Over at the always-interesting and usually-civil Game Day Thread at Lowetide, yours truly has been happily eating crow about Ales Hemsky. In my defence I pointed out he really did turn the corner in early November -- about ten seconds after I published some critical remarks about Hemsky's early-season play -- but Ales has been rolling for the most part the last five weeks. As we all hoped he would.
But while I'll take the hit for being “wrong” about Hemsky over there, I think I will claim being “right” about something else in the safety of my own blog. All year long I’ve been saying the best LW for the Horcoff-Hemsky duo is one Dustin Penner, who happened to first join the line on Nov. 1 in Carolina, just as Ales was getting it together. (Coincidence? Or not?) Hemsky played real well in Carolina before finally breaking the goose egg with two spectacular goals in Philly on the 2nd and then going on a bit of a run. His linemates have changed from time to time, with Penner finding time in the dog house in between, but for the most part the 27-10-83 have played the lion's share of time as a unit.
As we first saw last Dec.-Jan., “Horpensky” have been a very effective combination, compiling some eye-popping shots data. Assuming (dangerously) that I have entered the Timeonice.com code correctly and understand the results, this tells a tale:
27-83 together:
+9/-3 = +6 goals
+72/-41 = +31 shots
+139/-79 = +60 Corsi
27 apart:
+7/-4 = +3 goals
+71/-86 = -15 shots
+152/-157 = -5 Corsi
83 apart:
+7/-10 = -3 goals
+111/-125 = -14 shots
+206/-212 = -6 Corsi
Then there's this from Hockeyanalysis.com (all numbers are GF/GA per 20):
But while I'll take the hit for being “wrong” about Hemsky over there, I think I will claim being “right” about something else in the safety of my own blog. All year long I’ve been saying the best LW for the Horcoff-Hemsky duo is one Dustin Penner, who happened to first join the line on Nov. 1 in Carolina, just as Ales was getting it together. (Coincidence? Or not?) Hemsky played real well in Carolina before finally breaking the goose egg with two spectacular goals in Philly on the 2nd and then going on a bit of a run. His linemates have changed from time to time, with Penner finding time in the dog house in between, but for the most part the 27-10-83 have played the lion's share of time as a unit.
As we first saw last Dec.-Jan., “Horpensky” have been a very effective combination, compiling some eye-popping shots data. Assuming (dangerously) that I have entered the Timeonice.com code correctly and understand the results, this tells a tale:
27-83 together:
+9/-3 = +6 goals
+72/-41 = +31 shots
+139/-79 = +60 Corsi
27 apart:
+7/-4 = +3 goals
+71/-86 = -15 shots
+152/-157 = -5 Corsi
83 apart:
+7/-10 = -3 goals
+111/-125 = -14 shots
+206/-212 = -6 Corsi
Then there's this from Hockeyanalysis.com (all numbers are GF/GA per 20):
27/83 together
+1.500/-0.643 = +0.857
27 apart
+0.817/-0.350 = +0.467
83 apart
+0.594/-0.934 = -0.340
... which is just a bit of an eye opener, isn't it?
It's important to realize that Hockeyanalysis uses strictly GF/GA, which by definition is a small sample size this early in the season. Penner has achieved better results on the scoreboard, both for and against, than either of his highly-rated linemates, but it's early. Using the larger data sets of shots and Corsi available courtesy Timeonice, both players had virtually identical numbers when apart, and both are hugely improved since they are together.
Indeed, the Horpensky unit has been dominant in recent games. Here are shots and Corsi data as well as traditional +/- for forwards over the last five games:
Player ** = Shots / Corsi / Goals
-------------------------------
Penner ****** = +16 / +41 / +2
Horcoff ***** = +14 / +30 / +3
Hemsky ****** = +11 / +35 / +3
Gagner ****** = +4 / +7 / EV
Schremp = *** = +3 / -1 / +2
Nilsson ***** = +2 / +5 / EV
Cogliano **** = +1 / +3 / +1
Pouliot ***** = 0 / -8 / EV
Brodziak **** = -1 / -4 / EV
Cole ******** = -2 / -1 / EV
Stortini **** = -4 / -11 / -1
Moreau ****** = -5 / -7 / EV
Reddox ****** = -9 / -17 / -2
Note this five-game stretch began after the series against Corsi kings Detroit. It’ll be interesting to see how we stack up against the powerhouse Sharks tonight.
Speaking of which, it’s game on. Goilers!
Penner ****** = +16 / +41 / +2
Horcoff ***** = +14 / +30 / +3
Hemsky ****** = +11 / +35 / +3
Gagner ****** = +4 / +7 / EV
Schremp = *** = +3 / -1 / +2
Nilsson ***** = +2 / +5 / EV
Cogliano **** = +1 / +3 / +1
Pouliot ***** = 0 / -8 / EV
Brodziak **** = -1 / -4 / EV
Cole ******** = -2 / -1 / EV
Stortini **** = -4 / -11 / -1
Moreau ****** = -5 / -7 / EV
Reddox ****** = -9 / -17 / -2
Note this five-game stretch began after the series against Corsi kings Detroit. It’ll be interesting to see how we stack up against the powerhouse Sharks tonight.
Speaking of which, it’s game on. Goilers!
19 comments:
Souray is the new Pronger, Penner is the new Smyth, Roli is the old Roli and MacT seems to have decided to stop screwing around. We still have a hill to climb, but there are some good signs right now. The Oilers have managed to keep themselves in the mix, and everything seems to be coming together for a bit of a run. If we can start getting consistent contributions from our 2nd line, and get anything of value from the 4th, we should be in good shape.
Interesting another solid analysis from Bruce. What do you think about Horcoff's role on that line?
Could Penner's improved play also have to do with playing with the best center on the team, not St. Fernando?
I know all of the time that Penner played with Hemsky, he also played with Horcoff, so there is most likely no numbers to use, but any thoughts?
B.C.B.,
I'm not Bruce, but I think it's clear that Horpensky is more than the sum of its parts.
None of the three players are as good when you take out any of the three and put in a replacement. Nilsson did ok in substitution of Penner (small sample though), but no one else had. Both Penner and Hemsky stepped off a cliff last year when Horc went down. I don't think we have seen much of Penner-Horcoff-X, but I'm willing to wager it would be a significant drop-off. And, I'm pretty sure our PP has been better since they were reunited.
It really is the classic case of size- all-around play - skill; just a hell of a well-balanced line. Besides hearkening back to the brighter days of childhood, the 'Nintendo-line' moniker really sums it up nicely.
Bruce can you please write a negative post about Cole and Gagner?
I know all of the time that Penner played with Hemsky, he also played with Horcoff, so there is most likely no numbers to use, but any thoughts?
B.C.B. Yeah, there's numbers, in fact the exact same type of numbers for 27-10 as for 27-83. Originally I approached this as "what changed for Hemsky?" when issuing my mea culpa on Lowetide. To which the obvious answer was "Penner". Horcoff OTOH has been pretty much a constant with Hemsky all year, therefore the Horc-with-and-without-Penner numbers should be pretty similar to Hemmer's. It's one game later, but let's check them out:
Timeonice.com
27-10 together
+9/-2 = +7 goals
+85/-51 = +34 shots
+155/-98 = +57 Corsi
27 apart
+8/-5 = +3 goals
+66/-84 = -18 shots
+145/-155 = -10 Corsi
10 apart
+7/-8 = -1 goal
+113/-115 = -2 shots
+201/-215 = -14 Corsi
(Hockey Analysis.com)
27-10 together (92:18)
+1.300/-0.433 = +0.867/20
27 apart (172:25)
+0.928/-0.464 = +0.464/20
10 apart (227:24)
+0.704/-0.792 = -0.088/20
So similar results as with Hemsky; Penner has substantially better GF/GA when the two are apart, but both improve when they are together.
Note I added in TOI together and apart from Hockeyanalysis, which shows that Horcoff has spent less than 30% of his total ES TOI with Penner, yet the majority of his GF ON have occurred with Penner on his line.
For completeness I did the same analysis with Horc and Hemmer.
Timeonice
10-83 together
+13/-8 = +5 goals
+141/-115 = +26 shots
+267/-204 = +63 Corsi
10 apart
+3/-2 = +1 goal
+57/-51 = +6 shots
+89/-109 = -20 Corsi
83 apart
+4/-5 = -1 goal
+49/-60 = -11 shots
+89/-104 = -15 Corsi
Hockeyanalysis
10-83 together (218:58)
+0.913/-0.822 = +0.091/20
10 apart (100:44)
+0.794/-0.397 = +0.397/20
83 apart (110:00)
+0.727/-0.909 = -0.182/20
So in this case Horc and Hemmer play togther about 2/3 of the time, and are more successful when together. The difference in Corsi numbers is pretty extreme with any two of the three players.
I'm not sure if a mechanism exists for considering when all three are on the ice together, the above sites only work for pairs of players and not trios. However, it's clear from all these stats that Hemsky in particular appears to struggle in the absence of either of his current linemates, and has been "the bomb" when the line is together as a trio.
Sean: Using the same metric, Cole's performance with both Horc and Hemmer is pretty dreadful. In fact 26-83 have played together for 76:35, during which time they clicked for exactly Zero GF, which is a damning indictment of Cole's play and/or Hemsky's slow start. He's played a little more than two hours with Horcoff and the Oilers have scored 2 ES goals when both are on the ice together which is also pretty brutal. Shots/Corsi data for Cole with either first-liner is pretty much flat, very close to zero across the board.
Cole's lack of offence has been a bitter disappointment over the first third of the season, especially considering his huge ticket. If you want to compare who is earning his $4 MM and who isn't, based on all these results I'll take Penner over Cole 10 days out of 10.
None of the three players are as good when you take out any of the three and put in a replacement.
PerformanceOil: Agree absolutely. I think the above comment and the original post show that any 2 of the 3 are considerably less effective than all 3 together.
Nilsson did ok in substitution of Penner (small sample though), but no one else had. Both Penner and Hemsky stepped off a cliff last year when Horc went down.
That's for sure. With Stoll I called them the Minus Touch Line.
I don't think we have seen much of Penner-Horcoff-X, but I'm willing to wager it would be a significant drop-off.
We saw a 27-10-26 unit for a couple of games and everybody was salivating about how they would be great PvP, but it didn't last long. Meanwhile Hemsky had a bit of a run with Gagner as his pivot, but for the most part it's been Horc. That pair has run through the whole gamut of possible LW, less so with Nilsson who you mention but very similar with each of the rest.
So let's parse it this way: who is the best LW for Ales? Here's the results so far for each of the following when paired with Hemsky. I have folded Timeonice and Hockeyanalysis data in together, even though I suspect the latter is a couple of games out of date. This little study isn't meant to be as precise as three decimal places implies. Digits in the mirror are definitely less significant than they appear, but they're what we got:
Penner 93:19
+1.500/-0.643 per 20
+10/-3 = +7 goals
+79/-48 + +31 shots
+147/-94 = +53 Corsi
Moreau 86:01
+0.698/-0.930
+3/-4 = -1 goal
+35/-56 = -21 shots
+71/-85 = -14 Corsi
Cole 76:35
+0.000/-0.783
+0/-2 = -2 goals
+36/-36 = +- 0 shots
+65/-67 = -2 Corsi
Nilsson 46:17
+1.296/-1.296
+3/-3 = +- 0 goals
+29/-18 = +11 shots
+48/-33 = +15 shots
... indicating that Nilsson indeed did fairly OK with Hemsky, Cole was meh in terms of shots and nonexistent in terms of actual goals, and Moreau continued to bleed shots on goal even when playing with the best winger on the team. (To be fair, I think the captain got saddled with a lot of the Gagner gig as opposed to Horc Time).
I don't know how you can look at these numbers and conclude that there could possibly be a better choice to flank 10-83 than Penner. Which (ahem, MacT) I've been saying all along. The coach's extensive experimentation with all possible options certainly seems to have driven the point home.
And, I'm pretty sure our PP has been better since they were reunited.
Oh, yeah. I've been begging for Penner on the PP since the summer as well. He is an absolute force in front of the net, doesn't need to see a lot of the puck to be effective, but when the biscuit does come within his (long) reach he's got nice soft hands to do some serious damage with it. To this point he's only seeing about half the PP TOI as Hemsky, Horcoff, Visnovsky and Souray, but he easily outstrips all of them in team productivity while he's out there. (Penner +10.58/60; Souray +8.20; Hemsky +7.05; Horcoff +6.77; Visnovsky +6.06; all other Oilers < +6). In Desjardins' rating system, which is based on the PP unit's success when the player is on vs. when he's off, Penner is far out front and only Souray is in the same ballpark.
I remember calling Gagner to be demoted back in late October or early November. Not because I don't like the player, but in order to light a fire under his ass.
MacTavish may not have heard me, but the Hockey Gods have and they sent us Schremp and Brule.
Something to consider in 10 with vs. without 27/83 is that when he was without either of them, he was on a checking unit with 26-34. Now, those are in theory quality players who can at least create a little bit more than your average third-liner, but at the same time, without looking too closely at it, I'd wager that particular trio saw a lot of defensive zone faceoffs and tough-minutes matchups, which will hurt your outshooting/outscoring numbers no matter who you are.
Didn't Vic create a tool not too long ago for comparing trios in some fashion? I remember it being mentioned on LT's or MC's, but I don't go to IOF personally, so I couldn't tell you.
Thanks Bruce, lets hope Cole can turn the corner in give or take 10 seconds ;)
Doogie: You're right, there is a Timeonice tool for trios. Say what you like about Vic (and be my guest!) but his work is top-drawer when you can figure out the secret handshake that some of us find computer coding to be.
Here's the similarly organized data for Horpensky as a unit:
+9/-2 = +7 goals
+70/-43 = +27 shots
+133/-78 = +55 Corsi
Horcoff otherwise:
+7/-8 = -1 goal
+128/-123 = +5 shots
+223/-235 = -12 Corsi
Penner otherwise:
+8/-5 = +3 goals
+81/-92 = -11 shots
+167/-175 = -8 Corsi
Hemsky otherwise:
+8/-11 = -3 goals
+120/-132 = -12 shots
+223/-230 = -7 Corsi
... and the only column that I wish was available -- not that I'll be making that request -- is ES TOI so that joint rates could be expressed on a per/60 basis. TOi data is readily available for the individuals elsewhere of course, but for the identified combination we can only guess in context of the actual shot rates, which of course change drastically when they are together. If, however, you express those goals, shots, and Corsi figures as ratios, the trio's results are awesome.
In the case of Horpensky it is clear that they are wildly more successful as a trio than when separated, even when 2 of the 3 play together their rates plummet compared to their dominance as a line. Early times yet and small sample size, but impressive results for this power line.
Your additional point about Horcoff is well taken, Doogie. Seems to me Penner also drew some tough minute assignments on other lines, particularly playing RW with Pisani and Moreau early in the season.
(Btw, Doogie, thanks for going to bat for me, consistently, since that blow-up of a few months ago. I won't be forgetting that anytime soon, but after needing quite a while to lick my wounds I feel it's time to move on. Personal garbage aside, the IOF types bring important data and interpretation of the game, at least when they stick to talking about hockey. I was sorry to see you and Ender as the temporary target of another juvenile attack out of nowhere the other night, but didn't see any point in adding fuel to that particular fire. I made the mistake of escalating things once before. Anybody who doesn't like what you bring to the discussion just isn't paying attention, full stop.)
Thanks for the analysis Bruce. I think the numbers proved a hutch I had: Penner-Horc-Hensky need to be together all the time. Horc needs wingers of skill and size, and the two wingers need a good center.
All you guys are great!
Bruce,
Thanks for doing all the analysis that I was too busy/lazy to do. Nothing too surprising in what you posted, obviously. Some reasons for caution though (for reasons Dougie mentioned, and others), but critics of Penner and his acquisition don't have a lot of ammo right now. It may change, and I think a lot of why MacT was so angry with Penner has to do with what the man is capable of, rather than what he has been accomplishing. Still, whatever his deficiencies, and whatever the price paid for him, he fits a huge need for this team (LW for Horsky). I doubt a better fit could be found for cheaper, and I think an argument could be made that he is a reasonable replacement for Smyth, with a lower salary, possibly higher upside, and Nilsson as a side-benefit.
Agreed completely about Cole, and considering his prior accomplishments, he is holding this team back in a big way. I can't figure out what the problem is. You can see the talent pretty clearly, but he almost looks like a raw rookie who hasn't figured things out yet. And, it is impossible to imagine that motivation is the problem, given that he's coming up to his big pay-day. The Pronger curse continues, I guess.
As far as IOF, the place is full of big ideas, and whether right or wrong, they generate thought and discussion, and thus, better understanding. I think they misunderstood the meaning of irreverent though, since it means neither close-minded nor disrespectful.
@Bruce: That was my bad in Friday's GDT. I just get so tired of the sanctimonious bullshit sometimes. I also dislike how Vic runs TOI.com: it's a great tool, but you have to know how to use it to use it; for all his blither about how the computer-illiterate should have access to this stuff, he basically writes his site for himself and his cohorts.
But honestly, I'm beginning to be past the point of caring. As I said late in that post, I think the best thing is just to ignore him when he gets like that. I'm given to understand that there are a lot more people than you, me, and Ender who think he's a twat, but no one's going to call him out as long as he keeps contributing, which doesn't seem right, but if those are the ground rules, I can adapt.
Anyway, that's quite enough Vic-bashing. I did want to say that when I get a chance after finals, Ender pointed out a standings widget written in Java that I might be able to use to create auto-updating baseball standings. I've wanted to get back into coding for a while, and when time permits, I might do just that. I also have a post about the physiology of goaltending (in response to a Vic question, as it happens) in the oven that just needs to be committed to virtual paper; that should be done and sent off for vetting by my physiology prof in the next couple of days.
As for Penner, I said a year and a half ago that I liked the player but not the contract. I'm still not totally sold on the contract, but if he can play like this for most of the rest of the year, then I can at least see it.
I also dislike how Vic runs TOI.com: it's a great tool, but you have to know how to use it to use it; for all his blither about how the computer-illiterate should have access to this stuff, he basically writes his site for himself and his cohorts.
Oilswell covered this in the first sentence of the first comment when he said:
"Well the interface is critical to snare non-programmer folk."
I have made this point (key words: "secret handshake") myself and been more or less labelled a whiner for doing so, and fair enough. Nobody out here has any obligation to make their stuff accessible, it just seems logically that they "should". Vic certainly expressed as much in that informative first post that you so kindly linked here.
The data Vic compiles is fantastic, and I have bookmarked a handful of them (shots, faceoffs, icings) so I can go in and retrieve stuff. But there's lot more where I literally don't know the code word. e.g. the point I made above about calibrating perfomance of trios with their TOI as a group; very likely -- esp. given the name of the website -- there is another opage in similar format that breaks down TOI where the same sort of "&danrik&heniel" trick can be applied to suss it out. Unfortunately, though, I haven't come across that application in a blog thread. Since there seems to be no "home page" like Behindthenet or Hockeyanalysis where one can simply click tabs and follow the logic, a technoflub like me is limited in access. Which is too bad, because I think I pretty much fit the profile of one "of the guys with ZERO programming skills, but good eyes, rational minds and a passion for the game". I know I'm making good use of the data I do know how to access. I am in Vic's debt for scraping the game sheets and making it available in whatever format, and I won't lose sight of that fact, regardless of other BS that might go down from time to time. I'm certainly not beholden, but I am appreciative.
But honestly, I'm beginning to be past the point of caring. As I said late in that post, I think the best thing is just to ignore him when he gets like that.
Good plan. I tried to change the subject four or five times ("First goal is Big") cuz it wasn't a situation that required more gasoline. I find the whole subject of Oilogosphere eugenics to be distasteful in the extreme, although I understand it has happened before (HF Boards). To me it is a big club of Oiler fans, pretty much everybody has something to add. You'll never hear me in the group that is telling Hunter or Traktor to screw off, because in the 10% of their posts where they're not just slagging MacTavish or Horcoff they bring alternate viewpoints of some substance. Of course I'd prefer that both the screeds and the "Big Ben" responses they often provoke would disappear unless they bring new information to the discussion (such as an update on Cogliano's current Sh% !) Lowetide serves a hugly important function by leaving the door of his one-room school open for absolutely everyone. His game day threads are gold for their range of commentary.
Those likeminded folk who share interest in say, a branch of statistics, will naturally hive off into smaller groups, I don't see the need to "cleanse" any particular location. For the most part these diversions turn out to be temporary and the group returns to adult conversation.
no one's going to call him out
I think he's been called out quite a few times, actually. No one's going to kick him out, however, and that's as it should be in my view. As I said, I'm opposed to Internet eugenics; someone would have to go a long way off the line to be banned from this blog.
I also have a post about the physiology of goaltending (in response to a Vic question, as it happens) in the oven that just needs to be committed to virtual paper
As an ex-goalie with a bit of a scientific bent, I'll look forward to that, Doogie. Feel free to pop by my comments board and post a link when you have it up somewhere.
It is nice to see rationalism and manners rule in some part of oilblogosphere.
After reading this, I have to rethink some of my behaviours, since they occasionally are based in rage, rather the reason.
As a non-stats guy, I like reading Bruce's and Lowetide's analysis, since I can get my head around it. I even once tried to do it myself (PIM/game as a measuring tool for aggression). but failed.
Again Thanks Bruce for allowing me to observe the relationship between Penner, Horcoff, and Hemsky through numbers opposed to gut feelings.
@Bruce: I certainly will. I may not have the stats background of some (including Ender; he's simply too busy to even bother), but I can kick the hell out of some sports science, and I intend to as the situation warrants.
@BCB: Hey, points for effort.
Bruce:
You saw this post, right?
Everything seems pretty clear explained for ordinary visitors and "the cohorts" alike.
P.S. Doogie:
Don't click that link! Don't click that LIIIIIIIINK!
I don't want you to compromise your principles;)
Slipper, yes I saw that outstanding piece, in fact the methodology revealed there allowed me to tackle the above.
The TOI pages I do know how to access (shots, faceoffs, icings), I use. But I'm pretty sure there are others that I simply don't know what they are or how to get at them. At one point MC79 undertook to develop some sort of klutz-friendly interface to these data, but as yet I'm not sure one exists.
Post a Comment