Current State of Analytics in the NHL

[Filthy Dangles]

I have a few questions regarding the current state of advanced stats in the NHL. Some questions will have more concrete answers, while some are more philosophical in nature. I would like to hear the opinions of other BTN'ers on the subjects below.

1) Have 'FancyStats' reached the point where all front offices of all 30 franchises in the game have an analytics department? If not, do most teams have a team/person dedicated to analytics.

2) Also, besides reaffirming the notion that puck possession is a) highly correlated with winning and b) able to predict with relative success the outcome of future events, what can analytics (as they exist today) if anything, actually afford NHL teams in trying to gain an advantage in winning games?

The way I see it, out of all the major sports, the actual visceral game at ice level of hockey is the most abstract or 'disconnected' from the realm using advanced analysis to modify your team's strategy, especially compared to baseball. Of course, this is due to inherent things to the games like the speed and dynamics of the hockey, as opposed to the slow, turn-based, pre-determined matchup iterations in baseball. Also, baseball is a much more cerebral game where a difference in strategy is exhibited from pitch to pitch, where hockey is sometimes simply a matter of 'who wants it more' or 'who's playing harder', which again is completely abstract from analysis.

3) Is there anymore substantial progress that can be made without RFID tags?

It seems to me the biggest problems/questions to be solved are 1) actual offensive zone time and 2) how zone entry affects your zone/possession time and your quality of shots/chances. These two things seem simply too cumbersome and rigourous to calculate in the long term for all teams, without machine.

Click to read more...

 

[Doctor No]

2) Also, besides reaffirming the notion that puck possession is a) highly correlated with winning and b) able to predict with relative success the outcome of future events, what can analytics (as they exist today) if anything, actually afford NHL teams in trying to gain an advantage in winning games?

I'll take a stab at this stuff first - once something enters the public domain, the "gain an advantage in winning games" disappears.

Teams are looking for arbitrage opportunities, and the things that teams are currently doing to win games are not in the public domain (so it's hard to talk about them here). I've had enough off-the-record discussions that I can confirm that these team efforts are happening.

 

[Canadiens1958]

Team Efforts

I'll take a stab at this stuff first - once something enters the public domain, the "gain an advantage in winning games" disappears.

Teams are looking for arbitrage opportunities, and the things that teams are currently doing to win games are not in the public domain (so it's hard to talk about them here). I've had enough off-the-record discussions that I can confirm that these team efforts are happening.

Nice to see that team efforts are finally being recognized. They have been around since before the NHL.

Conversely the "Team Efforts" are the only measures that accurately judge how the team on the ice executed the game plan, the developmental plan, etc that always remains internal to the team, never entering the public domain with the exception of snippets years later from retired or traded players.

Since the public will never have access to the game plan or the developmental plan, etc, the outsiders can only guess.

 

[Doctor No]

Nice to see that team efforts are finally being recognized.

"Finally"? I've been talking about them pretty much since joining the site (and prior to that). And I'm not the only one.

 

[Canadiens1958]

Team Efforts II - November 9, 1954

I'll take a stab at this stuff first - once something enters the public domain, the "gain an advantage in winning games" disappears.

Teams are looking for arbitrage opportunities, and the things that teams are currently doing to win games are not in the public domain (so it's hard to talk about them here). I've had enough off-the-record discussions that I can confirm that these team efforts are happening.

November 9, 1954, Dink Carroll column from the Montréal Gazette:

http://news.google.com/newspapers?id=5IIuAAAAIBAJ&sjid=UpkFAAAAIBAJ&hl=fr&pg=6900,1703628

offers insight into how the Montreal Canadiens were tracking shots for individual players - Bernie Geoffrion is used. Note how the tracked the provenance of the shots - how close to the net. How a players shooting habits change over time with experience - geoffrion more successful as he gets closure to the net.

Granted the article is filled with a few old bromides, etc that misdirect the reader from the analytics. Nor is any hard data provided that may be used by others.

 

[Doctor No]

Great link - and I love the old Carroll articles in general.

I think there's an excellent book to be written about how these sorts of techniques have evolved over the years.

 

[Canadiens1958]

Elmer Ferguson

The Montréal Gazette was the third English paper in town after the Montréal Star and the Montréal Herald - featured Elmer Ferguson as the sports editor/columnist. Elmer Ferguson joined the Star when the Herald folded in the late 1950s.

Elmer Ferguson was also the NHL statistician from day one in 1917, until Ron Andrews joined the league.

Nuggets similar to the Dink Carroll column were common in Ferguson's works.

Copies of virtually all of the Montréal Herald and the Montréal Star are available on microfilm.

Also Frank Selke Jr, wrote a article in the Canadiens or Forum Programs titled Frankly Speaking, in the fifties and sixties. Camil Desroches did likewise in French - different topics. At times these would offer statistical insight into the team. Also provided insight into how the Canadiens gathered and interpreted stats during that era.

Will try to find a few other examples, especially those that provide insight into how the Canadiens viewed +/-.

 

[Jumptheshark]

Oilers are proof that some people put too much reliance on fancy stats

Most fancy stats have the oilers in the Black

 

[Doctor No]

Oilers are proof that some people put too much reliance on fancy stats

Most fancy stats have the oilers in the Black

A sample of n=1 constitutes proof in your mind? Especially when you have no idea how the Oilers' use of "fancy stats" stacks up to the league? (All you know is what's released publicly).

 

[Canadiens1958]

Early Stats

Great link - and I love the old Carroll articles in general.

I think there's an excellent book to be written about how these sorts of techniques have evolved over the years.

A sampling of newspaper and literary works that shed some light on how stats were kept,used,interpreted during the 1940s using Maurice Richard as a point of reference:

1943 to 1945

http://hfboards.mandatory.com/showpost.php?p=89670305&postcount=302

1948
Prior to the start of the 1948-49 season:
http://news.google.com/newspapers?i...AAAIBAJ&dq=mickey mackay&hl=fr&pg=4644,352699

1949-50
Looking at the importance of goals:
http://hfboards.mandatory.com/showpost.php?p=89699393&postcount=363

 

[HamiltonNHL]

State of analytics in 3 years:

 

[Canadiens1958]

Consumer Stats

^^^Basically consumer stats. Entertainment and promotional value.

 

[ZeroPT*]

Oilers are proof that some people put too much reliance on fancy stats

Most fancy stats have the oilers in the Black

Or you know, the team is just horribly built with absolutely no structure.

 

[The Mighty Duck Man]

Or you know, the team is just horribly built with absolutely no structure.

Definitely, but with them, one can't help but wonder if the reason they haven't made any major moves is because of their advanced stats. Basically, they think they're better than they actually are because their Corsi isn't that bad and their PDO is, so they don't think they need a shakeup.

 

[Doctor No]

Or they know that they need to make moves, but they can't make the necessary moves at a price they feel is reasonable.

Or the front office folks aren't listening to the "advanced stats" guys - that happens in a lot of organizations. "No nerd is going to tell me what to do, I've been riding buses for the last 35 years."

As long as we're speculating without any reliable information, I mean. Everything bad in the league doesn't have to be the fault of "advanced stats", especially when that's not even a well-defined term.

 

[Canadiens1958]

Takes Two

Or they know that they need to make moves, but they can't make the necessary moves at a price they feel is reasonable.

Or the front office folks aren't listening to the "advanced stats" guys - that happens in a lot of organizations. "No nerd is going to tell me what to do, I've been riding buses for the last 35 years."

As long as we're speculating without any reliable information, I mean. Everything bad in the league doesn't have to be the fault of "advanced stats", especially when that's not even a well-defined term.

Takes at least two parties to make a move regardless of whether advanced stats are used or not.

Potential incoming players have "no Edmonton" clauses as part of their contract. Likewise free agents.

No different than post WWII when players - Roy Conacher, red Kelly and others refused trades to the NY Rangers due to the higher cost of living in NYC and the resulting extra cost of maintaining two households if the wife and youngsters stayed home in Canada.

 

[Shrimper]

Is there a list of NHL statistic websites here? Find that a lot have been shut and need to find alternatives.

 

[The Mighty Duck Man]

Or they know that they need to make moves, but they can't make the necessary moves at a price they feel is reasonable.

Or the front office folks aren't listening to the "advanced stats" guys - that happens in a lot of organizations. "No nerd is going to tell me what to do, I've been riding buses for the last 35 years."

As long as we're speculating without any reliable information, I mean. Everything bad in the league doesn't have to be the fault of "advanced stats", especially when that's not even a well-defined term.

It was a simple question, where I admitted I was just a fan wondering about it. It does make sense, MacTavish, in his press conferences, gave(to me, at least) an impression that he never thought their situation was that bad, and by the looks of things, only fired Eakins because of the mounting pressure to do so, and not because he actually thought Eakins was a problem. Obviously, something like corsi and PDO agrees with that thinking, so I don't think I was out of line with that question.

 

[Doctor No]

It was a simple question, where I admitted I was just a fan wondering about it. It does make sense, MacTavish, in his press conferences, gave(to me, at least) an impression that he never thought their situation was that bad, and by the looks of things, only fired Eakins because of the mounting pressure to do so, and not because he actually thought Eakins was a problem. Obviously, something like corsi and PDO agrees with that thinking, so I don't think I was out of line with that question.

What was the question part? The post of yours that I was responding to was a pretty clear statement of opinion.

Granted, it was couched in "one can't help but wonder" language. And if you truly were wondering, then I gave you several alternatives to wonder about.

 

[The Mighty Duck Man]

What was the question part? The post of yours that I was responding to was a pretty clear statement of opinion.

Granted, it was couched in "one can't help but wonder" language. And if you truly were wondering, then I gave you several alternatives to wonder about.

How would it be opinion? We, the general public, have no idea what MacT has tried to do, and given Lowe's previous statements about knowing how to win, it's just as believable that MacT doesn't think there's a problem just because the boys club is that arrogant and has nothing to do with advanced stats. It wasn't couched with anything, it's a legitimate question, and also fits with a scenario I thought could happen this past offseason(team approaches analytics the wrong way and it blows up in their face).

As for what you added, all could be legitimate points as well, never said otherwise. I was just responding to you clearly getting defensive. What, the idea that a team bought too heavily into a single stat like Corsi to negative effect is inconceivable to you?

 

[Doctor No]

As for what you added, all could be legitimate points as well, never said otherwise. I was just responding to you clearly getting defensive. What, the idea that a team bought too heavily into a single stat like Corsi to negative effect is inconceivable to you?

What made you think that I was being defensive?

Providing alternatives is being defensive? This isn't a discussion forum?

 

[The Mighty Duck Man]

What made you think that I was being defensive?

Providing alternatives is being defensive? This isn't a discussion forum?

Considering the original question wasn't addressed at all, absolutely I'd say that was a defensive post. It'd be one thing to continue the discussion and also include alternatives, but you didn't. Not to mention, I was pretty specific, not wondering why the Oil haven't made any moves, but wondering if this was why. That's not a statement that really calls for alternative ideas(at least not right away), yet that's all you offered up. No actual discussion, just a quick retort in defense of advanced stats. Of course you were being defensive, that was pretty obvious.

As for your alternatives, the first one is obviously quite possible, and probably the most likely case, but I don't know if I buy the second. They were pretty public about jumping into the analytics fray headfirst this offseason, and the signings of guys like Pouliot and Fayne showed they were definitely listening to their analytics guys at some point. I don't think that randomly stopped sometime during the season. Not saying they're doing everything they say, either, but I definitely can't see them flat out ignoring them by this point.

 

[GKJ]

If the Oilers really have sent Tyler Dellow back home, we know it has failed.

 

[Doctor No]

If the Oilers really have sent Tyler Dellow back home, we know it has failed.

We know *what* has failed?

 

[Ogopogo*]

We know *what* has failed?

The use of bad analytics - like Corsi - as a player evaluation tool.