Article: How the Calgary Flames are Laughing at Advanced Stats

[_Q_]

Originally Posted by CorsiHockeyLeague

This is utter nonsense as an argument, though. Points come from wins, which come from outscoring your opponent. It's essentially tautological to say "goal differential determines playoff seeding"; you might as well go one step further and say 16/16 playoff teams are predicted by their win/loss record. It doesn't "predict" playoff seeding, it describes playoff seeding, because goals for percentage cannot be relied upon to stay the same over time.

Put another way, if asked, "if I want my team to make the playoffs, what should they be good at", the person who understands analytics will say, "it's no guarantee, but the best thing would be for them to be good at possessing the puck". Your response appears to be, "make them good at scoring more goals than the other guys". That's completely unhelpful.

Building a team that has forwards that can score a lot of goals, a defence that can stop the other team from scoring and implementing a system that works for the players you have is exactly how you become good at hockey. Attempting to shoot pucks in the direction of the opponents net more than the opposition may be a symptom of that, but as has been proven this season it may not.

[V]

Originally Posted by Resolute 14

Which explains why mudcrutch became the buttmonkey of the site by penning a treatise called "the value of outscoring: a primer". It was completely unhelpful.

He also became the buttmonkey of the site by claiming that the Oilers were a #1 goalie away from being a contender in 06. I can still remember the ridicule on this site. t99 had it in his sig for quite a while, until they kept on winning.

I do think it's interesting to look at the outscoring primer, though, because at that time there was virtually nothing written on stats outside of the traditional G,A, +/-. It's pretty neat to see where the discussion has evolved to. With new technology hitting the sport I'm really excited to see just what the next 10 years is going to show us.

[Fire of the Phoenix]

The day cannot come fast enough where technology makes these 'proxies' useless. I want real time data to track actual possession, what they were talking about/testing at the ASG has me legitimately excited. This intermediary stage is just pervasive and annoying considering the how loose the link between winning and having a good corsi actually is. How people can be so arrogant and sure of such a metric that is going to be a footnote in history within 10 years is beyond me. Can't go one flipping day as a hockey fan without hearing the words corsi and fenwick, try as I might. It's good to know it won't always be this way as better methods are developed over time.

[CorsiHockeyLeague]

^Until it becomes a footnote in history, it's the best we've got, and it is pretty good at tracking possession. But I agree, the technology cannot come fast enough and I wish they'd implement it starting next season.

Originally Posted by _Q_

Building a team that has forwards that can score a lot of goals, a defence that can stop the other team from scoring and implementing a system that works for the players you have is exactly how you become good at hockey.

Again, if I ask you how to build a team and you tell me, "get players who score a lot of goals and stop the other team from scoring", you are useless to me.

Attempting to shoot pucks in the direction of the opponents net more than the opposition may be a symptom of that, but as has been proven this season it may not.

And every other season. There are teams every year that win in different ways. Possession is the most reliable path of those readily identifiable, and it should in theory be easier to do. Acquiring elite talent players who score on a significantly above average percentage of shots is hard to do, or everyone would have a Steven Stamkos. Acquiring elite goalies who consistently stop a significantly above average percentage of shots is hard to do, or everyone would have a Carey Price. Similarly, coaching guys in such a way as will give you a bunch of high-percentage chances or will bump your goaltender's save percentage up into elite territory isn't easy, or everyone would be coached by Dave Tippett or Ken Hitchcock.

Getting guys who will outpossess the opposition? It's do-able. Coaching a system that leads to more puck possession? Also do-able. And work's always being done to figure out ways to do a better job of it; e.g. the focus on controlled zone entries over dump and chase during the past couple of years.

[EldrickOnIce]

Originally Posted by CorsiHockeyLeague

Getting guys who will outpossess the opposition? It's do-able.

Yes. It's doable. The Oilers did. Its just not necessarily any guarantee of success, obviously.
And yes, you can coach a 'puck control' game if you have the horses. Darryl Sutter didn't become a genius between 2004 and 2014. The 04 Flames were regularly out shot by a 2-1 margin in the playoffs. But they did OK. Fast forward to the Kings, and they can play keep away all day. A coach's game is based on the strengths and weaknesses of his players.
And obviously, outscoring and out chancing the opposition is a greater likelihood of success

[edslunch]

This site was more fun before advanced stats. I don't see accurate tracking making it more fun

[PeteMoss]

Originally Posted by CorsiHockeyLeague

^Until it becomes a footnote in history, it's the best we've got, and it is pretty good at tracking possession. But I agree, the technology cannot come fast enough and I wish they'd implement it starting next season.

Again, if I ask you how to build a team and you tell me, "get players who score a lot of goals and stop the other team from scoring", you are useless to me.

And every other season. There are teams every year that win in different ways. Possession is the most reliable path of those readily identifiable, and it should in theory be easier to do. Acquiring elite talent players who score on a significantly above average percentage of shots is hard to do, or everyone would have a Steven Stamkos. Acquiring elite goalies who consistently stop a significantly above average percentage of shots is hard to do, or everyone would have a Carey Price. Similarly, coaching guys in such a way as will give you a bunch of high-percentage chances or will bump your goaltender's save percentage up into elite territory isn't easy, or everyone would be coached by Dave Tippett or Ken Hitchcock.

Getting guys who will outpossess the opposition? It's do-able. Coaching a system that leads to more puck possession? Also do-able. And work's always being done to figure out ways to do a better job of it; e.g. the focus on controlled zone entries over dump and chase during the past couple of years.

The problem is knowing a guy who outpossesses on team A playing their system and playing with the other players on that team will carry all that over to team B with a different system and different players.

Just to pick a random sample - the top Corsi guy on the Penguins this year is Chris Kunitz. If the Flames decide to use one of their extra picks to trade for the guy, do those possession skills carry over to the Flames? What about guys like Dimitri Jaskin? Jakub Kindl? etc. The Oilers went out and spent boatloads of cash on Benoit Pouliet with his sweet Corsi and he's a pretty middling player for them.

[CorsiHockeyLeague]

Originally Posted by PeteMoss

Just to pick a random sample - the top Corsi guy on the Penguins this year is Chris Kunitz.

Not counting goalies, Kunitz has played 200 or more minutes with eight guys. In five of those eight cases, when Kunitz is on the ice without them, his CF% drops. So you'd have to say he's not the only one moving the river there.

Further, isn't this true of basically all statistics? Say you need goal scoring, and you decide to go out and get, oh, I don't know, consistent 30 goal scorer Bobby Ryan. Is there any guarantee he'll score 30 for you? Ask the Senators.

As for Pouliot, 4M was too much money the instant they spent it (UFA, what else is new). But if everyone in that team's bottom 6 was Benoit Pouliot they'd be in better shape. Lowering the threshold because he's played fewer minutes, take the 9 guys with over 150 minutes with Pouliot. All but one are better, in terms of corsi, with Pouliot than without Pouliot (Nail Yakupov). Petry is the same.

Further, just in terms of pure goals for and against, when Pouliot is on the ice at even strength, the Oilers outscore their opponents to the tune of 50.9% of the goals. He's the only guy on the team who sees the Oilers outscore their opponents when he's on the ice. Overall, they score 39.7%.

[heep223]

Originally Posted by CorsiHockeyLeague

Lowering the threshold because he's played fewer minutes, take the 9 guys with over 150 minutes with Pouliot. All but one are better, in terms of corsi, with Pouliot than without Pouliot (Nail Yakupov).

This means nothing.

Originally Posted by CorsiHockeyLeague

Further, just in terms of pure goals for and against, when Pouliot is on the ice at even strength, the Oilers score 50.9% of the goals. Overall, they score 39.7%.

This, on the other hand, means a great deal.

[CorsiHockeyLeague]

The former is a part of the reason for the latter.

[edslunch]

Originally Posted by PeteMoss

The problem is knowing a guy who outpossesses on team A playing their system and playing with the other players on that team will carry all that over to team B with a different system and different players.



Just to pick a random sample - the top Corsi guy on the Penguins this year is Chris Kunitz. If the Flames decide to use one of their extra picks to trade for the guy, do those possession skills carry over to the Flames? What about guys like Dimitri Jaskin? Jakub Kindl? etc. The Oilers went out and spent boatloads of cash on Benoit Pouliet with his sweet Corsi and he's a pretty middling player for them.

Um, I think that's where actual scouting might be useful vs just crunching numbers

[kevman]

Originally Posted by edslunch

This site was more fun before advanced stats. I don't see accurate tracking making it more fun

It would be neat if the site had a "does not contain" search feature. One could chose to view all threads that do not contain the words corsi and fenwick and go back to enjoying hockey for hockey's sake.

[Vedder]

Originally Posted by PeteMoss

The problem is knowing a guy who outpossesses on team A playing their system and playing with the other players on that team will carry all that over to team B with a different system and different players.

Just to pick a random sample - the top Corsi guy on the Penguins this year is Chris Kunitz. If the Flames decide to use one of their extra picks to trade for the guy, do those possession skills carry over to the Flames? What about guys like Dimitri Jaskin? Jakub Kindl? etc. The Oilers went out and spent boatloads of cash on Benoit Pouliet with his sweet Corsi and he's a pretty middling player for them.

This is a context issue. A couple of guys are tinkering with methods to adjust corsi/fenwick for variables like quality of competition/teammates, icetime, etc, with the intent of creating something more useful than relative corsi/fenwick %. These ideas are from bulletproof, but I found the methodology interesting.

http://nhlnumbers.com/2014/7/19/dcorsi-introductions

http://ownthepuck.com/usageadjustedcorsi/

[Loyal and True]

I know the article makes reference to 10/11 Stars, 11/12 Wild, and 13/14 Leafs and Avs, but I don't think any of those teams are an adequate comparison to this team. Bad possession numbers, granted, but still very different in my opinion.

14/15 Flames
+22 Goal differential
SF/GP 27.6 (28th)
SA/GP 28.9 (9th)
G/GP 2.88 (6th)
GA/GP 2.60 (16th)

Special Teams
PK Shots Against 225 (1st)
PP Shots For 344 (15th)
+119 shot diff (344-225) average +1.61 PP shots/game
+117 minutes PP over PK

This is huge for the Flames. They may give up ground in 5 on 5 possession, but they compensate with high quality shot differential on special teams.

Leading Scorers
F - Hudler 66 pts (projected 73)
F - Gaudreau 58 pts (projected 64)
F - Monahan 56 pts (projected 62)
D - Giordano 48 pts in 61 GP
D - Wideman 47 pts (projected 52)
D - Brodie 40 pts (projected 44)
F - Bouma 31 pts (projected 34)
D - Russell 30 pts (projected 33)
F - Jones 28 pts in 60 GP (projected 32)
F - Backlund 25 pts in 44GP (projected 30)

Look at those numbers for top 4 D-men. Unbelievable. Obviously they are jumping into the play to compensate for the young, developing forward group. If your forward lines can't consistently overpower your opponent in the neutral zone/offensive zone, you can create odd-man opportunities with D-men jumping into the play.

The special teams and contributions from defenceman help turn the tables on a 5-on-5 possession battle.

Other teams

Minnesota, Colorado and Toronto were all brutal at giving up shots on goal. Pretty much bottom of the league. Calgary is nowhere near as bad as those teams defensively. In no way are they apples-to-apples comparables. Eventually, all those shots on goal turn into goals against and they lose a bunch of games.

Dallas is a bit more comparable, but still very different. Dallas started hot but ended 9th in the West standings and -6 goal differential. They still gave up 30.5 shots per game (14th). Their leading scorers were:
Richards (upcoming UFA) 77 pts
Eriksson 73 pts
Ribeiro 71 pts
Morrow 56 pts
Benn 56 pts
Ott 32 pts
Neal 39 pts (traded to Pit for Goligoski in Feb).

On defence they had
Robidas 30 pts
Daley 27 pts
Goligoski 15 pts in 23 GP

Lehtonen 2.55 GAA and .914 SV%

Dallas had a very good top line Richards/Ribeiro but a pretty old group. Defense was nowhere near the Flames top 4 in scoring.

Dallas had a modest special teams advantage due to lots of powerplays, but not as disciplined as the Flames. +47 mins PP time for the year, +50 PP shots (0.61/game)

Overall, Dallas had very little to compensate for the poor possession numbers. Unlike the Flames.

[Red John]

Originally Posted by getbak

Burke doesn't scoff at analytics. He does scoff at calling things like Corsi and Fenwick "analytics" though.

Yep exactly. Burke is actually VERY interested in analytics and has said he'd pay megabucks in cash for a system, algorithm, or anything that actually is proven to work in terms of predicting outcomes. However, nothing so far has actually been useful in achieving this.

As he mentioned, like a lamppost to the drunk. Supports but doesn't illuminate.

It's actually incredibly pathetic how in this "analytics" era that we live in, we still haven't come up with anything better than shot attempts that we refer to as "advanced." Shot attempts. All of which are considered equal.

It speaks to the incredible complexity and fluidity of the game that this is the best we have so far.

As a footnote, was approached by Kent Wilson years back after some of my tracking of zone entries to post this info on Flames Nation. So definitely not an anti-stats guy. Love the idea of exploiting market inefficiencies and finding an edge.

But Corsi? It's incredibly elementary. It doesn't even measure true possession, which would be defined by having the puck. Some players hang onto the puck longer (ie having possession) but cycle more and often don't get a shot away unless they create a perfect chance. Like the Sedins. Then you have morons like Dallas Eakins telling Jordan Eberle to shoot from anywhere anytime because shot attempts = GOOD.

Shot attempts is always going to be flawed until someone figures out how to account for shot quality and that's incredibly difficult to pinpoint. In the meantime, let's save the term "advanced" for stats that actually are.

[Mister Yamoto]

Originally Posted by Red John

.
Shot attempts is always going to be flawed until someone figures out how to account for shot quality and that's incredibly difficult to pinpoint.

+/- does that. It's no where near perfect, but if the puck goes in, that is a quality shot in my books.

+/- is pretty much Corsi where you isolate only shots that find the back of the net. So, in some ways, more valuable than Corsi.

I really think the main reasons why +/- gets crapped on so much is that

A) It is not new
B) It is very easy to understand (What's the point of even talking about it if you don't get to sound like a professor)

[heep223]

Originally Posted by Red John

Shot attempts is always going to be flawed until someone figures out how to account for shot quality and that's incredibly difficult to pinpoint. In the meantime, let's save the term "advanced" for stats that actually are.

Why can't draw a box from the posts, to the dots, to the high slot and shots/attempts from that box are worth more? Where shots have been attempted from is already tracked with those heat maps, I've never understood why the stats guys couldn't use that to measure shot quality and incorporate it into a stat.

[PeteMoss]

Originally Posted by heep223

Why can't draw a box from the posts, to the dots, to the high slot and shots/attempts from that box are worth more? Where shots have been attempted from is already tracked with those heat maps, I've never understood why the stats guys couldn't use that to measure shot quality and incorporate it into a stat.

They do (or attempt to) but in most cases having a bigger sample size is more accurate than reducing it down by eliminating other shots.

[Coach]

Originally Posted by heep223

Why can't draw a box from the posts, to the dots, to the high slot and shots/attempts from that box are worth more? Where shots have been attempted from is already tracked with those heat maps, I've never understood why the stats guys couldn't use that to measure shot quality and incorporate it into a stat.

Something on top of that is the quality of the player making the shot. Should Alex Ovechkin's shots be worth more than Bouwmeester's?

[Enoch Root]

it's also not just about the location of the shot - the amount of time/space the shooter has is also important. But completely impossible, at this point, to track.