(Without having looked at any of the scores from tonight, although they announced the Vancouver score on the flight: VAN, DET, PIT, BOS; I was taking VAN anyway.)
124 NHL forwards meet the following criteria:
- At least 50 ES shots in 2007-08
- At least 50 ES shots in 2008-09
- A shooting percentage better than 9.3%, the average for forwards in 2007-08
In 2007-08, they collectively shot 12.2% on 1988/16333 shooting. This year they took a collective 15826 shots. What was their shooting percentage?
Bonus question: The top 25 by this criteria scored 466 ES goals on 2784 ES shots in 2007-08. What was their shooting percentage this year?
Bonus fact: The Avs had Marek Svatos and Paul Stastny go 43/220 last year at ES. They followed that up with a 12/186 campaign. Add that to to the goaltending and it’s fairly easy to see why things went so wrong.
Even more random fact: Mats Sundin was a huge disappointment for the Canucks at ES, scoring a pedestrian 0.45/1.01/1.46 at ES. He was the same Mats on the PP though - 2.21/3.97/6.18. I wonder if conditioning was his biggest issue - it should be less of a factor on the PP.
I have no idea. I’m going to assume that some of those guys have consistently good SH% and belong in that pile while others just got lucky. Let’s say 10.2% For the bonus I’ll say 10.5%.
I’m thinking that the point of the post is that these things are near enough to random so I’ll guess 9.3% for both questions. Although Jonathan’s reasoning does make sense to me, I won’t be at all surprised if I’ve overbid.
I was thinking about 2/3rds of the way back to the mean for the first one, so that put me the same as jonathon. So I’ll reverse ‘Price Is Right’ him with a 10.1% guess, that should be the lowest stab you’ll get.
For the bonus question, hrm. I think that Cogliano and Horcoff were probably in that group last year, and both fell a whack this season, especially the latter. Ribiero wasn’t in the news much this season so the Gods must have lost their desire for him. You’ve given us the spectacular drops of Svatos and Stastny. And ‘top 25′ seems arbitrary, plus the drop must have been huge if you were to bother with a bonus Q. So I’ll go with 9.9%.
D’oh, so much for trying to sneak in with what I thought would be the lowest bids.
What are the prizes, anyways?
10.4 and 11.9
This brings up a thought. There is a way to do this with pure math, but it is very difficult, and unless oilswell is still hanging around in the shadows, I doubt that any of us will try.
You’ve looked at this EVshooting% thing enough, though, that you probably have an opinion on what the true distribution of shotting ability is amongst NHL forwards.
If you literally drew that curve on a sheet of paper with a pencil. Then ran a simulation with the real shot totals (skewed towards the higher% shooters, because they probably get more ice) … you should get something close to the answer to Q1 above. If it’s a bit off, then change the shape of your curve and try again. Rinse and repeat.
And every other phenomenon regarding the drift of EVshooting% for players should be explained, in the right measure, by the final curve. And it should work equally well on other seasons as well, and give us a probablistic expectation and valuation for each player.
We’d be cooking with gas, then, methinks.
10.6 and 11.8. (I know that’s not statistically different from Quain…)
As always, there will be no prizes other than a sense, entirely within yourself and not provided by me or guaranteed to be acknowledged by anyone else, of how smart you are.
@Vic: Yeah, that’s sort of where I’m headed. Shot rates are stunningly consistent. Just ridiculously so. Very few guys move by more than a shot or two/60 in the course of a year. If I were to apply JLikens stuff on the bias of home scorers in certain rinks, I’d probably have that even tighter. Complete tangent: his site is awesome. If people who care enough about this to read the comments at my site aren’t reading it, they should: objectivenhl.blogspot.com.
Coming back to my point, if you follow this through, a guy playing 900 minutes at ES with a constant shooting percentage of 10%, coming off a season in which he took 10 ESS/60, we’d be reasonably confident that he’d score 15 goals, +/- say 3 goals.
The shooting percentage drift, of course, screws things up. If you had some reasonable way of saying “OK, this guy’s mean is X, we know that
My best guess at the true talent of NHL shooters at ES is something like 6% to 15%. That, of course, includes the shot quality factor - part of the reason some guys have better true talent shooting percentages is that they tend to take more high percentage shots. While it might seem weird to have the long tail on the good side of the mean, you have to consider that NHL players are a subset of hockey players as a whole and there’s a cutoff at a certain level below average.
All of this said, I think that you could come up with a decent way to model the number of ES goals that a player will score in a year. I’m kind of toying with a way to do that…maybe I’ll come up with one before next year and invite the masses to try and beat it.
I would be cool to pin the ’shot quality’ issue into something that could be wagered against. That would be an accomplishment.
And of course the stuff we’re looking at in this thread has everything to do with winning your hockey pool and little to do with winning hockey games.
On ObjectiveNHL, I agree that’s extremely honest and well written stuff, one of my top five reads on the internet. And the post you link to perhaps explains why EV shots data never does overtake the Fenwick and Corsi numbers, even given a huge sample. I think when I originally started talking about this stuff, I thought that after a week, at team level, shots data would overtake corsi/fenwick, and about four times the timespan for player data. But it just never happens. It gets near as dammit eventually, but never outperforms it.
Maybe this is because of the scorer bias. I mean on some shots there can be a lot of disgreement as to whether it was really “on goal” or not, but few would dispute that it was a shot directed at net.
Just to add; I would love to go out for a night of hard drinking with any of Regier/Corsi/Ruff, I mean folks don’t keep working together this long unless they are on the same page.
There is something deeply sensible about the corsi metric that even the cold hearted seem to underestimate. I mean I hope I’ve done something to assure that Dennis was honest and reasonable in his measurement of scoring chances. Now go look at scoring chances for (SC+) against shots direct at the opposition’s net when you were on the ice (Corsi+). And look at SC+/SCtotal and Corsi+/Corsi-total … it’s so close to the simple truth of hockey that the poets have precious little space to write on.
I’d pay serious money, in fact I’d arrange it, to have a night of hard drinking with any of those Buffalo guys. I mean ’saw him good’ works pretty damn well in hockey, so we’re into fine brushstrokes. Still, these guys are off the hook.
And if I didn’t happen to catch a CBC radio inerview during the lockout when it was mentioned, absolutely nobody would be looking at it. Which begs the question: What else do these effers know?
Just to add, and unrelated:
If Joey “shit between our toes” Saputo ends up owning the Habitants, then all is lost. It would be unforgiveable for any thinking man to support a league that allowed this nutless, stereotype reinforcing, HBO subscribing, restaurateur intimidating, stripper bar owning piece of living crap, cheese mafia douchebag in their ranks.
Not that the bar is set that high in the first place, but sweet Christ, it would be like when Staples entered the the Oilogosphere … he’ll still be a douche, whether Willis and Tyler kiss his ass or not.
Vic,
“I mean on some shots there can be a lot of disgreement as to whether it was really “on goal” or not, but few would dispute that it was a shot directed at net.”
great, great point.
re: hard drinking
i think all you hockey blogger stat geek mofos should come to new orleans so we can all meet for a couple nights of hockey talk over tequila shots and scotch.
Vic: Why the hate for Staples? I don’t always agree with him but it seems to me he’s been as respectful of ideas from the blogosphere as anybody working for an Edmonton paper.
RE: we’d be reasonably confident that he’d score 15 goals, +/- say 3 goals.
Actually it would be 15 +/- 7 (19 times out of 20)
Secondly if you factor out the random variations, and assume that shooting % is distributed normally.
The true skill distribution would be:
6.6% - 9.6% (95% of skaters) or
5.6% - 10.6% (99.9% of skaters).
I’ll try to break it by forwards & defense later…
he’s been as respectful of ideas from the blogosphere as anybody working for an Edmonton paper.
That’s high praise Jonathan. I like Staples well enough but that “compliment” was too hard to resist.
Javageek, do you think it’s safe to assume that shooting% ability is distributed normally amongst forwards? I don’t know, but I suspect that Tyler’s sense is probably right, that there are a lot of guys bunched close together and then fewer guys with really good finish who drag the average up.
I suppose the enforcers might drag it back into a more normal-ish shape, if you include guys with fewer shots.
I don’t know if there is any data around with career EV shooting%s, but that should be close to the distribution of finishing talent.
RE: we’d be reasonably confident that he’d score 15 goals, +/- say 3 goals.
Actually it would be 15 +/- 7 (19 times out of 20)
Am I missing something here? Vast majority of guys within +/- 2 shots/60 year over year. 900 ES minutes is 15 ES hours. Guy with 10 ESS/60 has 150 ES shots. +/- 2 shots is plus/minus 30 shots. At a fixed 10% shooting rate, we’re talking about +/- 3 goals, no?
Javageek, do you think it’s safe to assume that shooting% ability is distributed normally amongst forwards? I don’t know, but I suspect that Tyler’s sense is probably right, that there are a lot of guys bunched close together and then fewer guys with really good finish who drag the average up.
The real problem, I think, is that shooting percentage doesn’t just measure finishing ability. It also includes a measure of the shots that a guy takes.
The answer to my questions, by the way, is 10.2% and 11.3%. There’s a shitload of randomness there.
And of course the stuff we’re looking at in this thread has everything to do with winning your hockey pool and little to do with winning hockey games.
Not sure I agree. Identifying the high shooting percentage guys is important because those are guys who can survive at a lower threshold of outshooting. If it turns out that goals are, within certain margins predictable, and you can find a measure of defensive play that’s useful, you can take a stop closer to
I keep meaning to write something about assists. Bizarre stat, kind of the RBI of hockey. I’m not even sure that they tell us much, except at the extremes. Guys like Thornton or Sergei Berezin, they probably have legit information contained in their assist rates. Other guys…I’m not so sure.
The answer to my questions, by the way, is 10.2% and 11.3%.
I just wanted to point out that nobody cares what any of you folks answered for the bonus question. It’s the initial one that counts
12.2 -> 10.2 (9.3)
16.7 -> 11.3 (9.3)
That’s a hell of a lot of change, considering there are a bunch of guys (Ryan Malone, Andrew Brunette, etc.) that consistently score in the 15%+ range.
It’s really too bad that Loui Eriksson’s contract ends next season and not this one. I think Brett Hull would be all over that.
I keep meaning to write something about assists. Bizarre stat, kind of the RBI of hockey. I’m not even sure that they tell us much, except at the extremes. Guys like Thornton or Sergei Berezin, they probably have legit information contained in their assist rates. Other guys…I’m not so sure.
Brownlee got me thinking the other day about the value of second assists; I’d be willing to bet that for defensemen they’re as good an indicator as any of ability to clear the defensive zone.
Of course I don’t know that, I’m just guessing, but I’d be very surprised if they weren’t (with a few exceptions for lucky years, etc.). All of the good puck-moving defensemen should consistently rank high in that category.
Am I missing something here? Vast majority of guys within +/- 2 shots/60 year over year. 900 ES minutes is 15 ES hours. Guy with 10 ESS/60 has 150 ES shots. +/- 2 shots is plus/minus 30 shots. At a fixed 10% shooting rate, we’re talking about +/- 3 goals, no?
Sorry, I didn’t realize you were assuming shooting percentage constant. My figures were assuming shots & icetime were constant, and shooting percentage was the result of random variations. In that case you are probably correct.
If you figure that the top 25 are the real top 25 shooters (which they aren’t) and that their true shot% is 11.3 (which it isn’t), that’s 20% better than average. If shot counts are even both ways, over the long haul they should score 20% more than their opposition.
Now take the top 25 in corsi from BTN. Pretend that both they and their opponents are average shooters. Looking at their average shot differential, they should score 15% more than their opposition.
Now this was really (*really*) quick & dirty. A better way to get a handle on the value of shooting% versus corsi would be to look at how much of an improvement you get in goal differential from a guy one SD above average in each category. If we’re not looking at a normal distribution there is no SD, but there must be some other statistical measure that could be used.
Tyler,
“I keep meaning to write something about assists. Bizarre stat, kind of the RBI of hockey.”
IMO goals may not be far off that mark either. When I was watching the Caps-Rags series, I couldn’t help but think of that Michael Lewis article about Shane Battier where he makes the point that in basketball, unlike baseball, a player can help himself while not necessarily helping his team optimally. I think hockey is similar, and it would not surprise me if we eventually find out that a player like Ovechkin is kinda overrated. (Qualification: “overrated” meaning he might be like a top 20 player instead of top 5.)
To my eye, Ovechkin takes a lot of opportunities (which happen to help his individual layman stats) that could be passed up for better opportunities (i.e. better for his team but worse for his individual stats, at least the ones people currently look at). After all, ultimately the measuring stick of a player’s value is how much he adds to his team’s expected goal ratio over another player.
For evidence that’s a little more objective/empirical re: Ovechkin, dig the fact that Ovie’s ES on-ice GF/60 was 3.72 and GA/60 was 2.91 for a ratio of 1.28 this season. Datsyuk’s was 4.28 and 2.30 for a ratio of 1.86. I realize those numbers are influenced by teammates/goaltenders/opposition, but I doubt the differences in those factors between the two players are nearly significant enough to explain the VERY significant difference in those two on-ice ratios. But how many people would say Datsyuk is more valuable than Ovie, let alone significantly more valuable. (Note that Datsyuk had 18 ES goals while Ovie had 33.)
Tyler,
I’m not saying that it isn’t important, it obviously is, just that form a team results point of view, we’d be more concerned with how they impacted the other players on the ice. Personally, I think this post, and the last one, are terrific and relevant.
So Pavel Bure, to use an extreme example, had a high shooting%, and got a lot of shots at evens. But when he was on the ice his teammates didn’t get that many chances, and the opposition got a lot.
And the Sedins, #22 must be the winger, because he had 194 shots on goal at evens, #33 had 108. Granted Vigneault is a button down coach, on a Nolan team they are probably 160 and 145 shots each, or thereabouts.
From a Fantasy League POV, that’s extremely important to know.
From a team POV knowing that 54% of the shots directed at net happened at the opponents end when the Sedins played at evens … that tells you that that very likely 54% of the scoring chances happened at that end as well. And the scoring chance rate, relative to total shots directed at net, will be near enough the same for everyone down the roster.
So a line of in-their-primes Malone/Comrie/Satan may well score as many goals as the Sedins, in a parallel universe where they took all the Sedin’s shifts, because all three have historically shown an ability to finish their chances at evens. But they would have a corsi rate of about 45%, and scoring chance rate the same, 45%, and therefore are for more likely to have been on the ice for way more goals against.
And perhaps less obvious, and this is stealing directly from Nielsen or Keenan via Lonard’s paper on the NYR database from 20 years ago … the shift after the Sedins is going to be much better that the shift after Comrie’s line, and even two shifts after there is a real effect. Call it momentum if you want, but scoring chances beget scoring chances, long after you’ve left the ice. Dennis’ scoring chances bear that out in compelling fashion.
I guess my point is that on-ice EVshooting% is a fairer comparator to Corsi rates, or zone time rates, or any other measure of territorial advantage (and surely there is a better one than Corsi, we just aren’t aware of it yet).
And while historic on-ice Evshooting% is hugely important, given my druthers, I’d take the guy expected to get the same results by historically driving possession, because he gives less back the other way, and sets up the shift after himself to have more success.
So, if I were owner of the Wild, I would hire a GM based entirely on their relationship with the Sedin twins.
Sunny:
New Orleans? How does a someone from Louisiana become a hockey fan?
And excellent point on Ovechkin/Datsyuk. Though weeding out the context is an enormous challenge, of course. And Ovechkin starts in the offensive zone a lot, and against tired legs a lot, and with the puck headed the right way a lot. Against DET he will probably play the lion’s share of ice time against DET’s best players, but in favourable circumstance. Much tougher icetime for Zetterberg, Lidstrom and/or Dastyuk than it is for Ovechkin … if that makes any sense.
Ovechkin is still a terrific player, of course. Just not as good as the counting stats imply IMO, or the underlying numbers either.
On basketball, Jeff Sagarin once talked about all the little things that a basketball player can do to help his team win, he listed off a bunch of things that I don’t remember, because I don’t know basketball well enough that they resonated with me. Anyway, he lumped these things together in a category called ‘nuance’. And he felt that the only sensible way to capture that was to look at the overall results when the player was on the court.
The thinking being that if a team consistently scored more when Player “X” was on the court, even though he wasn’t showing up on the scoresheet, teammates and opponents considered. Then he was helping you win, and was probably more affordable relative to his value to the team, because NBA salaries are based largely on the counting stats.
He never considers the individual player, just a thousands of 5-man units (the team at any given moment is considered on 5-headed beast). Then he divorces the individuals from the beast after the fact, trying to measure their individual effect on the ability of the beast to outscore opposition.
Granted, there are many, many more ’shifts’ in hockey. This would be an onerous task for the NHL. Hell, I’m sure that it’s an onerous task for the NBA as well.
JeffJ:
If you’re point is that guys with a proven ability to finish, and to help their linemates do the same … that for those guys’ success is more dependent on their EVshooting% and EVon-ice shooting%, I agree. And there are a lot of players like that, good players. Plus those guys are often the ones that make powerplays tick.
Often small players as well (see Montreal, Minnesota and Buffalo since the lockout). You’d better have good special teams and good goaltending though, and sooner or later the territorial advantage will bite you. Still, there’s more than one way to skin a cat.
It seems to me though (based on nothing other than intuition) that you pay nearly as much in salary for a guy that creates his offense by influencing on-ice% as a guy that influences on-ice territorial advantage. And it seems to me that the latter is a lot more likely to help your team win. Which is a big part of the reason I liked the Oilers heading into 2005 but have never been that enamoured with the Oilers rosters since the summer of 2006. They looked like teams that could finish, but the October rosters never looked likely to have the majority of possession. (Same goes for you Habs rosters btw, maybe even moreso, but they play in a weaker conference and have had terrific PPs, so they got away with it).
Javageek:
Jim Albert separates the ability from the luck by assuming that ability takes the form of a beat distribution (without explanation) though a google shows that he is a Bayesian mathematician by profession, so I suspect there is some sort of pure mathematical reasoning there somewhere.
Then he solves for the shape of the ability distribution by solving for ‘k’ in the equation:
g( p) ∝ p^(Kη −1) * (1− p)^(K(1−η )−1)
Where eta is the average ability of all players, and he solves for kappa somehow, I assume by trial and error until he has a best fit.
Clearly it is some deviant form of the binomial distribution, I don’t follow how he gets there though. But the proof is in the pudding, if you take his results for 2003 (from Curveball) for ‘ability distribution of home run power, relative to balls put in play’, he arrives at K=70.
You can plot that out, build a list of player abilities for guys with over 300 at bats, then run simulated seasons assuming that the chance of a ball in play going out of the yard is random (binomial) … then the result is near as dammit the same as actual. The squiggliness in results follows the squiggliness of that seasons randomness, but you would be hard pressed to pick the real season from one of the simulated. And it’s heavily skewed to the left with a thick tail dragging out to the right trying to reach Barry Bonds. Which is the sort of distribution Tyler is sensing for EVshooting% in the NHL, if I have understood him properly.
Does that make sense, and have you done anything similar for hockey stats?
That should read ‘beta’ distribution.
And one more thing and then I’m out, I promise.
The vast majority of the baseball stats stuff I have read seems to be working on the assumption that ability is distributed normally (Guassian). This seems unlikely to me, for any ability in any sport.
Though I don’t think that there is anything magic about the Beta distribution either. It just happens to be closer, nature is still leading us around by the nose here.
If I had the programming chops, I’d write a website where someone who had kicked at a stat (say Tyler with EVshooting% or on-ice EVshooting% could go to a webpage and build a curve of his best guess at how the ability is distributed aongst NHL regulars, just by dragging a few points around, the program would smooth the lines.
Then he would click enter, and if we worked on the assumption that ‘when they actually go in’ is near enough random (and Tyler’s rolling average streak data suggest that strongly, granted he just used four players) then the program would simulate a few parallel universe seasons of results and plot them against the actual.
Then he could tweak the ‘ability’ curve and try again until he got something that worked best, just by eye.
And if the same ability curve worked best for the seasons before and after (assuming no major rule changes, all the common sense stuff) then I think we could be reasonably confident that he had it sussed.
And from there we would be able to get a much better idea of the probability of “Cogliano actually being a natural 13% shooter who just had a bit of good fortune” and so on. Because we can’t do that sensibly without knowing the behavior of the whole population.
Makes sense, no?
Ovechkin is still a terrific player, of course. Just not as good as the counting stats imply IMO, or the underlying numbers either.
600+ shot attempts at even strength. His linemates probably suffer for this, and the players who shift after him, aswell. But I can’t even imagine this player at the age of 27, 28, 29…
If he was drafted in 1979 he would have broke 100 goals in 1982.
Yeah, and a lot of those are good shots, too. He shoots from a bit further out than a lot of guys, but he has such a quick release and hard shot that he can.
I remember back in about January, Moreau had a game with a tonne of shots on goal and the commentator said that he was ‘just firing shots at every opportunity, like Ovechkin’. Moreau WAS firing shots from everywhere, a lot of those were from sharp angles, outside the faceoff dots. A few missed wide and whistled around the boards back out of the offensive zone.
After that I started watching to see if Ovechkin fired them from bad angles, and he doesn’t. Of all the highlight reel goals and saves I’ve seen, only one was from outside the dots. And I only saw one shot from outside the dots in the playoffs as well, it was saved by Lundqvist in Game 2. So I don’t think the guy is wasting many. Granted, Boudreau gets him over the boards on the fly a lot when the puck is heading into the offensive zone, that helps his shot totals. Plus he’s really, really good.
I wonder how many goals this guy could have scored in 1985? Knowing how many soft goals (by today’s goalie standards) were scored back then. Obviously he’d lose the composite stick in time travel, that would slow his release, maybe the shot speed too. Still, I agree, surely he’d have way over 100 goals in a few different years.
Looking back at those 80s stats, I’m impressed at how many shots Gretzky took. An absolute tonne. My memory shows me Gretzky passing most of the time, but hockey-reference.com shows that even Kurri was a distant second in shots most years.
@Sunny: Yeah, I can definitely see that as a possibility, although 8 is such a player that the circumstances in which it makes sense for him to be the shooter are probably wider than the normal player because of his ridiculous gifts.
I do think it’s a valid point though, along with Vic’s Sagarin discussion that followed. I’m going to be really interested to see the Hart voting this year, although I don’t really care about the awards. I’m curious to see where Crosby ends up. If you pull out the second assists, he actually finishes second in scoring, ahead of Malkin but behind OV. Just no second assists, to the point that it’s a little weird - like 24 firsts and 4 seconds at ES or something. Pens were better offensively with him on the ice too, IIRC.
Hard to believe, given that we’ve all known about him since he was 13 but he might be underrated, particularly compared to Ocho. The standards seem to be different for him, at the very least.
On that not, he’s been a consistently better ES scorer than Ovie. Ovechkin had a bit of a good year on the PP and Crosby had a (relatively) off one. If I was betting on next year’s Art Ross now, 87 would probably be my pick.
Ovechkin is still a terrific player, of course. Just not as good as the counting stats imply IMO, or the underlying numbers either.
600+ shot attempts at even strength. His linemates probably suffer for this, and the players who shift after him, aswell. But I can’t even imagine this player at the age of 27, 28, 29…
After Monday’s game, are y’all still up with this idea that his counting stats are misleading? I think the idea that his linemates are “suffering” because of his puck possession ignores a very important facet of hockey: chaos. Ovechkin creates it, constantly. He unhinges defenses not by slick passing (although he can make very slick passes, something he doesn’t get credit for), but by being the bull in the china shop. In a way that Crosby/Malkin/Datsyuk/Zetterberg/whoever cannot do. It doesn’t count as an assist, but Ovechkin is still the reason the goal was possible.
I still don’t know whether 8 or 87 is the better player, but I do know I would take either over Datsyuk in a heartbeat. I have great respect for Datsyuk, who is an excellent, excellent centerman. But I think people spend a lot of time trying to manufacture criticisms of Crosby and Ovechkin, and in the end, those arguments still don’t hold water.
I doubt the differences in those factors between the two players are nearly significant enough to explain the VERY significant difference in those two on-ice ratios.
I don’t doubt it. Detroit’s defensive corps is miles ahead of Washington’s as a defensive unit, and Ovechkin’s GA as cited comes in front of Jose Theodore. I think you’ll find that the GF/GA ratio for Ovechkin in future years, as he matures, his team matures, and Varlamov settles in to the long-term starter’s role, will surpass what Datsyuk is doing right now.
After Monday’s game, are y’all still up with this idea that his counting stats are misleading? I think the idea that his linemates are “suffering” because of his puck possession ignores a very important facet of hockey: chaos. Ovechkin creates it, constantly. He unhinges defenses not by slick passing (although he can make very slick passes, something he doesn’t get credit for), but by being the bull in the china shop. In a way that Crosby/Malkin/Datsyuk/Zetterberg/whoever cannot do. It doesn’t count as an assist, but Ovechkin is still the reason the goal was possible.
That was only one game. I mean, Datsyuk had a hell of a game last night too. Even better than Ovechkin’s, in the things that matter to me (effective backcheck, good puck movement forward, and doing it all against a very tough Anaheim checking line). It didn’t lead to any points but that’s more a fault of his teammates - for example, twice on the same shift Datsyuk sprung Franzen with a clear path to the net, and he made an unscreened slapshot from the top of the faceoff circle both times…
The point is, Datsyuk drives possession against tough competition in a way that Ovechkin and Crosby can only dream of (with Malkin not even being in the same conversation). That’s measurable, and the difference is significant. And he’s doing it in front of Osgood and Conklin - not exactly contenders for the Vezina. Zetterberg does the same. I’d take either of these guys over Ovechkin, Crosby or Malkin.
Fair enough. Backchecking and puck movement are important to me, too. But I still don’t think the numbers, or the observations (and obviously not just from one game, that’s just illustrative), are nearly as clear a proof as claimed. Ovechkin and Crosby *are* driving possession against tough competition, but by different means and in a different team environment, and it’s not (IMO) a differential they “can only dream of” equaling.
Another year or two of data as Detroit, Pittsburgh, and Washington evolve ought to clarify things. Until then, we may have a tough time making a dispositive case either way.
I keep meaning to write something about assists. Bizarre stat, kind of the RBI of hockey. I’m not even sure that they tell us much, except at the extremes. Guys like Thornton or Sergei Berezin, they probably have legit information contained in their assist rates. Other guys…I’m not so sure.
What exactly do you mean by this? In one sense, I think it’s obvious that assists give information about a player’s offensive production. Some players create opportunities for themselves more often and some create opportunities for others more often; the second group will have their offensive contributions show up as assists.
Do you mean that assists don’t tell us much beyond on-ice goals, and they are mostly a function of how many goals a player was on the ice for and didn’t score himself? That seems more reasonable to me.
I’m more open to that idea at even strength, as it seems like forwards are often slotted into shooting roles or passing roles on the power play, and some players really seem to have an “assist ability” on the power play.
I would agree that 2nd assists, at least, give no information beyond on-ice goals for. I have the assists for the 1980s broken down into 1st and 2nd assists, and Gretzky’s huge assists totals were almost all 1st assists. His 2nd assist totals, for the most part, were the same as other top forwards. For example, in 1983-84 he had 90 1st assists and 28 2nd assists. Hawerchuk was 2nd in 1st assists with 51, and Trottier led forwards in 1st assists with 33. I think 2nd assists could be lumped in with the other on-ice goals for that the player didn’t score or directly assist without losing any information, at least at even-strength.
To my eye, Ovechkin takes a lot of opportunities (which happen to help his individual layman stats) that could be passed up for better opportunities (i.e. better for his team but worse for his individual stats, at least the ones people currently look at). After all, ultimately the measuring stick of a player’s value is how much he adds to his team’s expected goal ratio over another player.
I wouldn’t penalize Ovechkin too much for the number of shots he takes, as he creates a lot of those chances himself and converts them at a reasonable percentage. On the other hand, it’s true that he doesn’t set up as many goals for his teammates as he could, but including his assists and on-ice goals for will solve that. So I guess I agree with you, I just don’t think that Ovechkin is a “selfish player”.
Incidentally, someone like Brett Hull would be the classic “selfish player” who was scoring a lot of goals and not really helping his team win. Over his career, his plus-minus was basically equal to his teams. That’s not surprising when you consider that he
1. Wasn’t known for his defensive play.
2. Took a lot of shots, most of which came from opportunities that had to be created by his teammates.
3. Had a good shooting percentage, but not at a Bossy/Kurri level.
4. Didn’t set up a lot of goals.
If someone tracked possessions for his teams, I bet they would find that he didn’t start many of them, and a lot of them ended with him. That meant that he basically had to score 50 goals just to break even. Anyway, that’s a bit of a digression but I think Brett Hull is the profile of the classic “goal leech”, and Ovechkin doesn’t match in that he creates many of his opportunities on his own.
I would agree that 2nd assists, at least, give no information beyond on-ice goals for.
Overpass: It has always bugged me w.r.t. Assists that the second guy gets just as much credit as the first, and the third gets none at all. Glad to see folks like Gabe Desjardins begin to pull apart the info on first and second assists, even as the third and subsequent are beyond reach except for on-ice goals for.
That said, I think second assists tell us a little more than that. Bryan Trottier made a lot of fine passes among those second assists and did his share in creating the goals that resulted. Ditto Gretzky’s second assists, many of which were the result of him setting up a 2-on-1 situation which needed one more pass to finish the job. We all know true three-way passing plays are not uncommon, and all three guys deserve more or less equal credit on at least some occasions. Occasionally more than three.
That said, while each goal has its own merits, I would suggest a general rule governing a large sample size, is that contribution to a goal starts with the guy who touched the puck last (the goal scorer) and diminishes from there. An imaginary scoring race awarding points on a 3-2-1 basis would likely do a better job of identifying the dominant offensive talents than the official 1-1-1 system.
Btw, Overpass, I’d be keen to see that ASST1/ASST2 data for the 80s, which remains an era of great personal interest. Have you published it anywhere?
Bruce: I don’t mean to say that second assists have zero value, just that I think they are closer in value to on-ice goals where the player didn’t have a goal or assist than they are to goals or first assists. I agree that a weighted scoring system with different weights for goals, 1st assists, and 2nd assists would be good. I also think that type of system doesn’t need to stop at second assists, as you could give credit for on-ice goals that weren’t goals or assists also.
Regarding the data, no, I haven’t published it, but I’d be happy to send it to you. Do you have an email I can send it to? I checked your blog but didn’t see a way to contact you.
Maybe it’s just me noticing it in spots, but I’m very skeptical of the primary data w.r.t. 1st and 2nd assists.
I’ve never looked for it or studied it, but many times I’ve checked a Flames boxscore after a game and noticed offhand the two in the “wrong” order on a given goal. Has anyone ever assessed this?
I mean, I assume it’s subject to the same error as every stat (sometimes the wrong players are given a plus or minus on a goal, e.g.), but do the scorers even care which name they put first? Is it reviewed? I need a ruling.
Matt - I wouldn’t be surprised at all if the primary data on 1st and 2nd assists was wrong at times, although I don’t know one way or the other. I can only say that looking at the aggregate data player by player, it looks right. Forwards get more first assists and defencemen get more second assists.
@Matt - As far as I know, they do it in the order that the assists happened. I’m out on the road at the moment - Calgary, actually - complete aside but it should be legal for me to kill the Calgary douchebag who put a big scuff on my rental car when he kicked it after I almost hit him as he jaywalked across the road, dressed in black at midnight - but it’s always seemed to me that the Oilers guy, at least, gets it about right. As overpass said, the breakdown between first assists and second assists as between D and F seems to make sense.
@Overpass: I kind of see assists as being like RBI. RBI have been shown to be a function of a player’s batting skill and the opportunities presented to him. I kind of see assists as being the same, with maybe more emphasis on the opportunities than is commonly thought, except at the extremes, guys like Thornton or Berezin. Havren’t bothered trying to prove it yet.
Do you have an email I can send it to? I checked your blog but didn’t see a way to contact you.
Hi Overpass, just checking back in now and saw your question. If you happen to come back, I have added an email contact on my blog. Good suggestion (I think).
As for recording errors in assist order, I seem to recall that in 1979-80 they used to frequently announce Gretzky’s assist second (”… and number 99 …”) to get the bigger roar from the crowd, regardless of what order it happened in. Seems to me at one time the league issued an edict about that, not specific to the Oilers, just generally that it must be done in the right order. My memory is pretty vague on when that happened, but I have not noticed misordered assists anywhere for a very long time. Phantom assists, maybe.