Friday, November 18, 2011

The Opposite of Adults

[Note: The Q(ual)C(omp)R(ank) column shows the position on the qualcomp list the player occupies on his own team. So 1 means the guy faces the hardest competition on his team, 17 (generally) the easiest competition. I limited everything to players with at least ten games on the season.]

Charts are beautiful things: so reasonable, so tidy, with all their neat straight lines and regularly proportioned boxes. Looking at a good chart, one can forget all about postmodernism altogether and drift happily along in a rational universe, buoyed by the sheer sensibility of everything. And sometimes, if you stare at them long enough, they start to tell you things.

This chart, for example, gives you a basic typology of the talent-developing practices of contemporary NHL coaches. Their philosophy of child-rearing, if you will. You have your neglectful dads, who just toss the kids down on the fourth line with the goons, play them eight fairly soft minutes a night and hope they learn something along the way. You have your warm, fuzzy, doting dads, the Mr. Mom types, who give their precious little boys everything they could ever ask for, stuff them full of offensive zone time and weak opposition until they fairly burst with goals. And then there’s the gruff, old-school, trial-by-fire dads, who just throw ‘em in the deep end and figure them as don’t drown are good enough to keep.

I’m messing around with a longitudinal project on young players and development techniques, the relationship between methods of use and trajectories of production, etc etc. Will anything substantive ever come of it? Given that my lifetime ratio of intended projects to realized projects is about 9323:1, probably not, so I’ve decided to post some of the bits and pieces along with way for contemplation and comment. For example, here are some comments:

1. Who exactly is Roman Horak and why are the Flames trying to kill him? Is this some kind of Calgarian ritual of sacrifice used to revivifying the skating corpse of Jarome Iginla with the blood of the young?

2. Tom Renney does not love all his children equally.

3. As much as PDO still creeps me out, the juxtaposition of Tyler Seguin and Erik Gudbrandson’s respective output/input is pretty persuasive. Now I just need somebody to tell me the PDO for my life.

4. If the Avalanche fuck up these kids, Quebec City will have a moral obligation to send a covert pack of slinky bombshells with sexy accents to seduce, drug, and recapture the team, because Colorado will have proven itself unworthy. Duchene is the only one getting anything close to any kind of shelter and they’re still getting tremendous results.

5. Unless I am missing something, Jared Cowan does not look ready for prime time.

6. One of my big questions going forward: what, exactly, are the virtues of the eight-minutes-per-sixty strategy? Couturier is the only one who’s doing well with it, and the smart money says it’s gonna get harder for him before it gets easier. If you’ve got a kid who you think is worth putting on an NHL team before he’s legal to drink in most NHL cities, it must be because you think he can contribute something awesome now, because cap management, ELC, blah blah blah. But nobody is going to contribute anything awesome lining up with George Parros two shifts a period. Wouldn’t anyone whose long-term projections are high-end benefit more from playing on the high end of a lower tier team than from warming the bench in the NHL? Hockey skill is not transferable by butt-chair osmosis, otherwise I would have learned something from Vincent Lecavalier. The fourth line is useful as a final stop for mid-range NHL guys aging out of their skills and as a prize for the hardest of the hard-working AHLers, but I can’t see how it helps development.

7. In data form, one Larsson looks much like another. Perhaps it was the same Lars.

8. With the remarkable exception of Colorado, every other team gives their young players at least a slight degree of protection from the top competition. In some cases, like Edmonton, I suspect the mid-range QualComp rankings are the result of an average between being very sheltered at home and deliberately targeted on the road. The kids who are bottoming out the QualComp on their team are mostly the 8-minute crowd, which goes beyond any deliberate protective intent by their coach- it suggests to me that coaches like worst-on-worst action in any rink.

[Please note that I am interested only in usage and performance thus far. This information is not necessarily predictive- luck, small sample size, ‘choice’, whatever- and I’ll be revisiting all this down the road to see how it pans out over the whole season.]

All data courtesy of


Clare said...

The real question that stuck out to me is--when did your butt meet the same chair as Vinny's butt and could you feel the awesomeness percolating up through your bones? :)

(Note: I have nothing useful to say because I don't really understand what those numbers are supposed to tell me. I plan to work on that. When I get time.)

E said...

movie theater, montreal, 2007. someday i'll write up the full story, although it isn't really much of a story.

as to explaining the numbers, though, you're probably not the only one confused, so here's a quick summary:

GP- games played (self-explanatory)

P- points (also self-explanatory)

TOI/60- the number of minutes a player averages per sixty minutes his team plays at even strength (yes, i know, you can't play 17:75 because there are only sixty seconds in a minute, so i'm just going to assume that means 17 and 3/4 minutes, or 17:45. but weird, anyway).

P/60- the number of points a player averages per sixty even-strength minutes he plays.

ZSO%- the percentage of a player's zone starts are in the offensive zone.

ZFO%- the percentage of a player's zone finishes are in the offensive zone (i'm pretty sure the two zone metrics only count face-offs for starts and whistles for finishes, so changes on the fly aren't recorded)

QCR- Quality of Competition Ranking (explained in the post)

CorsiRel- Corsi +/-, basically- the number of shots for (including blocked shots and missed shots)his team got when the guy was on the ice, minus the number of shots/missed shots/blocked shots against.

PDO- on-ice shooting percentage plus on-ice save percentage. basically a measure of luck, because it tends to regress for all players towards 1000 the more minutes they play. so a guy with a PDO under 1000 is experiencing bad luck and likely to see his production improve, whereas a guy with a PDO over 1000 is experiencing good luck and likely to see his production drop.

so, simplified, TOI/60, QualComp Rank, and ZSO% measure inputs- those are the things that the coach is giving the player, in terms of ice time, ease/difficulty of opposition, and offensive opportunities. p/60, CorsiRel, and ZFO% measure outputs- what the guy is doing with what he's given. PDO measures luck.

so, for example, ryan n-h is getting treated pretty generously, but is producing pretty well with it, and (probably) can sustain that kind of production with that kind of treatment. jared cowan is also getting treated generously, but is producing poorly- he's been a bit unlucky and will probably improve somewhat, but not necessarily enough to justify the amount of sweet soft time he's getting. on the other end, ryan o'reilly is playing significantly harder opposition and starting a lot more shifts in the defensive zone and STILL kicking ass in production, and that even with a low PDO, which looks very promising (small sample size, yes yes yes, i know).

saskhab said...

The thing is, the Bruins PDO numbers should generally be above average, they constantly are, because their goalies are quite good and stop more shots than the league average. PDO does show luck over a career, but year to year fluctuations can sometimes be explained by the fact that they played in front of great goalies.

When your starter has been voted the best goalie in the league the past 2 years and your backup appears to be a legit top 10 guy on his own, high PDO's are quite normal. Those bastards.

E said...

that actually raises a big question for me, which is how wide is the range of a 'sustainable' PDO? i've been doing quite a bit of looking through the explanatory/discussory posts, and everyone says that it regresses strongly towards 1000, generally within the number of minutes a relatively healthy player would skate in a single season, but obviously most players still end up a bit below or above. but the range of PDO is not that big to begin with, so what's the range of 'normal' PDO? is it more like 990-1010 or 950-1050? i suspect the latter, based on the way people talk about it, but i'm not 100% sure.

either way, seguin's still got the 2nd highest PDO on his own team, and indeed the third highest in the league, so if that shit ain't due for some regression than nothing is. since 2007-8 (as far back as behindthenet goes), nobody who's played more than 30 games has finished the season with a PDO of 1100 or better.

Doogie2K said...

Well, if you're someone who historically shoots well above average - a Crosby, or even a Tanguay - you're going to have a higher expected PDO. Ditto if you play in front of a goalie who's historically well above average - Tummy Thomas being the easy example. But over the course of a career, or over the NHL as a whole, everyone will average out to 1000.

As for an expected range about said expected PDO? Fuck if I know for sure - ask one of the numbers guys - but generally it seems to be closer to +/- 10 or 15 than +/- 30 or 50.

Word verification: curitylo, which looks like something entirely different in this font.