• Moderators, please send me a PM if you are unable to access mod permissions. Thanks, Habsy.

Advanced Stats Thread

Zeke, you holding out on us?

Saw this on HFboards. Great post.

Originally Posted by zeke
they have basically taken a stat like fenwick which actually has a fairly weak correlation with future wins, and are treating it as if it is a much stronger predictor than it actually is. It's correlation with future wins isn't strong enough to suggest its a a clear "cause" separated from all significant "effects", yet they still treat it as if the numbers have shown that we can actual model predictions with fenwick as the "reality", and factors such as sh% and sv% as just luck or small sample variation factors.

fenwick and corsi are nice stats that tell us something, but the conclusions being drawn from them are far too strong to be justified.

and this year is ESPECIALLY glaring because they are insanely drawing these conclusions from a the tiny half-season sample we had last year. even a full season is a small sample for the advanced stats with a weak correlation to future wins - a half-season sample is even weaker by orders of magnitude.

to give a layman's example, let's look at, say, Bozak.

Bozak Career:

Career: .23gpg, .56ppg
12/13: .26ppg, .61ppg
11/12: .25ppg, .64ppg
10/11: .18gpg, .39ppg
09/10: .22ppg, .73ppg

What we see there is a fairly consistent producer over his career. A guy with a hot rookie year, a bit of a sophomore slump, who then settled in between there the next two years, just a tad above his career average.

We have a solid track record to work off of here, which would give us very good reason to predict him to come in again around .25ppg/.60ppg level this year.

But an "Analytics Guy" is not allowed to draw this fairly simple conclusion, because last year Bozak had a 19.7sh%, an 11.1on-ice sh%, and a 1027 pdo. The "analytics guy" will be forced to claim that Bozak is very likely to suffer significant regression this year from last year's production - despite the fact that that would mean Bozak would "regress" to a level well below his relatively consistent career norms - career norms that were established before last year, even with less "lucky" advanced statistics.

So for an Analytics Guy, Bozak overachieved last year and is due for regression this year - even though Bozak simply continued on scoring at his typical career level last year.

And in fact, when you actually look at all the leafs' career production, weighted for recentness and adjusted for minutes played, it's very hard to argue that many leafs overachieved offensively last year, despite what the advanced stats say. About the only two leafs who played a significant amount of games that could legitimately be said to overachieve were kadri and franson - but even then, those were young guys gettting their first legit offensive opportunities, so their lack of track record is hard to hold against them. And moreover, a number of players, either due to poor performance or injury, clearly underachieved last year to balance those guys out - Lupul, Grabo, gardiner in particular.

What we've ended up with is having so many people with a fairly poor understanding of statistics, who have developed unjustified tunnel vision when it comes to these possession stats - they have literally convinced themselves that last year's half-season sample of possession stats has to be the "base" of their analysis, and are actually completely disregarding contrary facts like the Bozak example I showed above. For them, it doesn't matter that Bozak scored at his usual rate last year - he has to be due regression, because the advanced stats from a half-season sample size say that he was lucky last year.

They're making a big mistake, IMO.

But hey, maybe I'm wrong.
Click to expand...
 
Using Corsi/Fenwick to predict the standings is like using SV% or GF in isolation to the predict the standings. It's just one piece (stat) of a very complex puzzle.
 
hockeylover said:
Mirtle only ever cites SV% at ES though, which was 10.7 % last year and 8.8% this year.
Click to expand...

And the difference in the drop has been made up by an increased amount of shots, at least so far.

I don't think many people expected the Leafs to maintain their shooting % from last year, I slight drop is to be expected.
 
PlayerToBeNamedLater said:
And the difference in the drop has been made up by an increased amount of shots, at least so far.

I don't think many people expected the Leafs to maintain their shooting % from last year, I slight drop is to be expected.
Click to expand...

Sure, I'd agree, I was just trying to figure out where your numbers were from...
 
One thing to look at also is, the Leafs are good at drawing penalties. They generated the 10th most PP chances last year. They were also shorthanded the 13th fewest times last year. Good at drawing PP's and good at not giving up PP's. They spent a lot more time up a man, then they did down man. They've continued this success this season, so far.

The Leafs can offset some of their deficiencies at ES (or expected decline in shooting % at ES) by spending less time at ES and by outproducing their opponent on special teams.
 
It's still early but a quick look at that site I posted seems to suggest some good returns from some of the new guys in terms of their corsi/fenwick. Bolland specifically. One thing about the advanced stats guys is that they don't seem to look very much at any additions made, they just take last years numbers and seem dismissive about any changes we've made to that club.
 
Bolland's haven't been good, but it's tough to really evaluate him because his strength of comp has been so strong typically.

Raymond's been about average.
 
I hate the name "advanced stats" because totalling up shot attempts isn't exactly "advanced" stuff.

here's another new stat, that's not really "advanced", but is a simple and excellent idea.

Much like OPS+ and wRC+ in baseball, which adjusts these stats to league average of that season (i.e. 100 = league average)....may I introduce to you....


.....SV%+


http://www.hockey-reference.com/leagues/NHL_2013_goalies.html#stats::15



an excellent much needed stat given the changes in sv% over the years.

we now know, for example, that Reimer's .924sv% (115sv%+) last year was not quite as impressive as Belfour's .922 in '03 (117sv%+).
 
zeke said:
I hate the name "advanced stats" because totalling up shot attempts isn't exactly "advanced" stuff.

here's another new stat, that's not really "advanced", but is a simple and excellent idea.

Much like OPS+ and wRC+ in baseball, which adjusts these stats to league average of that season (i.e. 100 = league average)....may I introduce to you....


.....SV%+


http://www.hockey-reference.com/leagues/NHL_2013_goalies.html#stats::15



an excellent much needed stat given the changes in sv% over the years.

we now know, for example, that Reimer's .924sv% (115sv%+) last year was not quite as impressive as Belfour's .922 in '03 (117sv%+).
Click to expand...

At first blush, I really like this stat. Really, really like.
 
zeke said:
I hate the name "advanced stats" because totalling up shot attempts isn't exactly "advanced" stuff.

here's another new stat, that's not really "advanced", but is a simple and excellent idea.

Much like OPS+ and wRC+ in baseball, which adjusts these stats to league average of that season (i.e. 100 = league average)....may I introduce to you....


.....SV%+


http://www.hockey-reference.com/leagues/NHL_2013_goalies.html#stats::15



an excellent much needed stat given the changes in sv% over the years.

we now know, for example, that Reimer's .924sv% (115sv%+) last year was not quite as impressive as Belfour's .922 in '03 (117sv%+).
Click to expand...

Fantastic.

ERA+ and OPS+ are probably the two best metrics for evaluating individual performance in baseball. To have a statistic at our disposal that averages out factors to create a league aggregate of performance could prove to be incredibly useful. Not sure if there's ever been a comparable state derived for hockey to be honest.
 
http://www.grantland.com/story/_/id/9799967/toronto-maple-leafs-advanced-stats

Collision Corsi
This year's test case for advanced stats in the NHL? It's the old-school Toronto Maple Leafs, so … great. We're sure everyone will be super chill about this.


It's a battle sports fans come to know well over the years. On one side, you have the relatively recent wave of advanced analytics proponents, using new types of stats and theories to challenge the conventional wisdom. On the other side are the old-school thinkers, many of whom have been around the game since before these stats guys were born and question how much insight can be pulled from a spreadsheet.

Hockey was late to the party, but the debate has arrived here, too. There have been some ugly skirmishes, often flaring up when a particular team's on-ice results didn't match what the numbers predicted. Every year seems to bring at least one example. In 2009-10, it was the Colorado Avalanche. In 2010-11, the Dallas Stars.1

Perhaps most memorably, the 2011-12 Minnesota Wild managed to lead the Western Conference for much of the season's first half, despite horrible underlying numbers. The stats guys predicted doom. Wild fans scoffed. Insults were hurled. And in the end, the stats guys were right. The Wild plummeted all the way to 12th, missed the playoffs by 14 points, and firmly established themselves as the cautionary example for anyone who'd dare roll their eyes at hockey's analytics community.

This year, we've moved into new territory, and the stakes suddenly seem higher. In any battle, real or metaphorical, things don't get really ugly until you threaten sacred ground. And there may be no more sacred ground in the hockey world than the Toronto Maple Leafs........
Click to expand...
 
You must log in or register to reply here.
Back
Top