+/- anomalies

[tarheelhockey]

Some will people will always try to totally discount +/- as an individual stat. It's far from perfect and I would never judge a player defensively merely by looking at his +/-. I do think in the proper context and combined with other stats it may be useful in some cases. I think Larry Robinson's +730 in the proper context and combined with other stats may suggest he was an above average Dman while fellow Dman Rhett Warrener's +73 in the proper context and combined with other stats may suggest that he was not quite as good as Robinson defensively. Let's the flames ignite on me.

Can you give an example of how plus-minus can be combined with other stats to produce a more accurate analysis than if we split up GF and GA?

 

[begbeee]

Absolute poppycock. There are exactly zero stats that require no context in order to interpret properly.

If you disagree, please tell us a hockey stat that does not require context to interpret.

Goals.

 

[Bear of Bad News]

Goals.

The currency of goals was much lower in the 1980s than it was in the late 1990s. Fifty-goal scorers were not as valuable in the 1980s.

That context is important.

 

[seventieslord]

Goals.

Ilya Kovalchuk's goals need to be critically analyzed because he played for a poor run and gun team that used him for the entire PP - in fact he led the NHL in PPTOI multiple times.

Pavel Bure's goals need to be critically analyzed because he floated around at the blueline looking for a breakout pass in Florida and it's arguable that his presence was a net negative regardless of his goal totals there.

Kevin Stevens' goals need to be critically analyzed because he had Mario Lemieux passing him the puck in his prime. What does he score on 19 of the other 20 teams?

Mike Ricci's 243 career goals are pretty respectable, but the guy actually only ever scored 125 ESG in his career - he almost couldn't score without the PP time he was given.

there are just 4 examples of context.

 

[Iain Fyffe]

Goals.

Warren Young, 1984/85. Tell me how context is not important to goals again?

The two posts above also make salient points.

 

[85highlander]

I fail to see how Dallas Smith supports your overall point that "only the truly great two-way players [are] represented at the top of the list". Dallas Smith is not a "scrub" per se, but he is an unremarkable player. Yet there he is at the top of the list. The "clear causal connection" you refer to, otherwise known as leeching off a linemate, is a major problem with the +/- stat... not a virtue. It's designed to present frequent false positives and false negatives, Dallas Smith being an excellent example.

With Orr's extremely high disproportional +/- this would be expected -- no "surprises" here. I don't see this as a "problem" with the stat, rather I see it as reinforcing it - again, to be expected, as someone had to be the beneficiary of Orr's insane +/-.

Also, the assertion that Guy Lafleur is a "truly great two-way player" is questionable as well. Lafleur was a one-way player backed by an incredible defensive unit; his disproportionate +/- is nearly as much a reflection of his team situation as Smith's, as evidenced by his unimpressive post-dynasty +/-. In 1984, when the Habs finished 4th in their division with Rick Wamsley in goal, Lafleur was -14.

Are you ready to make the same argument with Gretzky, as much of the same could be stated concerning him...

During those 3 seasons in the 1960s, Savard had a +.24 per game. Orr had a +.79 per game.

During the 1970s, Savard had a +.74 and Orr had a +1.10. Some of that can be chalked up to entering their primes, but let's get real about the era effect here. A list of raw +/- per game leaders is basically just a list of the best players on the best teams of the 1970s.

Yes, the best players have the best +/- stats -- good observation.

And then he never had a losing season again. He played on 5 teams which had a winning percentage over .700.

I'm not sure where you're trying to go with this point. Clarke was clearly on a mini-dynasty team and his best +/- seasons were at the height of their dominance. There's obviously a direct correlation between team +/- and individual +/-.

And even on the three losing seasons when the Flyers were horrendous, Clarke NEVER had a negative +/- -- so again, maybe the individual player has something to say about a "team" stat.

 

[begbeee]

I was wrong.

 

[sr edler]

Pavel Bure's goals need to be critically analyzed because he floated around at the blueline looking for a breakout pass in Florida and it's arguable that his presence was a net negative regardless of his goal totals there.

wow, bure was third for the hart in 99-00 [which is the last time the panthers made the playoffs], he got most goals and easily most points on the team, best +|− on the team, second in league scoring, but i guess "his presence was a net negative"

even if you cut off roughly 10 goals those are still 50 goal seasons in the dead puck era on the florida panthers

i don't think you need to "critically analyze" bure's goal scoring ability more than you need to critically analyze the goal scoring ability of say jari kurri, or brett hull, or anyone really

 

[nik jr]

Cool project.

Without digging into the details too deeply, my first thought is that by working backwards from results instead of forward from causes it may underrate players who do a lot of little things well, especially in puck and territorial possession, and give their teammates more opportunities to make good plays and fewer opportunities to make bad plays.

This stat would certainly be more reliable than regular plus-minus at the game level or playoff series level. Maybe at the season level too. At the career level I'd probably want full EV plus-minus to catch those little things. But as a fan of a team you want something that works in-season, of course.

In any case, I wouldn't throw it out because it's not 100% perfect. Like any hockey stat or metric it's an imperfect model of a fluid game.

that is one of the main ways in which +/- can conflict with visual evidence.

it is not usually anyone's fault when the opponents have the puck more.


brad stuart looks better defensively than his stats. but he is not very good in transition or in keeping the puck on offense, so he spends too much time in the defensive zone. thus he was scored on more (throughout his career). i have heard similar analysis by a calgary fan of regehr.

i did not even think of that explanation of stuart until 2010, i think.



i think this is also an important part of lidstrom's +/- "anomaly" (definitely not all, though).

'08-'10: with rafalski (+40, +31, +22)
'11: mostly with stuart (lidstrom -2)
'12: with white (lidstrom +25)


lidstrom has become slow. rafalski and white are both faster puck movers and help lidstrom much in transition. stuart is clearly better defensively than rafalski and white and is not slow at all, but his pair with lidstrom spent too much time in the defensive zone, and was not as good at keeping play in the offensive zone, or at scoring.

but i don't want to scapegoat stuart, b/c lidstrom was also just not great defensively. imo, he should not have been a norris finalist and should not this season.

another part of it is datsyuk vs zetterberg. in '11, datsyuk was injured or on 2nd line for much of the season. GA were quite a lot worse when datsyuk was injured, and babcock said something like "we have the puck a lot more" when datsyuk is playing. datsyuk is a lot more effective at controlling play than zetterberg. from '08-10 and in '12, datsyuk has been on 1st line.

team D and goaltending were also worse in '11, though partly b/c lidstrom was worse. howard is better than last season (and helped by better team D). assistant coaches were changed (RIP mccrimmon), and F's are playing better in the defensive zone.



a DRW blog has a similar project to tarheel's. http://www.wingingitinmotown.com/2011/10/4/2467289/the-common-sense-scoring-index-for-2011-12

Sure they do.

it should be more accessible. +/- (+ and -) as in sweden and finland, instead of having to look at calculate +'s and -'s.

 

[seventieslord]

it should be more accessible. +/- (+ and -) as in sweden and finland, instead of having to look at calculate +'s and -'s.

I don't know what you mean though. Because the official numbers are released each year, no need to calculate anything.

 

[tarheelhockey]

With Orr's extremely high disproportional +/- this would be expected -- no "surprises" here. I don't see this as a "problem" with the stat, rather I see it as reinforcing it - again, to be expected, as someone had to be the beneficiary of Orr's insane +/-.

What you are describing is the definition of a false positive. How is this not a problem?

Are you ready to make the same argument with Gretzky, as much of the same could be stated concerning him...

Are you saying Gretzky was a poor two-way player? Or that he played in front of a great defense?

Either way, I'd say this argument is going off the rails already.

Yes, the best players have the best +/- stats -- good observation.

You missed the crucial prepositional phrase: "the best players on the best teams".

And even on the three losing seasons when the Flyers were horrendous, Clarke NEVER had a negative +/- -- so again, maybe the individual player has something to say about a "team" stat.

The Flyers weren't "horrendous" in those seasons, just weak offensively. Their defense was middle-of-the-road all three years. Clarke was their scoring leader by age 22. A high-scoring player on a decent defensive team is going to have a positive +/-.

In any case, you can't argue with a straight face that Clarke didn't have a good team situation in his career. The Flyers won more than 60% of their games and two Stanley Cups while he played there.

 

[nik jr]

I don't know what you mean though. Because the official numbers are released each year, no need to calculate anything.

where? (again my point about accessibility)

only place i know to find them is here: http://www.nhl.com/ice/playerstats....12&sort=plusMinus&status=A&viewName=plusMinus

but it does not simply split +/- into + and -. it gives GF, GA, PPGF and PPGA. hockey-reference keeps same numbers.

 

[seventieslord]

wow, bure was third for the hart in 99-00 [which is the last time the panthers made the playoffs], he got most goals and easily most points on the team, best +|− on the team, second in league scoring, but i guess "his presence was a net negative"

even if you cut off roughly 10 goals those are still 50 goal seasons in the dead puck era on the florida panthers

i don't think you need to "critically analyze" bure's goal scoring ability more than you need to critically analyze the goal scoring ability of say jari kurri, or brett hull, or anyone really

So that must be why Bure's Panthers were so good, because what their best forward was doing was best for the team and not his own personal stats. Right?

where? (again my point about accessibility)

only place i know to find them is here: http://www.nhl.com/ice/playerstats....12&sort=plusMinus&status=A&viewName=plusMinus

but it does not simply split +/- into + and -. it gives GF, GA, PPGF and PPGA. hockey-reference keeps same numbers.

right, exactly. what more are you looking for?

 

[nik jr]

right, exactly. what more are you looking for?

more accessible + and -

it should be like this: http://estat.hockeyligan.se/c/LPlayersPlusMinus.aspx?LId=248

 

[plusandminus]

Seventieslord, the + and - is available where nik jr say it is, but it is not an easily accessible stat. One has to dig it up. IIHF and domestic leagues in Europe lists it next to the points, for example:

Name|Team|Pos|GP|G|A|Pts|PIM|+|-
Sidney Crosby |PIT |cf |26|20 |30 |50 |14 |40|22

or

Name|Team|Pos|GP|G|A|Pts|PIM|+|-|+/-
Sidney Crosby |PIT |cf |26|20 |30 |50 |14 |40|22|+28

It's still not perfect, and have weaknesses. But surely you realize that it's easier than having to have to calculate it oneself, by combining TotGF, TotGA, PPGF and PPGA?

Anyway, for NHL's "+/-" specific reports, I think the NHL should list situational stats in different columns, for example:

TotGF| TotGA| ESGF| ESGA| PPGF| SHGA| SHGF| PPGA| ENGF| ENGA

or

TotGF| TotGA| ESGF| ESGA| PPGF| PPGA| SHGF| SHGA| ENGF| ENGA

Then one would see instantly, without having to do calculations or studying gamelogs, how players have performed during each situation.

By the way, would you find an even strength only +/- useful? (That is, one that only consider even strength goals where both teams have a goalie on the ice.)

 

[sr edler]

So that must be why Bure's Panthers were so good, because what their best forward was doing was best for the team and not his own personal stats. Right?

i don't even know if i get what you're saying here but bure's panthers wasn't that good because hockey is a team game, not because he was a general liability to the team, only trevor kidd would say that

even if bure transformed himself to 93-94 fedorov the 00-01 and 01-02 panthers wouldn't have made the playoffs, have you seen those rosters? marcus nilson was a key offensive player on those teams

 

[thom]

plus and minus

Response to Lafleur being a one zone player is the most incorrect line I'v heard.As someone growing up and watching most of his games I can tell you he came into the defensive zone all the time.No superstar of that era ever got slashed highsticked or roughed up.In junior career he was in over a dozen fights and in the early part of his career he was in over 8 nhl fights.Scotty Bowman messed his early career by playing him at center in his first 3 seasons.As a rookie Guy Had 3 hat ricks as a rookie the most in nhl history at the time mostly playing 3rd and fourth line.Ask Jim Roberts or Howie Meeker who is the best competitor they ever saw his name was le blond Demond.

 

[plusandminus]

Response to Lafleur being a one zone player is the most incorrect line I'v heard.As someone growing up and watching most of his games I can tell you he came into the defensive zone all the time.No superstar of that era ever got slashed highsticked or roughed up.In junior career he was in over a dozen fights and in the early part of his career he was in over 8 nhl fights.Scotty Bowman messed his early career by playing him at center in his first 3 seasons.As a rookie Guy Had 3 hat ricks as a rookie the most in nhl history at the time mostly playing 3rd and fourth line.Ask Jim Roberts or Howie Meeker who is the best competitor they ever saw his name was le blond Demond.

(Just clarifying: the subject line of the previous post has nothing to do with me or anything I have written. My chosen nickname here has some plus and minus too, in this case a minus.)

 

[nik jr]

Anyway, for NHL's "+/-" specific reports, I think the NHL should list situational stats in different columns, for example:

TotGF| TotGA| ESGF| ESGA| PPGF| SHGA| SHGF| PPGA| ENGF| ENGA

or

TotGF| TotGA| ESGF| ESGA| PPGF| PPGA| SHGF| SHGA| ENGF| ENGA

Then one would see instantly, without having to do calculations or studying gamelogs, how players have performed during each situation.

By the way, would you find an even strength only +/- useful? (That is, one that only consider even strength goals where both teams have a goalie on the ice.)

that would be much better, and EN would be a very good addition.

you have a good name for this.

 

[tarheelhockey]

Response to Lafleur being a one zone player is the most incorrect line I'v heard.As someone growing up and watching most of his games I can tell you he came into the defensive zone all the time.

Lafleur was the third best defensive forward on his own line. You can say he was still good defensively, but in the grand scheme of NHL history he is not a significant two-way player.

 

[seventieslord]

more accessible + and -

it should be like this: http://estat.hockeyligan.se/c/LPlayersPlusMinus.aspx?LId=248

Seventieslord, the + and - is available where nik jr say it is, but it is not an easily accessible stat. One has to dig it up. IIHF and domestic leagues in Europe lists it next to the points, for example:

Name|Team|Pos|GP|G|A|Pts|PIM|+|-
Sidney Crosby |PIT |cf |26|20 |30 |50 |14 |40|22

or

Name|Team|Pos|GP|G|A|Pts|PIM|+|-|+/-
Sidney Crosby |PIT |cf |26|20 |30 |50 |14 |40|22|+28

It's still not perfect, and have weaknesses. But surely you realize that it's easier than having to have to calculate it oneself, by combining TotGF, TotGA, PPGF and PPGA?

Anyway, for NHL's "+/-" specific reports, I think the NHL should list situational stats in different columns, for example:

TotGF| TotGA| ESGF| ESGA| PPGF| SHGA| SHGF| PPGA| ENGF| ENGA

or

TotGF| TotGA| ESGF| ESGA| PPGF| PPGA| SHGF| SHGA| ENGF| ENGA

Then one would see instantly, without having to do calculations or studying gamelogs, how players have performed during each situation.

By the way, would you find an even strength only +/- useful? (That is, one that only consider even strength goals where both teams have a goalie on the ice.)

I see what you guys are saying. I can honestly say it's never been a big concern of mine. I can add and subtract those numbers on the fly if I have to. and if I want to do major work I just throw them into a spreadsheet and I have all pluses and minuses in 2 seconds.

i don't even know if i get what you're saying here but bure's panthers wasn't that good because hockey is a team game, not because he was a general liability to the team, only trevor kidd would say that

even if bure transformed himself to 93-94 fedorov the 00-01 and 01-02 panthers wouldn't have made the playoffs, have you seen those rosters? marcus nilson was a key offensive player on those teams

First of all, I said it was "arguable" that he was a net negative presence, not that he definitely was. Secondly, it was sarcasm, obviously the team wasn't good, and though I typically don't blame the "best" players on a team for that, Bure might be one of those exceptions. if the team was shorthanded when he was on the ice with the exception of the moments he got hold of the puck and rushed up ice, that's arguably a negative presence. "Feed the Panther", his teammates said. That doesn't sound like a healthy environment/

 

[sr edler]

First of all, I said it was "arguable" that he was a net negative presence, not that he definitely was. Secondly, it was sarcasm, obviously the team wasn't good, and though I typically don't blame the "best" players on a team for that, Bure might be one of those exceptions. if the team was shorthanded when he was on the ice with the exception of the moments he got hold of the puck and rushed up ice, that's arguably a negative presence. "Feed the Panther", his teammates said. That doesn't sound like a healthy environment/

i get it if we're talking about the 01-02 season, he wasn't good that year but it shouldn't affect what he did in 99-00 or even 00-01

 

[plusandminus]

I see what you guys are saying. I can honestly say it's never been a big concern of mine. I can add and subtract those numbers on the fly if I have to. and if I want to do major work I just throw them into a spreadsheet and I have all pluses and minuses in 2 seconds.

Do you really? Isn't it rather impossible to know the event strength + and - based on the data provided there, because one doesn't know about short handed goals?
OK some say, short handed goals are rare and doesn't affect things much anyway, but if one think they do (like I definitely do) there is not enough data provided to know about even ESGF and ESGA.
Even if things like strength of opposition, etc., isn't considered, the more factual numbers we do have, the better I think it is.
So why not show factual ESGF and ESGA instead? After all, that's what many are looking for anyway.

A major problem with +/- today is that it favours players playing more PK than PP, and vice versa. So why include special teams play in the +/-? Let's just let it show ES performance.
If we want to know about PP performance, there are other things, like for example PP points scored, or personalized net PP%. SH performance is more problematic to show with just one number, but percentage of icetime might be a fairly indicative number.