+/- anomalies

[Iain Fyffe]

Thank you for taking my bait and proving my point.

10 games played is EXACTLY my point. 11 players had better +/- than Orr that year because they played more games on a good team. +/- has so many external factors that are not controlled by the individual that it is a completely worthless number.

That's just ridiculous. If you think the number of games counts as an externality that invalidates the statistic, then the same externality invalidates all recorded statistics.

If that's the point you were trying to make, it is the weakest point I think I've ever seen, and does nothing to support your argument.

Again, you're going to need a better example.

 

[Nalyd Psycho]

I agree, but if corsi is probably adjusted for shot quality we probably have a winner.

(maybe it already is? I haven't researched it hard enough)

If so then absolutely that is a great efficiency stat.

 

[Prophecy35]

If you really consider how statistics actually work, especially in hockey, every stat needs some context, even goals. In 84-85 (random 80s season) there were 56 players that scored 30 goals. Flash forward to 03-04 and there were just 20. Take someone like Warren Young who scored 40 goals, and then any one of Iginla/Kovy/Nash at 41. Are they nearly equivalent at face value? Sure, goals are goals after all. Was Young as effective? Well if just reading the stat line was that straightforward, then you would have no reason to notice that he played with Lemieux and never had a season like that before or after.

I don't agree with adjusted stats, and I'm not treating it like the bible, as some do, but context is always necessary when looking at any stat, even something as straightforward as goals.

One player that really is amazing is Herb Cain. What a generational talent. I mean, he won the NHL scoring race in 43-44. Fabulous. That is precisely what you get when not looking at the complete picture.

The reality is, that was during the extremely watered down war years, it was an anomaly season for him to an extreme degree, and he has no real accomplishments of note outside of that season, as far as I can tell.

Simply put, no matter what the stat, it can almost never be taken at face value. I still maintain that +/-, though a troublesome stat, can be useful when used in the right contexts.

 

[Ogopogo*]

That's just ridiculous. If you think the number of games counts as an externality that invalidates the statistic, then the same externality invalidates all recorded statistics.

If that's the point you were trying to make, it is the weakest point I think I've ever seen, and does nothing to support your argument.

Again, you're going to need a better example.

Read post 23

 

[Iain Fyffe]

Read post 23

Did that already, though with the personal attacks it contains on anyone who disagrees with you, it's difficult to take it seriously.

Besides that, of course, is the obvious fact that the Orr example was presented by you after the post you refer to here, so surely it was meant to make an additional point of some kind, yes?

 

[Iain Fyffe]

Simply put, no matter what the stat, it can almost never be taken at face value. I still maintain that +/-, though a troublesome stat, can be useful when used in the right contexts.

This is the essential point. +/- is more troublesome than certain stats, of course (though less than some others), but writing it off as useless is naive.

 

[Johnny Engine]

Here's the problem, though.

...links...

The major outlets including the NHL present raw +/- as THE basic measure of defense. Even if you go a step deeper and look at the NHL's main stat sheets: http://www.nhl.com/ice/playerstats.htm#?navid=nav-sts-indiv

it gives you GP-G-A-P-plusminus-PIM as the basic stat line with 7 more columns until you get to TOI... which apparently ranks with shifts/game and faceoff% in their list of priorities. This sends a message to people looking at leaderboards that raw +/- is a meaningful statistic in and of itself. The people running stats for the league should surely be expected to know better, even if the media outlets don't.

Yeah, you're definitely right about the way +/- is promoted. That's stupid.
However, I'm just going to throw it out there that when I look at those main stats sheets, the first things I look at, in order, are either points or goals (because one of them will be highlighted), GP (because it's on the hard left), and then TOI (because it's over on the hard right). I'm not sure if it's laid out that way on purpose, or whether my eyes do that because I've already internalized it as one of the more important stats on the sheet.
Either way, when someone says "advanced stats", they aren't talking about TOI.

 

[Ogopogo*]

Did that already, though with the personal attacks it contains on anyone who disagrees with you, it's difficult to take it seriously.

Besides that, of course, is the obvious fact that the Orr example was presented by you after the post you refer to here, so surely it was meant to make an additional point of some kind, yes?

Read carefully.

My additional point was: A player who gets the most ice time on a team that scores more than it allows will likely be the highest +/- player on the team.

Because +/- is a team stat, getting more playing time on that team will affect your +/- more. So, a player that is completely mediocre could have the highest +/- on a good team simply by being on the ice a lot. Being good may have nothing to do with it.

Charlie Huddy comes to mind immediately.

 

[seventieslord]

could have the highest +/- on a good team simply by being on the ice a lot. Being good may have nothing to do with it.

Charlie Huddy comes to mind immediately.

Being put out on the ice a lot and being good are closely related.

 

[tarheelhockey]

Over on the Canes board, I have embarked on a season-long project to review every ES goal scored in our games and assign a plus or minus only to players who had a causative effect. The project is crowdsourced by allowing any forumer to post objections or additions, and contains a Margin of Error option in case of disagreement (of which there is surprisingly little).

See here: http://hfboards.mandatory.com/showthread.php?t=1013809

Of course it won't be finished till next month, but I hope it can be used to benchmark the accuracy of standard plus-minus.

One interesting discovery that is already apparent: forwards are near-universally "plus" players and defensemen are near-universally "minus" players. The nature of their positions limits the impact they can have on each end of the ice. So, one apparent weakness of standard +/- is that it distributes pluses (defensemen) and minuses (forwards) in a manner that is completely team-dependent and can't be adjusted.

 

[Iain Fyffe]

Yeah, you're definitely right about the way +/- is promoted. That's stupid.

Absolutely. Plus-minus is not a defensive stat at all. The plusses come from offence, and the minuses from defence, basically. It's equal parts offence and defence.

Read carefully.

My additional point was: A player who gets the most ice time on a team that scores more than it allows will likely be the highest +/- player on the team.

Think harder. Who gets the most ice time on a team? Wouldn't that generally be its best players?

Coaches do not send players out at random.

Because +/- is a team stat, getting more playing time on that team will affect your +/- more. So, a player that is completely mediocre could have the highest +/- on a good team simply by being on the ice a lot. Being good may have nothing to do with it.

No kidding, that's why you don't take it at face value. You're still well short of demonstrating that there is zero value in the stat.

 

[tarheelhockey]

Oh -- and another thing about tha project which I think is an improvement. It excludes ENG and SHG.

 

[plusandminus]

Over on the Canes board, I have embarked on a season-long project to review every ES goal scored in our games and assign a plus or minus only to players who had a causative effect. The project is crowdsourced by allowing any forumer to post objections or additions, and contains a Margin of Error option in case of disagreement (of which there is surprisingly little).

See here: http://hfboards.mandatory.com/showthread.php?t=1013809

Of course it won't be finished till next month, but I hope it can be used to benchmark the accuracy of standard plus-minus.

One interesting discovery that is already apparent: forwards are near-universally "plus" players and defensemen are near-universally "minus" players. The nature of their positions limits the impact they can have on each end of the ice. So, one apparent weakness of standard +/- is that it distributes pluses (defensemen) and minuses (forwards) in a manner that is completely team-dependent and can't be adjusted.

Interesting. Do you prefer feedback here in this thread or in the thread in the Carolina section?

I have only glanced, but can imagine it being controversial to align + and - arbitrarily. (I know you are serious and provide videos and gives explanations, which is good.) For example, not being involved in the play can be a minus when allowing a goal. Sometimes a goal against may also be the goalie's fault. But as I say, I have only glanced, and maybe you are considering that.
At first look, it seems that you could do well to combine this with icetimes. That is because forwards generally gets +, and defencemen generally -.

Something I miss, is two columns showing the "factual" + and - (for even strength only, excluding empty net goals, which I understood was what you're looking at). Then one can compare how "favoured" or "unfavoured" different players are.

 

[Ogopogo*]

Absolutely. Plus-minus is not a defensive stat at all. The plusses come from offence, and the minuses from defence, basically. It's equal parts offence and defence.


Think harder. Who gets the most ice time on a team? Wouldn't that generally be its best players?

Coaches do not send players out at random.


No kidding, that's why you don't take it at face value. You're still well short of demonstrating that there is zero value in the stat.

If you need to "find the context" for every player to figure out what their +/- really means, then the stat is useless. You could just "find the context" and not even care about their +/-.

If the +/- cannot give you a realiable indication of anything for an individual, it is garbage.

It is like trying to run a science experiment without controlling the variables. Your conclusions mean nothing because it could be one of a dozen things giving you the result.

 

[Ogopogo*]

Being put out on the ice a lot and being good are closely related.

True but not my point.

 

[Bear of Bad News]

If the +/- cannot give you a realiable indication of anything for an individual, it is garbage.

Plus-minus gives you a very accurate indication of the number of (non-power play) goals that a player was on the ice for, compared with the number of (non-power play) goals that a player was on the ice against.

In fact, it's 100% accurate. Very reliable.

I'm not sure what sort of magic you're hoping for, but your threshold of "any stat that isn't 100% perfect is completely worthless" must make for a nice world view.

 

[Johnny Engine]

If you need to "find the context" for every player to figure out what their +/- really means, then the stat is useless. You could just "find the context" and not even care about their +/-.

So you're saying that if two guys play the #1 defenseman role on teams that respectively, lead their divisions, and they both put up 40 points and go out against top lines, one of those guys having a far superior goal differential when he's on the ice doesn't add anything to the comparison?

(by the way, my example doesn't even illustrate the nonsense you just came up with. Context, by defintion, doesn't exist if it isn't context for something.)

 

[seventieslord]

Plus-minus gives you a very accurate indication of the number of (non-power play) goals that a player was on the ice for, compared with the number of (non-power play) goals that a player was on the ice against.

In fact, it's 100% accurate. Very reliable.

I'm not sure what sort of magic you're hoping for, but your threshold of "any stat that isn't 100% perfect is completely worthless" must make for a nice world view.

And isn’t this the guy that has the best players of all-time all nicely lined up in a list based entirely on a points system including an offensive dominance metric that is in complete slavery to top-10s in scoring?

 

[Ogopogo*]

Plus-minus gives you a very accurate indication of the number of (non-power play) goals that a player was on the ice for, compared with the number of (non-power play) goals that a player was on the ice against.

In fact, it's 100% accurate. Very reliable.

I'm not sure what sort of magic you're hoping for, but your threshold of "any stat that isn't 100% perfect is completely worthless" must make for a nice world view.

It would be so much more effective to abolish +/- in its current form and simply have a guy watching the game and crediting players with a + when they do something on the ice that is clearly a significant positive contribution and giving them a - when they do something that is clearly a negative contribution. Actually watch them play and give individual marks for individual accomplishment or mistakes.

Scrap this garbage of painting everyone with the same brush. Some players make a positive difference and some make a negative difference - giving every single player on the ice credit or blame is the wrong way to do it.


P.S. It is very reliable to tell you what?

 

[Ogopogo*]

And isn’t this the guy that has the best players of all-time all nicely lined up in a list based entirely on a points system including an offensive dominance metric that is in complete slavery to top-10s in scoring?

Please do not post false/inaccurate things about me or things I have done. It is perhaps the very worst form of trolling.

 

[Bear of Bad News]

It would be so much more effective to abolish +/- in its current form and simply have a guy watching the game and crediting players with a + when they do something on the ice that is clearly a significant positive contribution and giving them a - when they do something that is clearly a negative contribution. Actually watch them play and give individual marks for individual accomplishment or mistakes.

Scrap this garbage of painting everyone with the same brush. Some players make a positive difference and some make a negative difference - giving every single player on the ice credit or blame is the wrong way to do it.

Ah, abolish! Let's have less information! I can't interpret a statistic properly, so no one should have access to it!

Are you intending to have the same person watch all games, or different people? Because the first is untenable, and the second introduces all sorts of biases.

 

[Ogopogo*]

Ah, abolish! Let's have less information! I can't interpret a statistic properly, so no one should have access to it!

Are you intending to have the same person watch all games, or different people? Because the first is untenable, and the second introduces all sorts of biases.

You mean like referees? How does the league survive with those guys if it couldn't possibly have a guy scoring "real" +/-?

 

[Bear of Bad News]

P.S. It is very reliable to tell you what?

The number of (non-power play) goals that a player was on the ice for, compared with the number of (non-power play) goals that a player was on the ice against.

Like I mentioned in the post you quoted. Are you acting dense intentionally? It is perhaps the very worst form of trolling.

 

[Bear of Bad News]

You mean like referees? How does the league survive with those guys if it couldn't possibly have a guy scoring "real" +/-?

So you're okay with biases now, just not the biases that are inherent in plus-minus?

 

[Ogopogo*]

The number of (non-power play) goals that a player was on the ice for, compared with the number of (non-power play) goals that a player was on the ice against.

Like I mentioned in the post you quoted. Are you acting dense intentionally? It is perhaps the very worst form of trolling.

Is this for real?

You are saying the current +/- is a reliable measure of a player's effectiveness on the ice - knowing full well a player is given a plus or a minus whether he was involved in the play or not?

Then you speak about dense?