Shanny gets “puck possessedâ€

[Pookie]

I'm not a big fan of corsi saying anything about puck possesion and the tink to it's armour is from my own studies of learning the old red army soviet hockey as to why that is. Those Russian teams would get out shot every game do to they wouldn't shoot unless there was a a scoring chance but they had the puck in the offensive zone forcing you to play defense for the majority of a game.

This is what Bowman had happen to him a long time ago as the article pointed out and they got spanked bad by puck possession hockey to create offensive scoring chances.... You can't score if you don't have the puck but you also wont score often if all your shots on goal are of the get it on net look for tips, rebounds variety which is LA hockey and Boston Bruin hockey. They have strong corsi numbers but actual poor scoring chances attempts which is why they've never been strong offensive teams. They'r strong at puck possession but again not by the means to create offense. They're puck possession hockey is the new defense. Hold on to pucks, grind the wall on the cycle tire the opponent with safe shots on net looking for tips rebounds. A big chunk of the NHL play this way that's why scoring is down not because of goalies and there equipment as many think. This is why the game has got boring.

Russian puck possession hockey is what the greatest of all time played Wayne Gretzky. There was nothing boring about that individual and his ability to what you call play keep away hockey (puck possession hockey) that you describe as boring.

What is boring is the Don Cherry love for grind hockey that has plagued the league leading to Lemieux calling the NHL a garage league many years ago and it has progressively gotten worst. Lest grind more skill & vision in the team game will increase scoring chances league wide. That is what needs to change for the better of the game to increase scoring and the entertainment value of the game.

When this rebuild is over it will not only be a strong puck possession team that would make them strong defensively in result but also one of the most entertaining teams in the league do to skill & creativity playing fast paced keep away hockey and scoring goals in result.... Detroit Red Wings hockey, Chicago Black Hawks Hockey... Ron Wilson hockey that we stupidly let Don Cherry manipulate as a problem with the Leafs that lead to his replacement Randy Carlyle and why this team went backwards not forward with him in many ways even to the point killing young players in the process.

Look, interesting view here. I do find it interesting that you reference "Ron Wilson Hockey" when Ron's 2009-2010 team was amongst the top of the NHL in Corsi metrics.

Sometimes we try to paint teams with a brush. Babcock's team is supposedly a Corsi coached one. Yet their 2 wins come from games in which they were outshot. The 3rd win was when the shot total was basically even. Is this what Jim Corsi was trying to describe when he devised the ratio?

A decent portion of the fan base is buying this "possession based hockey" as some new religion.

This "improved play" people cite has occurred as the team has gone from 3rd overall in Shot % Attempt Close to 17th. Interestingly NHL.com suggests is a "significant indicator of which teams make the playoffs" (Leafs were 30th overall at that point which is worth a chuckle).

An improvement, I guess, over Carlyle's teams. But they are falling. And falling fast. Yet people are buying that there is some significant improvement. It started that way but the trend is downwards. Net result may be higher but let's not pretend that you can play Keep Away and win. The key is converting chances. That's why teams can win while being outshot. They convert the chances.

 

[Pookie]

5v5 SV% has an 'r' (correlation with Pt%) of 0.483; 5v5 Sh% has an 'r' of 0.267; 5v5 CF% has an 'r' of 0.532 (increases to 0.584 when score situation is 'Close').

You do realize that the data you are using here to try to put me down suggests a "medium" level of correlation at best? 0.267 is actually considered "Weak." Closer to 1 or -1 and we have something. Closer to 0, we've got nothing. Generally accepted interpretation of the Pearson Product Correlation… Correct?

I'm not sure why the daggers come out when someone says, hey, you know, there might be more to this NHL prediction game than Corsi.

I started my post with the following statement: "It's something to consider but there are more things to consider."

I then proceeded to highlight why I felt there were "gaps" in this Jim Corsi derived theory of shots for and against leading to points using a high profile team.

You've taken it upon yourself to call me a kid, and throw some "r" values out there that indicate a weak to medium level of correlation…. which I assume you did assuming that perhaps I wouldn't understand "r" or "p" values?

Weak to medium level of correlation. Yeah. That's what I am saying. Something to consider but there are more things to consider.

Enjoy your evening.

 

[91Kadri91*]

You do realize that the data you are using here to try to put me down suggests a "medium" level of correlation at best? 0.267 is actually considered "Weak." Closer to 1 or -1 and we have something. Closer to 0, we've got nothing. Generally accepted interpretation of the Pearson Product Correlation… Correct?

Yeah, the statistics YOU thought were meaningful have a weak (or moderate) correlation with little to no reliability.

'CF%, Close' has the second strongest correlation with Pt% among non-GF related (that being any statistic without GF/A in the title) statistic, and the highest reliability (thereby it is both valid and reliable to a moderate degree) among all listed statistics. As I said in my initial post, Corsi/Fenwick is not perfect, but it's significantly better than the statistics you presume to be important.

I'm not sure why the daggers come out when someone says, hey, you know, there might be more to this NHL prediction game than Corsi.

I started my post with the following statement: "It's something to consider but there are more things to consider."

And I didn't dispute that.

You also said that teams that finish or save shots are more likely to win, which simply isn't true. Teams with a better aggregate goal differential are more likely to win, but you can't build a team around great goal differential players, because goal differential is fickle.

I then proceeded to highlight why I felt there were "gaps" in this Jim Corsi derived theory of shots for and against leading to points using a high profile team.

Jim Corsi created Corsi to measure goalie workload. It was subsequently shown to have a strong correlation with TOA, and a moderate correlation with Pt%.

You've taken it upon yourself to call me a kid

I assumed you were, since you make the same arguments interminably despite being shown that your arguments are often baseless.

and throw some "r" values out there that indicate a weak to medium level of correlation…. which I assume you did assuming that perhaps I wouldn't understand "r" or "p" values?

No, I did it to show that the statistics you find meaningful are not all that meaningful (and are much less meaningful than the possession statistics you denounce).

Weak to medium level of correlation. Yeah. That's what I am saying. Something to consider but there are more things to consider./QUOTE]

Nobody said they're the 'be-all-end-all', and they're far from perfect, but they're much better than the alternatives. They should certainly be used in conjunction with GF metrics, but I'd rather have a player with great possession metrics than a player with a great TkA/GvA ratio, or a high one-year Sh%.

 

[dimi78]

Look, interesting view here. I do find it interesting that you reference "Ron Wilson Hockey" when Ron's 2009-2010 team was amongst the top of the NHL in Corsi metrics.

Sometimes we try to paint teams with a brush. Babcock's team is supposedly a Corsi coached one. Yet their 2 wins come from games in which they were outshot. The 3rd win was when the shot total was basically even. Is this what Jim Corsi was trying to describe when he devised the ratio?

A decent portion of the fan base is buying this "possession based hockey" as some new religion.

This "improved play" people cite has occurred as the team has gone from 3rd overall in Shot % Attempt Close to 17th. Interestingly NHL.com suggests is a "significant indicator of which teams make the playoffs" (Leafs were 30th overall at that point which is worth a chuckle).

An improvement, I guess, over Carlyle's teams. But they are falling. And falling fast. Yet people are buying that there is some improvement. It started that way but the trend is downwards. Net result may be higher but let's not pretend that you can play Keep Away and win. You can but the key is converting chances. That's why teams can win while being outshot. They convert the chances.

There's many ways of playing hockey All that puck possession hockey has as direct source of advantage is defensively. You have the puck the other team can't score it's that simple. LA and Boston the myth is that they're good defensively. They weren't any better than any other team in the league in the defensive zone. The Defensive zone is something I would say is survival mode for all teams that depends on goalies.. This is where the impact of Rask & Quick come to play in what they meant to those teams and why as per example Wilson's teams faltered especially on the PK and were pined as a poorly coached defensive team when Wilson was touted as a freaking defensive guru. No defensive zone game plan will work without goaltending thus the rather huge importance to every team in hockey. Both Boston and LA were machines on the offensive cycle playing keep away but not with the means to create offense but as means to there strength defensively IN THE OFFENSIVE zone to win there cups so to suggest you can't win play keepaway your wrong. Teams who play defense in the D zone all the time end up like Carlyle's Leafs.

Team Canada won gold in the Olympics by dominating puck possession not by this myth they were super strong defensively. They barely had to play in there own D zone was there means to winning. That's what puck possession hockey is first and for most about but like I said there's different ways of playing it. It has very little to do with shot's on goal vs shots against. In fact teams that take more shots are teams that can't score so yes finishing chances is rather huge that's why I hated the Phill Kessel drama that can't win with this guy as a core member of your team.

As for corsi like I said I'm not big on it but it is information that if you could control context is like anything else as a tool to help you eliminate bias from your opinions. That's what advance stats is all about. Most poor decisions are made by biasness advance stats is an attempt to help eliminate it as much as possible from your decision making.

If I was to follow an advance stat it would be a team stat of offensive zone time %, neutral zone time % and defensive zone time % and I'm sure because in spite of difference of opinions on many things, you are knowledgeable when you want to be so I'll let you make your own opinion as to why that is.

 

[teeder333*]

One thing about puck possession is that unless its being played by supremely fit guys, a great defensive counterattack rope a dope system can wreck havoc on it.

Pat Burns teams were sublime at letting teams run all over the offensive zone tiring themselves out(puck possession is exhausting) then finally grabbing the puck and running it up the length of the ice and potting a counterattack goal.

 

[dimi78]

One thing about puck possession is that unless its being played by supremely fit guys, a great defensive counterattack rope a dope system can wreck havoc on it.

Pat Burns teams were sublime at letting teams run all over the offensive zone tiring themselves out(puck possession is exhausting) then finally grabbing the puck and running it up the length of the ice and potting a counterattack goal.

I'd argue you have it backwards. Just by playing the game from my own experience it was much more tiring playing in the D zone chasing play rather than dictating it... more times than not your not moving up the ice in counter attack your changing lines.

You know this whole drama about fitness under Carlyle wasn't fitness it was Carlyle's old school approach to defensive hockey that was very similar to Burns in some ways but with added stupidity in not utilizing PMD, stretch pass tip in and chase hockey as a means to a break out into the offensive zone that had them play practically 70 % of every game in the D zone. Being able to score goals is what lead the Leafs to positions where they were playoff teams but burned out from goalies, D men, forwards for playing too much in the D zone in an 82 game schedule... The short year they were falling in the standings and just survived in result. Reimer didn't get them in the playoffs, toughness didn't get them in the playoffs, scoring goals did.

 

[yubbers]

Great read

 

[teeder333*]

Its impossible to expend as much energy merely skating without the puck than carrying the puck. I played a lot of high end hockey and set up a four on four league that had tons of NHLers and AHLers and some OHLers , participating. In fact the NHLers loved it so much, that league I started prompted the NHL to go four on four due to the participants years of exclaiming to their NHL brethren about how much fun it was to play and entertaining to watch.

And on top of that, if the defenders box perfectly, they barely have to do much at all, but play perfect positionally.

Watch film of Burns teams they would tire the offensive team out, time and time again.

Sure if you have five guys playing no position and skating recklessly it could be tiring, but this isn't minor atom we are talking about.

 

[Pookie]

Yeah, the statistics YOU thought were meaningful have a weak (or moderate) correlation with little to no reliability.

'CF%, Close' has the second strongest correlation with Pt% among non-GF related (that being any statistic without GF/A in the title) statistic, and the highest reliability (thereby it is both valid and reliable to a moderate degree) among all listed statistics. As I said in my initial post, Corsi/Fenwick is not perfect, but it's significantly better than the statistics you presume to be important.

I don't think you are understanding. I highlighted that shots and shots against have a weak correlation. And the "r" values on that blog you put forward clearly indicate that. Those "r" values for the other metrics are the same. Medium to weak correlation.

If CF% close has an "r" value of .5xy that means essentially that if you took 10 teams, 5 of them would show a picture that we'd expect with the Corsi theory.

5 would not. There would no picture that resembles what Corsi stats profess to project. With a 30 team NHL, that's 15 teams that support the picture and 15 teams that don't.

Now, GF and GA metrics have much stronger correlations as we'd expect. Some over .800. But that's about conversion. As I said originally, that's the key. Not possession or shots. Here's what I said:

"Other variables, such as goaltending and conversion of opportunities seems to be more important."

Are you in disagreement with that conclusion?

 

[91Kadri91*]

I don't think you are understanding. I highlighted that shots and shots against have a weak correlation. And the "r" values on that blog you put forward clearly indicate that. Those "r" values for the other metrics are the same. Medium to weak correlation.

'Shots-for percentage' has a moderate correlation (0.523) with Pt%, actually.

Also, as I've stated several times, the original article was in reference to Corsi (possession), so why do you continue to bring up 'shots-for'?

If CF% close has an "r" value of .5xy that means essentially that if you took 10 teams, 5 of them would show a picture that we'd expect with the Corsi theory.

Again, it's 0.584 (closer to 60% than 50%). This correlation, as I've stated countless times, is significantly greater than the correlation shown by statistics others find important (Sh%, TkA, GvA, SV%, FO%).

5 would not. There would no picture that resembles what Corsi stats profess to project. With a 30 team NHL, that's 15 teams that support the picture and 15 teams that don't.

Which, even if that were the case, would be good (considering it also has moderate reliability). A statistic should be valid and reliable, and Corsi/Fenwick/xGoals is the closest we have to a perfectly valid and reliable statistic (goal metrics are valid but unreliable).

Now, GF and GA metrics have much stronger correlations as we'd expect. Some over .800.

Uh, yeah, that's what I said... like five times.

But that's about conversion. As I said originally, that's the key. Not possession or shots.

Here's where you're wrong.

GF% and possession are closely associated, and xG/Corsi/Fenwick are all significantly better at predicting future GF% than current GF% is. Goal metrics may indicate how well a team has played to this point, but they're not predictive (the sample size is simply too small), and as such you have to question how indicative they are of a team's abilities.

Another reason is simply goals are extremely variable. While Corsi tends to be stable, shown to have a split-half season reliability around 70%, goals are around 40% for the same sample. This is essentially the root reason why Corsi is a stronger predictor. When you are making predictions of goals based off of goals, you are trying to predict a heavily variance influenced variable with another heavily variance influenced variable. When you try instead with Corsi, you at least have one variable that is steady in shot attempts.

Possession statistics are simply a much more stable account of how well a team has played, and will continue to play.

Also, you said conversion; 'conversion', by definition (and taking into account context) is the number of goals scored by shots taken, which is shooting-percentage. You can see in the chart that Sh% has a significantly worse correlation with Pt% than CF% does.

Here's what I said:

"Other variables, such as goaltending and conversion of opportunities seems to be more important."

Are you in disagreement with that conclusion?

Yes, because conversion (Sh%) and goaltending (Sv%) have a weaker correlation with Pt% than Corsi% does.

I agree that the goal should be to outscore your opponent (obviously), but I made that abundantly clear in my very first post.

 

[Damisoph]

Same author of that awesome Timashov article right?

Great talent.

Yep, Nitro whatever loll that was a great article too.

 

[Bullseye]

Nice seeing some of the stupid journalism in this city get their due Chesfuk - perfect.

I loved that era for the Wings - Konstantinov was my favourite non-Leaf player on the Leaf.

Great article - thanks for sharing it.

 

[Le Cobra]

One of the most eye opening articles regarding the direction the management are taking the Leafs. We are so used to the old Leafs that when you take a really close look at them, they are actually starting to look more like the Red Wings from the Bowman years. Can't wait for our own version of the Russian 5 to drop with Nylander, Kapanen, Marner, Brown, and Soshnikov.

 

[Nithoniniel]

I wonder how many times the same thing needs to be said before certain posters stop selectively forgetting it the next time they want to argue something.

It's sad how often people make up their minds first and judge the merit of arguments accordingly.

 

[Pookie]

I think it's a little sad how metrics that generally show a .2 to .5 correlation (weak to medium) get trumped as gospel by those claiming others don't understand.

Clearly some folks here are seeing exactly what a weak to medium correlation results in. A team that is outshot but wins or "plays their best hockey" as one poster observed.

You yourself said these are only part of the picture. So, what's wrong with highlighting why they aren't THE picture?

 

[Pookie]

91kadri91,

If you want to round .584 up to .6, that's fine. That's still telling is that 4 of 10 teams (or 12 in the NHL) won't show any correlation whatsoever.

It's why we see upwards of 4-5 top ranking Corsi teams miss the playoffs every year.

Though this isn't advanced, go to NHL.com and sort the league table on goal differential. Notice that the higher you are the more points you have?

Now do the same for every season? Notice that virtually all of your top 16 in goal differential make the playoffs?

Without doing a calculation, I would suggest that Goal differential as a single metric is about as close to 1 as we can get in terms of correlation.

What does that mean?

That scoring more than you let in leads to points.

Simple. But that's what the game is about.

That comes down to being skilled in conversion and goaltending. Find me players that excel at that and I will find you teams that have great goal differentials and the resulting points.

If the best keep away player in the world can't score or the goalie lets in goals from the blue line, you have a team that struggles.

It won't sell papers or get blog hits but that's the reality.

 

[leafs in five]

anyone else getting kind of bored of being comfortable with the direction of the team and confident that the guys in charge know what they're doing?

 

[amack081]

This way was a fantastic article and three things stuck out to me:

1) Even though the Wings in '95 were the most dominant team in the NHL, they couldn't get by Patrick Roy and the Avs. Even though they had a winning formula it required tinkering. The next season they added Shanny and had a gritty bottom (Lapointe, McCarty, Draper and Maltby). These bottom end guys were still talented and brought different attributes to the team. The article talks about Russian 5 man units, and its interesting because its rare that we see a defensive pairing match up with the offensive trio on a regular basis in the NHL now (except on the PP and in a defensive match up capacity).

2) If you look back at the late 80s and early 90s drafting by the Wings, they were extremely successful . If you look at 89 to 91, they had 11 players play over 400 NHL games. They were able to create a nucleus of players that would end up being their philosophy for years to come.

3) Veteran core and solid youth. The 96-97 and 97-98 winning team had a Larionov, Fetisov, Murphy who were in the late end of their 30s. Yes the game has changed in terms of speed and youth movement, but the point is veteran presence must be served and that all 3 of those mentioned, thought the game at an extremely high level. I think we all get caught up in having a team full of prospects/young players, that we forget the value of veteran pedigree.

Just my two cents.

 

[DeclanK]

Part II of this series from Quinn MacKeen is up:

"Brendan Shanahan, The Russian 5, and the New Generation Maple Leafs – Part II: Reversing the curse"

Read more at: https://mapleleafshotstove.com/2015/11/16/brendan-shanahan-the-russian-5-and-the-new-generation-maple-leafs-part-ii-reversing-the-curse/

 

[91Kadri91*]

91kadri91,

If you want to round .584 up to .6, that's fine. That's still telling is that 4 of 10 teams (or 12 in the NHL) won't show any correlation whatsoever.

I didn't round up to .6, I said that it's closer to .6 than the .5 you rounded down to.

Though this isn't advanced, go to NHL.com and sort the league table on goal differential. Notice that the higher you are the more points you have?

Now do the same for every season? Notice that virtually all of your top 16 in goal differential make the playoffs?

Without doing a calculation, I would suggest that Goal differential as a single metric is about as close to 1 as we can get in terms of correlation.

What does that mean?

That scoring more than you let in leads to points.

Simple. But that's what the game is about.

How have you not been banned? How is this not against the rules?

I've said six ****ing times now that the goal should be to outscore your opponents; it's common ****ing sense. The only thing I've argued (in respects to goal metrics) is that goal metrics are unreliable (fact), and that while outscoring your opponent should be the goal, it can't be the plan (because goals are so fickle). You build a system that pushes the puck up ice and puts your team in a position to score, and the opposing team in a position to, y'know, not score. You can't build a system where the opposing team just directs shots and controls possession. That's not a sustainable model of success.

That comes down to being skilled in conversion and goaltending. Find me players that excel at that and I will find you teams that have great goal differentials and the resulting points.

Of course that requires skill, just as possession requires immense skill. What goal metrics inherently include, however, that 'conversion' and 'goaltending' don't, is volume. The top goal scorers also happen to be among the league leaders in shots on goal. Additionally, goals against inherently includes the defensive system deployed, and the ability to suppress shots/possession. You're arguing that 'conversion' (Sh%) and 'goaltending' (Sv%) are more important to winning than possession/volume- they're not.

 

[91Kadri91*]

I think it's a little sad how metrics that generally show a .2 to .5 correlation (weak to medium) get trumped as gospel by those claiming others don't understand.

The only metrics showing weak correlation are the metrics that YOU find noteworthy!

 

[Pookie]

You know, an interesting read is a book called Soccernomics, written by a statistician and an economist.

Basically, talking about advanced stats and the world of football.

There is the usual stuff like are you more likely to score on an in swinging corner kick and the value (or lack thereof) of possession.

The real interesting key and the type of information not available on the web are the metrics dealing with player performance.

When a player is at their peak and moving them when they are on the decline.

To do that, wearable tech is entering into the game... as we are seeing starting to be explored in hockey. We aren't there yet.

To my point, if you can find skilled goalies or skilled shot converters and out them on a team where they all peak around the same time you win.

Interestingly, in looking at players from possession based- run and dominate the midfield clubs, they tend to decline in skill faster. More miles on the old legs so to speak.

This movement towards metrics to evaluate players is key. It isn't available in blogs or public data.

That is the stuff of real meaning. If you want to continue to devide numbers to come up with ratios and argue about them, I will leave you to it.

The real data is knowing when players are peaking and a variety of metrics such as health, reaction times, training intensity, etc, etc are key.

 

[Discoverer]

Without doing a calculation, I would suggest that Goal differential as a single metric is about as close to 1 as we can get in terms of correlation.

What does that mean?

That scoring more than you let in leads to points.

Simple. But that's what the game is about.

Scoring more increases your chances of winning the game. Playing a strong possession game increases your chances of scoring more. Obviously goals has a stronger correlation to wins, but "Let's go out and control the puck" works much better as a strategy than "Let's outscore these guys!"

 

[Gallagbi]

To my point, if you can find skilled goalies or skilled shot converters and out them on a team where they all peak around the same time you win.

So high PDO?

 

[Pookie]

Scoring more increases your chances of winning the game. Playing a strong possession game increases your chances of scoring more. Obviously goals has a stronger correlation to wins, but "Let's go out and control the puck" works much better as a strategy than "Let's outscore these guys!"

Actually it doesn't. Ask the 2009-2010 Leafs or the 4 top 4 Corsi teams that will miss this year's playoffs.