trade away players after great shooting% years … there would be a statue of you in front of an NHL arena somewhere.

  12. Vic Ferrari says:

    Another bit on shooting% (sorry to be going off topic LT, it’s just fascinating to go through the list though, and I don’t want to even think about how many of these guys cashed in on these).

    But some Oiler fun ones:

    Buchy had a 20 goal/44 point year in 91/92 (Damn! Good for Buchy, I was living abroad that year, so I can’t comment on his game back then, but he was up there amongst the hall-of-famers in shooting percentage that year).

    Same with Selivanov in his 27 goal year for the Oilers (far from a complete player I know, I’m not making comps to Cogs here). The next year in CBJ he fell right back to the 8% he’d averaged before.

    And since the time is right: In Gretzky’s first year in L.A the entire group of top six forwards shot 16.8% or better. Damn! It wouldn’t even surprise me if the Kings were outchanced when Gretzky was on the ice either, when the opposing goalies put up sub-sub-sub-Racicot save%’s when you’re out there, it just doesn’t matter. How would you even coach a guy like that?

  13. Dennis says:

    It’s tempting to say that the Cogs/Marchant comparisons seem apt because Todd once had a pretty good half season of offense that lead people to believe he had a little bit of a scorer in him. Now, I don’t know how that looked compared to league averages and I’m not saying that Cogs won’t ever hit 25 goals in a season but from limited Cogliano viewing – WJHC — he looked positively Todd-esque, ie he has the speed to get himself in the high pct scoring areas but doesn’t have the hands to take advantage.

    Now, what difference would it have made if Marchant had gotten to play with guys like 12 and 89?

    I’m not sure where he’ll land counting stats wise but given that he’s speedy, hardworking and intelligent, he looks like a prime candidate to be an outscorer and you can never overvalue these guys.

    On that note, it’s gonna be interesting to see just how much he’s gonna be making come his next contract. I don’t think there’s any way to “keep down” 89′s counting stats so you can be sure he’s gonna get his before the 2011 season begins. But with 10 in the fold and 89 as well, if 13 starts playing in one of the PVP roles, then his counting stats won’t be all that impressive and I think we’ve yet to see a ~45 point outscorer get a big payday.

  14. Vic Ferrari says:

    Yeah, Dennis, that’s a really valid point I think. Now that first time RFAs are getting huge dollars, commensurate with their counting numbers, can you really afford to let young guys eat soft icetime and develop their offensive games? Even if it hurts them and the team a bit in the short run, maybe the Oilers braintrust has to accelerate him into that role a bit sooner, for the sake of the 10/11 roster and beyond.

    As you say, they’re going to have to pay Gagner no matter what, besides injury. No stopping him I think.

  15. Vic Ferrari says:

    On the Marchant comp … I don’t have the numbers in front of me, but I think that his first part-year came with a wicked shooting%, giving us all hope that he would be an NHL scorer. Then poor after that until the Sakic-in-a-very-good-year 17% he put up before Lowe cut him loose.

    The reason I mention it though, is like Selivanov, Doug MacLean acquired him for the CBJ roster. Doug just couldn’t stay away from these guys.

  16. Traktor says:

    “I mean if he could really finish like that, surely to God he would have kicked the ever living crap out of Kariya’s college numbers.”

    Has anyone even came close to matching Kariya’s 1992-93 season? Off the top of my head I can only think of Peter Sejna’s 80+ points for Colorado.

  17. Scott says:

    Vic:

    So if you go out acquiring guys after high shooting percentage years and trading away guys after low shooting percentage years, you get a job in broadcasting?

  18. Traktor says:

    “from limited Cogliano viewing – WJHC — he looked positively Todd-esque”

    I never agreed with that perception. He missed a couple nice scoring chances but still finished with something like 6 points in 7 games. He also had to play with that idiot Downie who instead of giving the puck to Cogliano with speed would try and go through 3 zones by himself. If he doesnt’ score on any of his 3 chances people can him Marchant but if he scores on all 3 it isn’t sustainable and he is a fraud…

  19. Vic Ferrari says:

    Vic:

    So if you go out acquiring guys after high shooting percentage years and trading away guys after low shooting percentage years, you get a job in broadcasting?

    I guess so. Doesn’t end badly either way. I suppose the trick is getting the GM gig in the first place, it’s all good after that.

    BTW: That’s just a terrific blog you were doing for a while there. I never noticed it before, not until I clicked your name from a comment on IOF the other day. A nice mix of old and new testament biblical references and a terrific post on Tambellini that has swayed me more than a bit. But the St. Pierre analysis you promised never came, which is a shame, because I think he’s an interesting player, and possibly the unluckiest guy around.

    Hopefully the rest of the Oilogosphere shames, bullies or charms you into blogging again when hockey starts.

  20. Scott says:

    Vic:

    Thanks for all the credit, but that’s not me. I’ve noticed another ‘scott’ posting over at jonathan’s blog, so maybe you saw him.

    I’ll gladly take some more credit for it, though ;-)

  21. Vic Ferrari says:

    If he doesnt’ score on any of his 3 chances people can him Marchant but if he scores on all 3 it isn’t sustainable and he is a fraud…

    Who called him a fraud? Am I missing something?

    Probably something you read on HFboards. I took a brief wade through that cesspool the other day, and I can see why the hoards of slobbering orcs would grate your nerves.

    Strangely, I would have thought that a one-off season driven by shooting percentage is exactly the kind of thing that would get the HF clowns hard. Apparently not.

  22. Alice says:

    Let’s remember that 17 – 12 = 5%, so while it looks like a +40% swing the difference only amounts to one more goal per twenty shots on net. And he took, what, 98 shots all year?
    So yes, the shooting% puts some color to the boxcar numbers, which is helpful as long as we don’t forget which ones count and which ones are the context. It’s kind of a trailing indicator, and I just hate to see people make categorical statements about what can or can’t be sustained. Cogs is going to have a nice season, and hopefully we’ll see it show up in his +/- column. That’d make a few of us pretty happy.

  23. Sean says:

    In Marchant’s first 2 seasons (1994-1996), the Oilers were to shambles. They were a horrible team, with poor management and coaching. Cogliano’s draft position indicates a higher skill level IMO and he will be playing with players like Hemsky, Horcoff, Gagner and Nilsson who are all high skill level. The Marchant comparison isn’t exactly a knock to Colgiano. If Cogliano plays 1000 NHL games and wins a cup I’m sure he’d be happy. I just think there are enough factors that indicate to me that (barring injury) Cogliano should develop into a better player.

  24. mc79hockey says:

    Let’s remember that 17 – 12 = 5%, so while it looks like a +40% swing the difference only amounts to one more goal per twenty shots on net. And he took, what, 98 shots all year?

    Right…but that’s something like 5 goals disappearing. That’s a rather large chunk of value for him. Without it, I’d guess his offence is close to replacement level (ie. a $450K NHL forward who gets ice time is going to score at least some goals – a player’s value is what he gives you above that).

  25. Vic Ferrari says:

    Alice:

    There are a lot of ways to look at that. I was just talking EV here, which is more or less the same difference, because the guy didn’t get a whack of PP time and I don’t remember any empty netters.

    So he was 19.8% at evens. 16 goals on 81 shots.

    As you say, there aren’t that many shots, hell there just weren’t that many shots by Oiler players when he was on the ice last year (that shortfall was accounted for by the extra shots the opposition got, btw. Which makes basic sense, puck has to be somewhere).

    So if you’re suggesting he’s an EVshooting% 12% kind of player, I can buy that. What if the cookie had crumbled the other way, or does next year, and he falls 5% short. They just don’t go in for him. No softies, no lucky breaks with rebounds, too many five bell saves from opposing goalies on him. So that takes him down to 7%, and all the smart money in the world is going to say that he’ll have at least one year that low at some point in his career.

    That would’ve translated to 4 EV goals this year. Would you e defending him, saying that “stuff happens, they’ll go in for him eventually”? I would. And as I say, it’s going to happen eventually. And when it does, Kelly Hrudey will reach into his bag of hockey magic and pull out a dozen reasons that seem plausible at first blush.

    I guess it will be me and you that think he was probably just a bit unlucky. :)

  26. toqueboy says:

    re: cogs SH %:

    it seems to me that mgt is indicating that they will keep the kid line together. i believe that everyone here has accepted that gagne and mini magic will be the ‘handsy’ players on that line…ie. account for the assist, so i think we’d have to assume that cogs will be doing more shooting. if he’s not, then i believe he’ll be replaced on that line with someone who does.

    that being said, with gagne’s ability to pass from below the goal line and nilsson’s ability to stutter step the game’s tempo, i can’t see any reason why cogs shouldn’t see a big swing in the number of shots taken, and frankly, from continued high quality zones.

    more than anything i hope that cogs has been watching video of hull finding the seam and rifling a seeing eye one-timer.

    i think that LT’s comments on why the oiler’s have interest in potulny (re: shots) is bang on and hopefully there will be a managerial/coaching focus on delivering some rubber this season.

  27. dave says:

    Well I do think Cogs is most on the bubble to hit the minors.

    In order for him to stay up he needs to continue to put up goals with the second line.

    I think we currently have or will trade for centers that can fill a better checking role than him.

    That said, these kids (sans Poo) say the right the things, with an 80′s swagger so as a fan I will make a nature / nuture arguement, that his surroundings will support him as a shooter.

  28. Scott says:

    I don’t think Cogs is on the bubble to drop to the minors at all. Even when he wasn’t producing last year, MacT was still singing his praises and talking about ‘finding ice time for Cogs’. He has an elite skill that translates to all facets of the game.

    That said, I think he has to (and will) take more shots this year. As was pointed out, the kid line was badly out shot last year. Increased defensive awareness and ability should lead to more puck possession and more shots for the kids. That should make up for the inevitable dip in shooting percentage, or at least close.

  29. Alice says:

    As a rookie he might have been a bit shy about just unloading the thing too – so he probably swung his own number North just by being careful with his shot selection, trying to look as wily and composed as any veteran 20yr old :-)
    With increased confidence he’ll probably pull the trigger more often this year, and his shooting % will normalize.

  30. Jonathan says:

    Compraing Cogliano to other rookies with high shooting percentage post-lockout, there isn’t any doubt that he’s due for a drop. On average, these kinds of players lose over a third of their shooting percentage, and score 28% fewer goals/game.

    With that said, I doubt very much that Cogliano is a comparable offensive player to Todd Marchant- Marchant’s career high in shooting percentage was 13.7%, 5% higher than his average. I don’t see Cogliano’s career average being much below the 11-12% range.

  31. Kris says:

    The only way to know what Cogliano’s SH% means about him as a player and his future as an offensive contributor is to count how many of his goals were due to breakaways.

    Why? Well, speedy players -Briere for example- tend to have their career SH% inflated by breakaways. (Of course its common sense that you score less often if there’s a defenseman between you and the goalie.)

    Moreover, it’s possible that Cogliano displayed a rookie tendency to not shoot and thus held on to the puck when he had low percentage chance to score; i.e. he routinely didn’t take the Jason Blake shot from the top of the circle. If so, Cogliano actually should have scored MORE goals AND had a lower SH%. That is Cog’s high SH% might suggest he will score more in the future.

    So, until somebody does a more detailed analysis of Cog.s shot selection, his oddly high SH% is a very unreliable indicator of his offensive skills or lack thereof.

    In other words, it seems the thought most people have is that his high SH % means he got lucky last season, and this could be true. However, nothing is shown by his SH% alone, since there are a variety of plausible ways of explaining his SH%.

    On the other hand, I do remember Cogliano getting at least 3, lucky, Johnny-on-the-spot, open mouth goal crease tap-ins.

  32. Vic Ferrari says:

    The kid line is the one to watch for sure.

    Gagner is the guy who is going to make that line go, I think. My guess is that the other two will have a good year as long as #89 stays healthy and continues to improve.

    Nilsson is a guy who put up decent numbers, and decent underlying numbers as well … but I just can’t get on board, by eye he just seems to make so many stupid mistakes, especially at the blue lines. Probably just my bias, or ‘seen him bad’, or both.

  33. Traktor says:

    “Who called him a fraud? Am I missing something?”

    Nothing that was specifically said here, just a bunch of stuff over time.

    “Well I do think Cogs is most on the bubble to hit the minors.

    In order for him to stay up he needs to continue to put up goals with the second line.”

    Speed = transition.
    Transition = offense.
    Cogliano = speed.

    Most of the time he is on the ice means his linemates get to gain the blueline for free.

  34. Jonathan says:

    In other words, it seems the thought most people have is that his high SH % means he got lucky last season, and this could be true. However, nothing is shown by his SH% alone, since there are a variety of plausible ways of explaining his SH%.

    That may be true, however, looking at the eleven comparable players in terms of rookie shooting percentage, post-lockout, 10 of the 11 had their shooting percentage drop the season after. The one player who’s shooting percentage increased (Paul Stastny) had a lower shooting percentage than Cogliano both as a rookie and a sophomore. The odds are against Cogliano.

  35. Vic Ferrari says:

    kris:

    The thing is, if you took all the forwards in the league, gave them a set of dice weighted to roll 6, weighted by the EVshooting% average of NHL forwards, and gave them as many throws as shots on net … well some guys are going to have good stretches with the dice, others not so much. And the right numbers of guys will overachieve and underachieve by the expected amounts. That’s just dice rolling.

    Compare that to what happens with shooting% itself, and it’s near as dammit the same. A bit wider spread with the shooters, as you’d expect, it’s not JUST luck, just mostly.

    Now if you weighted those differently for each player, weighted them to the EVshooting% from their own previous four seasons, then those lines are going to be almost on top of one another, the spread on the shooters just ever so slightly outside (I haven’t done this, but you can bet the moon on it).

    That’s why you can’t swing a cat on the internet without hitting someone rationalizing why shtuff happened that revolves around EVshooting%, EVshooting%-while-on-ice, EVsave%-while-on-ice etc.
    But I’ve yet to read anyone on the web who can predict these things any better than they can predict the roll of the dice.

  36. Kris says:

    I just watched all 18 of Cogliano’s goals.

    Most of them were from right in front of the net.

    There are two ways to look at his highlight reel. Either:

    1. Cogliano used his speed to get behind defenders and score high percentage goals.

    Or 2: Cogliano got a lot of lucky bounces around the net and had a good passer to get him the puck when he was in position.

    If its two, I can see Cog scoring 10-12 goals next year.

    If it’s one, I see him scoring at a similar pace.

  37. Sean says:

    Vic:

    BTW: If you were a GM in that era, and just made the incredibly simple decision to acquire players after bad shooting% years, and trade away players after great shooting% years … there would be a statue of you in front of an NHL arena somewhere.

    So in light of Jonathan’s post. The Oilers should keep Stoll and Torres and trade Horcoff? ;)

    I think Cogliano’s shooting percentage is reflective of the play making ability of Gagner and Nilsson. On another line his SH% would likely go down and it would be up to him to create more. For some reason, I think MacT is going to force this on him.

  38. Oilman says:

    A few guys have built fairly solid careers on a low shot/high percentage game. Mark Parish and Andrew Brunette are two current players who tend to score between 20 and 30 goals per year and never take more than 120 or so shots – it wouldn’t be surprising to me to see Cogliano produce similar results….but the smart money is likely on a drop in shooting percentage and about 15 to 18 goals on 120 shots.

  39. Vic Ferrari says:

    sean:

    Thanks for the link to jonathon's blog, I wasn't reading much in April, and not reading C&B at all back then. An excellent blog though.

    In short: History tells us that Torres will bounce back, I think it was a poor trade. And I think that Stoll should have been traded last summer, not this one.

    Though perhaps other reasons came into play with these guys, bad boy millionaires aren't finding traction on a Katz-owned roster it seems.

    On Horcoff: I've always rated him, he brings a lot more to the table than counting numbers. So if you can sign him long term, you do just that IMO. In a way, it's unfortunate for the Oilers that he had a big goals/points year this past one, he might have been a bit cheaper to sign otherwise.

    And of course the total Shooting% we're looking at there include PP time and empty netters as well.

  40. Vic Ferrari says:

    Oilman:
    I hadn’t thought of Brunette, I’ll check into him on day. Parrish for sure though is a PP guy, those in-the-crease PP forwards should have a decent shooting percentage. He doesn’t bring much else to the table though.

    Brunette is a good player though, always criticized for being slow, but he’s as hard on the puck as anyone, and he can pass and shoot. That was a good pickup by Minny at a good price, assuming he still has some gas in the tank.

  41. Jonathan says:

    Mark Parish and Andrew Brunette are two current players who tend to score between 20 and 30 goals per year and never take more than 120 or so shots – it wouldn’t be surprising to me to see Cogliano produce similar results….

    Although such players tend to have two things in common: a) bad wheels and b) tons of powerplay icetime.

  42. PerformanceOil says:

    Vic,

    Behindthenet.ca has had a rough time with Corsi numbers for some reason. For a most of the past twelve months they were semi-random, with blocked shots numbers reversed in the script. Now they look to be Fenwick numbers.

    There is something wrong with the Corsi calculations given on BtN – for some players – I agree with you there. I think it might have to do with rounding in some cases, but in others it is less certain (it is certainly not a Fenwick number being calculated). However, calculating the Corsi numbers manually still gives the same conclusion: No strong correlation between Corsi and EV +/- (and Pronger’s Corsi is actually worse than what is given, wherever you ‘feel’ he should fall relative to the rest of the Ducks).

    This assumes the numbers given for SF/SA etc. are correct on BtN. Doing some quick checks, they seem to be.

    Interestingly though, your own site’s numbers appear to be flawed.

    Specifically, the numbers appear to reflect all situations, rather then only EV, at least in some cases.

    Two examples:
    For Cogliano, game# 20038 (vs. Det), your site has Cogliano as: Shots for = 5 and Shots against = 8. Conversely, BtN has it as 4 and 7.

    Going through the game log, I scored it as 4 and 7 (in agreement with BtN), however Cogs was on for 1 PP shot for and another PP shot against, giving your totals of 5 and 8.

    The second example involves game#20073 (vs. Van). In this case, your site has Cogs as being on for a GF, however the only time Cogliano was on for an Oiler’s goal was during a Vancouver powerplay.

    So again, using the best available data, Corsi does not predict even strength success at the level of the player. I think coaching in general probably has a much larger affect on Corsi, then a given player’s offensive/defensive abilities. No need to rationalize a good player’s poor Corsi number, until someone proves otherwise.

  43. Vic Ferrari says:

    Just to reintroduce everyone to the elephant in the room.

    Cogliano was on the ice for 875 total shots on goal, by either team, at even strength. That’s near enough average for the league, and the Oilers have never been either a shot-happy team of a one-extra-pass team during MacTavish’s tenure. So it’s exactly where he should be.

    The problem is that 501 of those shots were taken by the other team. That’s the number that has to change. And when it drops the 375 oiler shots will go up naturally.

    What what happens with the shooting percentage will happen, no point worrying about it. If he spends less and less time in his own end of the rink as the season wears on, the rest will take care of itself eventually.

  44. Traktor says:

    jonathan: I just spent an hour replying to your post on O.N but it won’t let me reply and I lost all my efforts. errrr…

    In short, Clowe and Eaves aren’t credible comps.

    1 played 15 games the other played less than half the season and was injured when he did play.

    Using your comps, minus Clowe and Eaves, they averaged a 39 shot increase from 1st to 2nd season.

    Using your comp logic, Cogliano should shoot the puck 137 times next year. 25+ if he keeps the same %. He can drop a full 5% and still hit 18 goals.

    That said, none of Stastny, Staal, Radulov or Svatos are good comps as they shoot the puck 3/4 more than Cogliano.

    Oh and none of the players you listed shot the puck fewer times per game than Cogliano, in either of their 1st or 2nd season, which was the crux of the #2 point of view. He doesn’t shoot much but when he does it’s quality.

  45. PerformanceOil says:

    He doesn’t shoot much but when he does it’s quality.

    Based on last season, this seems to be true.

    The questions are: can he continue to generate quality chances, and can he continue to cash in on those chances?

    It will depend on his role, and just how good of a player Cogliano really is. Again, I think he was good last year, and I think he will be better next.

  46. Jonathan says:

    jonathan: I just spent an hour replying to your post on O.N but it won’t let me reply and I lost all my efforts. errrr…

    In short, Clowe and Eaves aren’t credible comps.

    1 played 15 games the other played less than half the season and was injured when he did play.

    Using your comps, minus Clowe and Eaves, they averaged a 39 shot increase from 1st to 2nd season.

    Using your comp logic, Cogliano should shoot the puck 137 times next year. 25+ if he keeps the same %. He can drop a full 5% and still hit 18 goals.

    That said, none of Stastny, Staal, Radulov or Svatos are good comps as they shoot the puck 3/4 more than Cogliano.

    Oh and none of the players you listed shot the puck fewer times per game than Cogliano, in either of their 1st or 2nd season, which was the crux of the #2 point of view. He doesn’t shoot much but when he does it’s quality.

    It’s true that the vast majority of the “comparable” players play different styles- there weren’t enough samples post-lockout to get a good look at Cogliano-type players, so I used what I had and gave a “quick and dirty” look at traditional progression, with the implied assumption being that the average progression of players with similar sh% would be comparable, despite different styles. It’s an argument with weaknesses, but the fact is there isn’t much to work with.

    With that out of the way, I look at Cogliano and see a player with virtually no powerplay time and can’t picture him scoring at the same rate. I just can’t.

    He won’t fall back far, because I think he’ll take more shots, but I am convinced there will be a dip. You’re evidently convinced otherwise, so I’d suggest we wait and see at this point. If I’m wrong, I’m more than ready to take my lumps for it (and it wouldn’t bother me a bit, given that it’ll mean better results for the Oilers). I’m just trying to be realistic, despite my own biases, and I think the majority of Oilers fans are showing their colours when projecting Cogliano.

  47. Vic Ferrari says:

    PO:

    I went through the play-by-play for 20038 and got 5 and 8 saved shots, and 1 and 0 goals, just as the script at timeonice churns out. That’s what that script uses, the play-by-play sheet. And all EV icetime (4v4 and 5v5) except empty net stuff, because that can throw things way out of whack.

    And it depends what you mean by predicting, if you’re looking at a spreadsheet and trying to predict what the points total is based on a player’s shots+/- or fenwick or corsi … well that’s not going to work well at all, or at least intuitively I wouldn’t hink so.

    If you’re looking at predicting performance (ability to outscore opposition) in the second half of the season at even strength, based on the first half of the season … then Fenwick works well. Or Shots+ rate for PP.

    On a team basis it’s even stronger. I mean if you use EV+/- from the first half to predict the second half, that’s reasonable. Of course you’re not accounting for trades and injuries, or maybe a goalie who was ridiculously hot during the first half … still, beats the hell out of alternatives. But if you use EV faceoff zones as a predictor instead (further away from what we care about, EV+/-, but a bigger sample size, and possession matters a tonne so there is reason in it).

    So if a team was EV+/- +0 in the first 40 games of a season, but had 100 more offensive zone draws than defensive ones, and it’s a year with 10 end-zone faceoffs for every goal at evens … then you’d predict +10 by faceoffs, and a repeat of the +0 by EV+/-. Makes sense, no?

    And it’s fair because of all the things that can happen (say Iginla returns from injury) it will affect both numbers for the team. i.e you’d expect them to have a better EV+/- and a better faceoff +/-.

    Anyhow, I did that for the last four seasons and was 18W-12L twice and 16W-14L twice. Faceoffs trumped EV+/- as a predictor of future EV+/- by a narrow margin. Though both are wholly sensible IMO.

    You could do the same with shots+/- as a predictor, you’d get more or less the same result I’m sure, maybe a touch stronger.

    There is no magic bullet here, and I’m not advocating that anyone start wagering on this principle. Because by Christmas Vancouver will probably have a positive EV+/- and a negative shots+/- and faceoff+/- … but they have Luongo, so they’ll have a decent shot at sustaining that.

    The startling revelation: puck possession in the right end of the rink matters, a lot.

  48. Traktor says:

    You’re evidently convinced otherwise, so I’d suggest we wait and see at this point.”

    The only thing I’m convinced on is that people are putting too much stock into shooting %.

    If the arguement is that Cogliano won’t get PP time and will have to play on the 3rd line with Moreau and Pisani and thus his numbers will drop, then I agree. I don’t don’t agree with the shooting % spin. As I’ve said in the past, the ability to generate offense is much more important than conversion rate. If he plays on the 2nd line and spends half the season on the 2nd PP unit I see no reason why he shouldn’t crack 20.

  49. PerformanceOil says:

    Denis,,

    I think it is safe to say that Cogliano’s SH%’age will fall. However, I’m not sure the comparison to Marchant is fair. Marchant’s best season was 13.7% (twice) and his average is 8.7%. So, his high is 5% above his mean.

    Even if 18.4% represents Cogliano’s career high SH%’age (and I don’t think anyone would predict otherwise), if he averages about 5% less for his career, he’d be right around Marchant’s highest.

    While there are probably many mitigating factors (era for one), and there is no way to predict just how far Cogliano’s SH%’age might fall, it can go a long long way before he is in Todd Marchant territory, at least as far as goal scoring.

    For reference, if Todd Marchant put up a career SH% of 13.4, he’d have averaged 27 goals a season. Given that 15 goal Marchant was a heck of a player, what would 27 goal Marchant be worth?

    Of course, this assumes that he can be as defensively reliable as Marchant, and that he can raise his shot total with out SH%’age tumbling more than that 5%.

    While that is probably best case, he still has a long ways beofre he proves the likely worst case (Marchant), and that is a pretty damn fine consolation prize anyway.

  50. Vic Ferrari says:

    Traktor:

    The Oilers took 375 shots when Cogliano was on the ice at even strength (according to the NHL.com play-by-play sheets,and not inclding empty net situations).

    Cogs took 81 of those 375 shots, or 21.6% of them. That’s low for sure, we’d all like him to shoot a bit more. But he’s not a net crasher like Gary Roberts or a pure score-from-anywhere shooter like Shanahan, so you can only grow that share so much. And just shooting for the helluvit won’t help the team.

    There’s only so much room to grow the shots without growing the possession first.

  51. Asiaoil says:

    As Vic points out – the bigger issue with Cogs is shots against this coming season. That’s the area that must improve since preventing goals against is just as important as putting them in the other team’s net. Sure he probably slips some in shooting percentage – he wasn’t an elite scorer in college or at WJC – but stranger things have happened than having two unusually good years in a row. He looks to have a bit better hands than Marchant – and if he can eventually play defense as well – he’ll be both valuable and under-paid relative to his overall contribution to team success. He’s a nice player who I’ve liked from the beginning because of his speed and smarts – but fixing his defensive issues will likely mean a step back in terms of offensive production. That’s a trade-off we should both expect and be pleased about longterm.

  52. Jonathan says:

    If the arguement is that Cogliano won’t get PP time and will have to play on the 3rd line with Moreau and Pisani and thus his numbers will drop, then I agree.

    But how could that be the argument when Cogliano didn’t get powerplay time this past season and few are projecting him to play tough minutes?

    I don’t think he’s the goal-scorer (yet) that he looked like last year; I think there was some luck and some situational stuff happening that won’t be repeated, and I think shooting percentage is a nice indicator of how lucky he was.

  53. PerformanceOil says:

    Vic,

    Yes, I know you script it off the logs. I counted game number 20038 again, and I still get 4 and 7.

    Not that it matters, since game 20073 is much easier to verify. The Oilers only scored 1 goal, and it was during a Vancouver PP. To be fair to you, the log does have it as an EV goal (though it is clearly SH, since two minutes hadn’t expired since Pitkenan’s penalty). However, BtN did not count it, so clearly their method is more accurate, at least in this case. Overall, there are probably many errors in both data sets (nature of the beast), however I don’t agree that only your data can be used to generate valid conclusions, or that there is evidence to suggest BtN’s data is flawed to a greater extent.

    As for the rest, Corsi does not predict EV+/-, period.

    As far as possession, of course possession matters. Trouble is, Corsi does not measure possession. In fact, it measures loss of possession much more directly. Obviously trading possession for a (good) chance to score is a great swap. Losing it for a low %’age chance on net isn’t.

  54. PerformanceOil says:

    Asia:

    the bigger issue with Cogs is shots against this coming season. That’s the area that must improve since preventing goals against is just as important as putting them in the other team’s net.

    Those are two separate things, and one is not a good predictor (by itself) of the other.

  55. mc79hockey says:

    My numbers on Cogs are 379 and 508, for what it’s worth. Might have the EN there, can’t remember. One of the guys on Yahoo threw up the PBP files for this year in a text file; I’m eagerly moving them to Excel so that I can put together the type of numbers I had for the past couple of years plus some new stuff.

    Point of interest: He got much better in the second half. 156/243 in the first half and 223/265 in the second. So much better is still bad but we’re moving forward.

  56. Vic Ferrari says:

    Asiaoil:

    If the mood ever strikes you, and if time permits, here are some links to help you to prove (or disprove) your point.

    http://www.timeonice.com/1EDM.HTML
    http://www.timeonice.com/2EDM.HTML

    That gives you the underlying numbers for the Oilers in the first half of the NHL season (games 20001 to 20615) and the second half of the season respectively.

    Hopefully somewhere gives the half-season splits for EV icetime. Failing that I’ll write something.

    If you’re right, and I’m sure that you are, then guys who’ve seen a significant change in EV-shots-against-while-they-were-on-the-ice from one half of the year to the next … they will, by and large, have seen a similar change, in the other direction, of their shots-for-while-on-ice number.

    As an aside, Jebus, I’m agreeing far too much with goalies lately, from Oilman and Bruce to dawgbone and yourself. That can’t be good.

  57. Vic Ferrari says:

    PO said:

    As for the rest, Corsi does not predict EV+/-, period.

    Yes it does, and I’ll boldly predict that when the NHL eventually publishes which team took the offside … that will predict EV+/- well also. Not because offsides are good (they obviously aren’t) but because they tell us who had the puck going forward at a moment in time.

    I think I’ve posted on this before, no? In any case, at some point I’ll hit this from a few different sides all at once.

    Shots +/- is near enough the same thing with big samples, and we’ve got years of that data on the NHL.com PBP sheets, should be easy enough.

  58. Asiaoil says:

    Vic – well it’s just nice to hang around here with people who understand that Cogs should probably build his game around ES outscoring and strong 2 way play. It’s good for the team longterm – maybe not so much for his bank account – but that dynamic is tough to grasp without inside info. All of the HF blubbering about how many points he may or may not get is silly as you point out. Thanks for the links – I’m on the road the next week but I’ll take a look after I get back home.

  59. Traktor says:

    Vic, you broke it down for me the last time Cogliano’s shooting % was being discussed. I remember it all, and appreciate the insight, but I just can’t look at that and determine the most likely or maximum shooting output increase.

    It would be interesting if you know what the number was for that same stat for Hemsky. He had 87 shots in 71 games in his 20 year old season, and while I assume your stat would say Hemsky isn’t a candidate for a big shot increase, it was painfully obvious from watching him play that he could be a 200 shot/year guy. I think he had 184 in 74 games this year. Sometimes you just have to believe in your eye witness events.

  60. Traktor says:

    “But how could that be the argument when Cogliano didn’t get powerplay time this past season and few are projecting him to play tough minutes?”

    Depends on who is projecting what.

    The way I see it

    2nd line and 2nd unit PP = 24 goals 57 points

    2nd line and no PP time = 19 goals 45 points

    3rd line and no PP time = 14 goals 35 points

  61. Traktor says:

    “As Vic points out – the bigger issue with Cogs is shots against this coming season.”

    I remember Edmonton getting blasted with 45+ shots against a game vs the Wings in our playoff run. How many of these shots were quality though? He had one of the best +- so they couldn’t have been can’t miss scoring chances could they?

    “he wasn’t an elite scorer in college”

    I was tied for the team lead in goals playing on the 2nd line and didn’t play on the 1st unit PP until half way though the year. I belive he was tied for 7th in the nation in goals scored.

    If Cogliano wasn’t an elite goal scorer at the college level, does that mean that Dany Heatley (8th) and Vincent Lecavalier (9th) are not elite goal scorers at the NHL level?

  62. RiversQ says:

    performanceoil said…
    As far as possession, of course possession matters. Trouble is, Corsi does not measure possession. In fact, it measures loss of possession much more directly. Obviously trading possession for a (good) chance to score is a great swap. Losing it for a low %’age chance on net isn’t.

    I think this misses the point.

    I think you’re correct that it doesn’t measure possession directly. However, Corsi reflects where the puck is more than anything else IMO. Sure Detroit Player X gave it up with his crazy shot from LW boards at the blueline on the rush, but where’s the puck even when this happens?

    Heck even if we use Detroit as the poster team for putting up crazy low percentage scoring opps in bulk, can anyone reasonably argue that they don’t keep the puck in the other end a lot more than the other team? Can anyone argue they don’t also have a large amount of possession at that end as well?

    I’m fully on board with the notion that the Babcocks and Carlyles of the NHL boost these numbers by their respective styles, but it’s just not reasonable to argue the puck isn’t most of its time in the right end of the rink when their teams are playing.

    Does Corsi tell you about scoring chances or shot quality or anything else directly? No, not really.

    However, if you’re watching your team get bootfucked by the Corsi numbers every night, what on Earth would make you think your team is good at playing hockey?

    Common sense says if I want my team to have a good chance at winning hockey games, I want the puck at the other end of the rink as much as possible. After that, I’d like to see them get it to the net a lot too.

    Of course that could just be me.

  63. PDO says:

    Vic.

    Love that you’re posting more lately.

    But you’re agreeing with goalies. And no offense to Bruce and uh… you other guys :D, but I hate you all. You’re all fucking bat shit crazy. You actually enjoy getting a frozen piece of rubber shot at you and attempting to get in the way. Think about that for a second. You’re all fucking insane. And this is coming from a guy who prides himself on his shot blocking ability on the PK at a decent level of hockey and has played with guys who have been paid to play the game.

    You agreeing with them is a definite problem. No matter if they’re right or not, they’re fucking bat shit crazy. Goalies are the Ozzy’s of hockey. They may move everyone forward, but they eat live bats FFS!

  64. Vic Ferrari says:

    mc79hockey said:

    My numbers on Cogs are 379 and 508, for what it’s worth. Might have the EN there, can’t remember.

    With empty netters in I get 380 and 507 for Cogs. So I imagine that you have ENs in yours. Important to filter the empty net stuff out though, even just the goals if its convenient.

    Important to remember as well that this data is being gathered on the fly by several different people at once, and as of the last two years the shifts have been rationalized (so that player A’s shift starts exactly the moment player B’s shift ends) which of course never really happens in the game.

    I would imagine that the difference with BTN is that they are just 5v5 but include empty net stuff, timeonice is 5v5 and 4v4 but excludes empty net stuff.

    I think that PO has missed my point entirely though, BTN’s problem was that he was calculating Corsi as:
    Fenwick + (team blocked shots) – (opposition blocked shots)

    So the +’s and -’s were in the wrong spot for most of the season, and lat I checked (a couple of weeks ago) they were being ignored altogether.

  65. Bruce says:

    Common sense says if I want my team to have a good chance at winning hockey games, I want the puck at the other end of the rink as much as possible. After that, I’d like to see them get it to the net a lot too.
    Of course that could just be me.

    Well I’m fucking bat shit crazy so common sense need not apply. I argued, vociferously at times, during the season that scoring chances against the flow of play tend to be better opportunities, largely negating the effect of shots on goal “advantages”. Some teams, notably Edmonton in recent years, consistently do better when outshot than outshooting. The Red Wings are the poster boys for outplaying, outshooting, and outscoring, but let’s face it, there’s only one Detroit so to use them as the example of how it Should be done is a lot different than what might actually work for a less exceptional club.

    My general observation is that teams that soak up pressure and shots and strike on the counter attack hold their own better than it appears they should. One result is I tend to look askance at the meaningfulness of Corsi numbers, and at the braying in some quarters that goalies on dominant teams should have both fewer shots and lower quality shots against. Typically (IMO) it’s the former and not the latter; it’s hard to maintain a top-drawer Sv% on a team that tends to carry the play and allow few shots. Folks who see Sv% as the be-all-and-end-all of goalie stats tend to underrate the contributions of the goalie in such circumstances, even if he has an excellent GAA and a ton of wins to support a middle-of-the-road Sv%. It’s a tough gig to stand around for 5 or 10 minutes, then face a sudden counterattack. A league-average Sv% in such circumstances is pretty damn good in my books.

  66. Oilman says:

    some sarcastically says “big saves” in 3….2…..1….

  67. Vic Ferrari says:

    AO:

    Just a quick note on Brunette, as you mention him earlier. This excerpted from the Minnesota Wild official website:


    Priority • Increase offense and puck possession without sacrificing the solid defensive identity of our team.

    … Brought back Andrew Brunette to help offensive zone puck possession and create more goals.

    I think that’s a fair assessment of the Wild’s needs (they were the only team from the west with a negative corsi and negative faceoff zones to make the playoffs iirc). And that’s a fair description of what Brunette brings to the table imo.

  68. PerformanceOil says:

    However, if you’re watching your team get bootfucked by the Corsi numbers every night, what on Earth would make you think your team is good at playing hockey?

    First, team-wide Corsi is a different thing then Corsi at the level of the player. One can be significant while the other is not, however contradictory that may seem.

    Secondly, I find myself becoming confused. It seems the major thrust of using things like Corsi (i.e. numbers) is to remove bias and subjectivity. Yet, to defend the numbers such as Corsi, qualitative statements are made, (or Vic’s case, incorrect use of statistics) which reintroduce the bias and subjectivity. Believing that outshooting the opposition is important to winning the game, doesn’t make it necessarily so, no matter how much sense it makes on the surface. If it were important, the numbers would say so.

    And the numbers say it is not as important as people are making it out to be. At the team-wide level, the correlation between goal differential and shot differential is only 0.216 (last season). This is a weak correlation, though it is certainly a factor in winning. Now, to be fair, the correlation may be better if you looked at EV GD vs. EV SD. Additionally, one season is a fairly small sample, so including additional seasons would possibly strengthen the relationship. Still, while I certainly think shot differential is a factor in goal differential, I don’t think it a particularly overwhelming one. You can be a winning team and routinely get outshot – so say the numbers. That said, it may be the biggest single factor, but there must be many many other smaller factors that contribute in that case.

  69. PerformanceOil says:

    *A correction -> The value of 0.216 was for GD vs. points last season. GD vs. SD is a bit better, at 0.287).

    Now the level of the player is entirely different. Here the correlation between Corsi and Ev+/- is quite weak (I've gotten anywhere from 0.14 to 0.19, depending on which numbers I use). One source of noise in Corsi is easy to identify – missed shots will never result in a goal for or against.

    Again, these are the numbers. It is a simple regression analysis. Why are you (Riversq) telling me that Corsi is important without actually verifying it? The numbers exist, use them. I may well be wrong in my analysis, but I can guarantee you that Vic doesn't know what he is doing when it comes to statistics, so if you are trusting his, you are in error.

    As for Detroit, I imagine why throwing everything at the net works for them, is because they are extremely adept at regaining possession, and preventing quality scoring chances. As Bruce mentions though, taking the most extreme example to make your case isn't very convincing.

    Finally, in regards to your point that Corsi is generally reflective of possession time, at the level of the player – that may well be true, and I certainly agree it makes sense. Now prove it.

  70. PerformanceOil says:

    Vic,

    I did not use BtN’s Corsi, I used the SF/SA MF/MA and caluclated the Corsi myself. But an example:

    According to BtN, Pronger’s Corsi is 0.5/60 min.

    His SF/SA(/60) are 24.3/24.2. His MF/MA(/60) are 9.5/9.3. His BSF/BSA(/60) are 9.4/10.2.

    According to these numbers his actual Corsi would be (24.3+9.5) – (24.2+9.3) = 0.3/60 min.

    His Fenwick would be (24.3+9.5+9.4)- (24.2+9.3+10.2) = -0.5/60. Neither number matches his given Corsi, although with rounding you could get to 0.5 using the first method. If you think you know how the number 0.5 was derived, show the math.

    Not that it matters, since like I said – I calculated the Corsi numbers myself in order to run my regressions. As long as the base numbers are correct (and they probably aren’t, but they probably are as good as any out there), the regression is accurate.

  71. PerformanceOil says:

    Bruce,

    My thinking is pretty much in line with yours. To me, Hockey is very organic, and difficult to quantify, except in very discrete chunks.

    While I don’t know that the majority of goals come on the counter-attack, it certainly seems like the best chances often come that way.

    Beyond that, I think it is pretty clear that NHL teams defend some areas of the ice much more vigorously than others. Those same NHL teams work hard to take the puck into those areas in the offensive zone. There’s a reason why, and it isn’t because all shots are equal, or that shots are essentially a slightly weighted roll of the dice (in terms of both location and quality). And I agree it goes a long way towards identifying why Corsi has the problems it does.

    However, while it is certainly a large piece of the puzzle, it leaves us no closer to being able to predict which player will do well with a poor Corsi, or poorly with a good Corsi, or everything in between, unfortunately.

  72. Oilman says:

    dem’s fightin’ words!

  73. Vic Ferrari says:

    Performance Oil:

    BTN had the corsi wrong before because the blocked shots for and against were entered backwards. I eventually got around to emailing Gabe on this a month or so ago, and I see that he has changed it.

    It takes seconds to show that he’s using Fenwick as Corsi now, just cut and paste a section of the code into a spreadsheet, calc the fenwick yourself in a new column … it’s the same as his corsi, we a touch of rounding error on the occasional number.

    It shouldn’t have required this much typing to clarify this.

  74. Vic Ferrari says:

    As I say, like mc79, I’m using the play-by-play sheets. I’m not personally verifying any of the PBP sheets or TOI sheets … if there are occasional errors in those, and surely there are a lot in the latter, caused by the people who entered the data, I can live with it.

    It looks like whoever entered the play by play for game 20073 vs the Canucks erred and put an EV in the column instead of a SH for Cogliano’s goal.

    The horror. I’ll get a man on it right away.

  75. PerformanceOil says:

    It takes seconds to show that he’s using Fenwick as Corsi now

    Then why didn’t you do it using Pronger’s numbers, like I just did? I got a Fenwick of -0.5, while his listed Corsi is 0.5.

    Write it out here, so I can see what I did wrong.

  76. PerformanceOil says:

    if there are occasional errors in those, and surely there are a lot in the latter, caused by the people who entered the data, I can live with it.

    Good, then we agree. The raw data at BtN is valid, and conclusions derived thereof are as well.

    Excellent! So, Corsi and EV+/- correlate somewhere around 0.15-0.2. Which is poor.

  77. Vic Ferrari says:

    Performance Oil, you need to take a Bayesian approach (which is why I’m jonesing for the offside stuff, in case it wasn’t obvious already).

    Did you listen to nothing I told you last time? You are time consuming, PO.

    If we look at the reason for Glencross’ success as an Oiler, you can just dump it into a spreadsheet and apply frequentist methodologies, you will come to the conclusion that shooting percentage was the principle cause of his success as scorer, and shooting% while on ice the principle driver of his ability to outscore. And nobody here will argue with you. That’s just plain obvious.

    When he was acquired he had a mediocre shooting% iirc, that’s surely part of the reason why he came cheap. Though he was fourth (again by memory) in CBJ forwards by Corsi. That was a good bet by Lowe there.

    My point is that there are elements here (most notably the save and shooting percentages) that contribute enormously to the result, even moreso over a short timeframe.

    In an effort to divorce these things, we can look at front halves and back halves of seasons. Because to use the analogy of a hybrid of trivial pursuit and craps with weighted dice … I want to know how much the dice are weighted for each player/team, and how good they are at trivia. And looking at the cummulative data for a season of trivia craps is going to tell me how important it is to be lucky at the table (unless you believe that thinking positively can affect the outcome of a dice roll).

    When you’re looking at teams, from any one section of the season to another, EVshooting% just has no sustain, hell even EVsave% has so much luck attached to it that poor goalies regularly outperform great goalies over short stretches.

    Faceoff zones and shots+/- (or use Fenwick+/- if your sample is small) does repeat, and it also predicts future success. Not a magic bullet, but a solid indicator.

    Turns out trivia craps unfolds the same way. For obvious reasons.

  78. PerformanceOil says:

    you need to take a Bayesian approach (which is why I’m jonesing for the offside stuff, in case it wasn’t obvious already).

    First, no I don’t. It is pretty simple. Given (only) a player’s Corsi number last season, could you predict their EV+/-? The correlation between the two is very poor, so the answer would be no, not with any accuracy.

    Secondly, I highly doubt you have any clue how to use Bayesian probabilities correctly, since even the simplest applications are reasonably complex. But again, if you are convinced you can, then show me an example for a single player and show the math here, where it can be critiqued.

    I’ll even start you out: Given Cogliano’s Corsi during the first half of the year, what is the probability that his EV+/- would be +7 for the second half?

    That is the sort of question you answer with Bayesian probability and I don’t think it is possible to do so, since his second half EV+/- is completely independent from his first half Corsi.

  79. Vic Ferrari says:

    For any one dice roller, or any one player, of course not.

    Repeat the procedure for the teams themselves as a whole, and for the players collectively, relative to teammates.

    There is surprisingly little repeatability in shooting% (you have to separate the F from the D of course) for players or teams. Intuitively you’d think that there would be more. And very little in EVsave% behind them either, though clearly there is a bit of something there, the noise is so much louder than that impact it’s crazy.

    As well, if we use Pearson correlation to infer causation, convenient as hell I know, but there are a lot of wacky things that can happen. You can split the Corsi+ and Corsi- and correlate it to EV+ and EV-, present or future (all relative to teammates) and both correlations will be much stronger.

    As we discussed before, you can take the correlation of goals to both shooting% and shots and the sum of causation falls a mile short of 100%. Since we know that, by definition, goals=shooting%*shots, we know that’s not right.

    And though we all know that shooting%, or especially a track record of it (or creating it for teammates) is really important, it’s going to shine through too strong here, because in the short run it just makes such a difference. Sometimes the puck falls to you in just the right spot, sometimes you make your shot, sometimes you don’t make you shot and it goes in anyways.

    I scan through the hockey numbers type blogs now and again, they are taking an identical approach to yourself. And the conclusions are off the wall to my intuition, and you would never wager with this information because you would surely lose your shirt. Probably outside the hold. In fact one of the guys who runs a numbers site had a long stretch where he tried to do this, it was going sideways when he stopped, declaring the game too organic. Props to him for having a go, though.

    I have a degree in engineering by the way, though I only used it briefly in a technical capacity before moving into business. And I was a good student, though admittedly I don’t remember much of specific classes from back then, and am generally too lazy to do a lot of legwork for my hockey hobby. I don’t find you intellectually intimidating, not in the least.

  80. PerformanceOil says:

    As well, if we use Pearson correlation to infer causation

    When did I discuss causation? Show me.

    As we discussed before, you can take the correlation of goals to both shooting% and shots and the sum of causation falls a mile short of 100%. Since we know that, by definition, goals=shooting%*shots, we know that’s not right.

    No, you think you know something, and you’re wrong.

    Using the Oiler’s Shots, Goals and SH%’age from last year (from Time on Ice), the correlations are as follows:

    Shots vs. Goals: 0.532
    SH%’age vs. Goals: 0.643
    Shots vs. SH%’age: 0.249

    So this means that the variance between shots and goals overlaps by 53.2%. Between SH’age and Goals it is 64.3%. Between Shots and SH’age, it is 24.9%.

    Further, the variance in goals scored explained by SH’age AND shots is = (0.532 + 0.643) – (0.249) = 0.926, or ~93%. True, it is not 100%, as it should be. However, including a larger data set (37 players is quite small), would almost certainly bring the correlation to 100%. As it is though, it is very close; certainly it is not ‘a mile short’ as you claim. There is a very strong correlation between the values, and the math proves it.

    There are many things linear regression is a poor tool to describe, but a linear relationship between two continuous values is not one of them (assuming accurate measurement).

    Try again.

  81. Oilman says:

    …so the square of the hypotenuse is equal to the the sum of the square of the other two sides?

  82. RiversQ says:

    The correlation discussion is interesting and I rather not interrupt but…

    performanceoil said…
    First, team-wide Corsi is a different thing then Corsi at the level of the player. One can be significant while the other is not, however contradictory that may seem.

    Well, it doesn’t seem contradictory to me. I would think the problem is mostly sample size – the team has 82*60+ minutes to develop a sample set while individual players have less than 82 games at around 20min per game or less to develop theirs. If you want to break it down by situation (and you probably should) then the numbers go south from there.

    If you concede that Corsi looks like a good idea for a team, then the argument for Corsi vs. individual EV+/- is largely a waste of time. (I probably never should have mentioned Detroit, because it’s separate from the point at hand.)

    Personally, I think Bruce is crazy – there’s no way the Oilers trot out there looking to get outshot and looking to spend most of their time in their own end. No hockey team would do that and there’s far too much chance in play for NHL teams to choose to emulate the Azzurri and really think that’s a reasonable idea.

    Thus far Bruce has latched onto a handful of playoff games which happen to be in a season where the Oilers posted their best shot differential in Lowe’s tenure. Go figure.

  83. RiversQ says:

    performanceoil said…

    Write it out here, so I can see what I did wrong.

    This one is so easy I’ll take it:

    Corsi = SF + MF + BA – SA – MA – BF.

    Fenwick = SF + BA – SA – BF.

    There you go. You did have that wrong.

  84. Vic Ferrari says:

    Riversq:

    Yeah, I think you’re saying the same thing anyways, but it bears repeating that there is no such thing as a “rope-a-dope” gameplan in hockey. Just isn’t. Never has been and never will be.

    Sometimes you win games that you don’t deserve to win, that’s all.

    And further to your comment about Corsi (or more accurately puck possession in general), with injuries and maturation, or even a different mindset of the coaching staff, that can change a whack for individual players. So if I decide at midseason next year that Cogs should be playing tougher opp and taking more own zone draws, either for the benefit of the team in the long or short term, it is completely inevitable that his possession numbers (where his shifts endsd, shots+/-, etc) are going to take a kicking. But someone else on the team, or an aggregate of them, will see the reverse effect.

    That’s what we see, no?

  85. Vic Ferrari says:

    Performance Oil:

    Terrific stuff, thanks. I have some issues with it of course, but props for hanging in here. Hopefully you’re still here (that LT is so damn prolofic, leave for a day or two and you last comment is in the archives at this place) and hopefully we can both stop being quite so pissy about it.

    The sample you’ve chosen is a bit quirky because there are no icetimes there, of course. But it’s cool. With the same sample (sans goalies, 32 and 25 for Garon and Roloson respectively):

    But since you chose this data set we’ll stick with it for now.

    Correlation of Shots+ to GF-while-on-ice for Oilers skaters: .98

    Correlation of Shots+ to Shooting%-while-on-ice for Oilers skaters: .40

    Correlation of GF-while-on-ice to Shooting%-while-on-ice for Oilers skaters: .31

    So my question, and I ask honestly, is how do you separate these out?

    I mean you could pop them into a matrix and multiply it through I suppose, but aren’t you just spreading the luck around? (if one believes that on-ice-shooting% over a short stretch is mostly just luck). Or is it handled a different way?

  86. Vic Ferrari says:

    Edit for above: Garon wears #35

  87. Vic Ferrari says:

    D’oh! Another edit for above:

    Correlation of Shots+ to GF-while-on-ice for Oilers skaters: .94

  88. PerformanceOil says:

    RiversQ,

    Thanks for correcting me in regards to Corsi/Fenwick – I had my numbers mixed up.

    You misunderstood be in regards to significance:
    I actually meant that one might correlate quite well, while the other might not. Yes, it could be a sample size issue, but a season’s worth of player data is still pretty substantial, and I doubt the low correlation is only explained by sample size.

    f you concede that Corsi looks like a good idea for a team, then the argument for Corsi vs. individual EV+/- is largely a waste of time. (I probably never should have mentioned Detroit, because it’s separate from the point at hand.)

    1) I never conceded that point in the first place – in fact said the team-wide SD vs GD correlation is quite weak as well (~0.29)

    2) Different players may have different roles. Some might be told to throw everything they have at the net, others might be instructed to play a more patient puck possession/cycling type game. Thus different players could be doing well in their role even if they have poor Corsi numbers relative to their team. On the other hand, their Corsi numbers might be good/bad relative to players who play a similar role on different teams. Thus team-wide Corsi may be a measure of success while at the player level it is not (also keep in mind at the player level, there are 9 others on the ice who influence the number as well as the given player of interest, this would be irrelevant at the team-wide level).

    Personally, I think Bruce is crazy – there’s no way the Oilers trot out there looking to get outshot and looking to spend most of their time in their own end.

    Again, you are using Corsi as a measure of possession time, when there is no concrete evidence for that. To be fair, I am sure Corsi is at least partially a measure of possession time, but the question is, how strong of a proxy is it? You need to answer that question, before assuming it is fine to substitute one for the other.

    Further, you may need to have possession to generate shots, but you don’t need to be generating shots to have possession. This means there is no way Corsi can be a perfect proxy for possession, though it could still be a very good one.

    Thus far Bruce has latched onto a handful of playoff games which happen to be in a season where the Oilers posted their best shot differential in Lowe’s tenure. Go figure.

    The burden of proof doesn’t lie with Bruce (or me) to prove that Corsi is less important than you believe, or that it doesn’t measure possession. The burden is on you. Now, if you simply want to believe what ‘feels’ right to you, rather than proving it, great. But again, why bother using numbers in that case?

  89. PerformanceOil says:

    Vic,

    I agree it would benefit us both to stop being pissy. If you keep an open mind about our discussions, I will try to do the same.

    there is no such thing as a “rope-a-dope” gameplan in hockey. Just isn’t. Never has been and never will be.

    No, I don’t think teams purposely build a game plan around allowing the other team to run around their end at will.

    However, teams might well allow easier access to some areas of the ice than others (and thus allow shots from some areas more readily than others). The bottom line is that shot differential is important, but so is shot quality (and execution, but that is what you attribute to luck).

    It is possible that you can be badly outshot, yet have the majority of quality chances. Is it probable? Could a game plan be built around such a strategy (and note, the team with the most quality chances may have also had the larger share of possession time)?

    I don’t know the answer to those questions, but they present possibilities which may explain why Corsi doesn’t seem to be a strong variable in determining GD.

    Until these questions are answered, (and any other critiques that come up) I don’t know how it can be considered wise to use Corsi to evaluate a player, rather than their actual results (EV+/-, quality of.. numbers, etc.).

    So, in the absence of these answers, my only conclusion is that Cogliano had a great season, backed by a likely unrepeatable SH%. Time will tell us whether he is the real deal, not Corsi.

    BTW, in regards to Corsi=possession time, I have seen your post on IOF, and will discuss it there.

  90. PerformanceOil says:

    So my question, and I ask honestly, is how do you separate these out?

    Forgot about this.

    I’m not sure exactly what you’re asking for. What are you trying to separate out?

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

© Copyright - Lowetide.ca