Five year study on the plus minus statistic...

[Preisst*]

I came across this article about a five (5) year study that was done on the plus/minus statistic. (did a search but didn't find another thread on it)

I implore posters to read it, I suspect many will find it highly interesting.

Some bottom line points are that 30% of pluses awarded on a goal were found to not involve the player at all.

Defensively it was far worse as it was found that 50%, yes 50%, of players did nothing to warrant getting a negative and it was very often that players were doing their job defensively and were given the minus because of another players error.

Please note that one of the words in the link is filtered so you will need to copy and paste it :


blogs.edmontonjournal.com/2013/05/13/just-how-horse-

Here are some other interesting articles on the plus minus.

http://islanderspointblank.com/news/the-farce-that-is-plusminusthe-most-thankless-stat-in-hockey/

http://www.shutdownline.com/hurricanes/statistical-analysis/fun-with-plusminus.html

http://blogs.thescore.com/nhl/2013/...us-etiquette-amongst-teammates-in-pro-hockey/

Enjoy and feel free to discuss.

Click to read more...

 

[LuGBuG]

What if a puck comes up the wall, winger tries to chip is out doesn't succeed, goes behind the net up the other wall, puck still doesn't get out, puck stays in for another 45 seconds, Dmen are gassed and the other team scores? Those 2 wingers didn't have a direct impact on the goal when it was scored but still deserve a minus. There are plenty small simple plays that can lead to a goal over a 30 second to minute shift. In my opinion I don't think there will ever be a definite way to decide whether it's meaningful or not and I can't see a better stat coming that I know of or can think of.

 

[Crosbyfan]

I came across this article about a five (5) year study that was done on the plus/minus statistic. (did a search but didn't find another thread on it)

I implore posters to read it, I suspect many will find it highly interesting.

Some bottom line points are that 30% of pluses awarded on a goal were found to not involve the player at all.

Defensively it was far worse as it was found that 50%, yes 50%, of players did nothing to warrant getting a negative and it was very often that players were doing their job defensively and were given the minus because of another players error.

Please note that one of the words in the link is filtered so you will need to copy and paste it :


blogs.edmontonjournal.com/2013/05/13/just-how-horse-

Here are some other interesting articles on the plus minus.

http://islanderspointblank.com/news/the-farce-that-is-plusminusthe-most-thankless-stat-in-hockey/

http://www.shutdownline.com/hurricanes/statistical-analysis/fun-with-plusminus.html

http://blogs.thescore.com/nhl/2013/...us-etiquette-amongst-teammates-in-pro-hockey/

Enjoy and feel free to discuss.

So on an average goal it is 70% effective on the plus side and 50% on the negative? Sounds like it is statistically very valid, especially over large sample sizes...

...and of course taken in the right context for the player under consideration

 

[tarheelhockey]

So on an average goal it is 70% effective on the plus side and 50% on the negative? Sounds like it is statistically very valid, especially over large sample sizes...

Doesn't 50% effectiveness make it totally invalid? I mean, you could take a team's total ESGA+SHGA, assign a corresponding number of minuses randomly* across the roster, and get an equally valid set of numbers, right?

* edit: Of course you'd have to adjust for ice time as well. But the point is, the likelihood of a defensive-minus being assigned correctly is 50/50... that doesn't sound particularly valid to me.

 

[Preisst*]

Doesn't 50% effectiveness make it totally invalid? I mean, you could take a team's total ESGA+SHGA, assign a corresponding number of minuses randomly* across the roster, and get an equally valid set of numbers, right?

* edit: Of course you'd have to adjust for ice time as well. But the point is, the likelihood of a defensive-minus being assigned correctly is 50/50... that doesn't sound particularly valid to me.

I also don't think the 70 - 30 split on the offensive side is very appealing. I think the bottom line is that the study verifies the impotence of the useless plus/minus stat.

 

[Ho Borvat]

One Thing People Forget
Whos going to have a worse plus minus?

- A guy playing infront of Steve Mason
or
- A guy playing infront of Henrik Lundqvist?

If your team is getting .900 goaltending, and your opponents are getting .930 goaltending then your getting an additional "-3" every time you and your opponent take 100 shots each.

Theres so little context with +/-. It just tells you who was on the ice when something happened at even strength.

 

[BraveCanadian]

I think this confirms to some degree what pretty much everyone suspected.. of course the solution of injecting a subjective decision on individual responsibility also creates another issue but it looks like the pros would outweigh the cons on that one.

Leave it to Captain Video to have found a decent working solution years ago.

Too bad it would be so labour intensive to have it kept as an official stat.

 

[MarkGio]

I don't think plus/minus is tracked so that professionals, GMs, and coaches alike can evaluate and analyse a player; its tracked so that the layperson can follow hockey with some analysis.

Imagine if broadcasters started using corsi and trying to explain it to people? Does anybody even know the math formula to it? At least +/- is simple enough for a fan new to the game, or even a child, exploring a new analysis to hockey.

 

[MastuhNinks]

I don't think plus/minus is tracked so that professionals, GMs, and coaches alike can evaluate and analyse a player; its tracked so that the layperson can follow hockey with some analysis.

Imagine if broadcasters started using corsi and trying to explain it to people? Does anybody even know the math formula to it? At least +/- is simple enough for a fan new to the game, or even a child, exploring a new analysis to hockey.

Corsi at its simplest is basically +/- but with shots, it's not complicated.

 

[tarheelhockey]

One Thing People Forget
Whos going to have a worse plus minus?

- A guy playing infront of Steve Mason
or
- A guy playing infront of Henrik Lundqvist?

Time to revisit this question

 

[Darth Joker]

+/- is a mildly useful stat, but only with large helpings of context:

1. Effective power play specialists should be given some leeway since they can easily make up for bad +/- with power play production (which doesn't help their +/- stat).

2. Guys who log a lot of minutes for bottom feeders will likely have a very bad +/- even if they're good defensively.

3. They're most useful for comparisons between two players on the same team, not between two players on different teams.


Corsi is definitely the better stat, as it factors more in. But +/- is a decent stat for the casual hockey fan as long as they keep a few things in mind (like the 3 I mentioned above).

 

[Ohashi_Jouzu*]

I don't think plus/minus is tracked so that professionals, GMs, and coaches alike can evaluate and analyse a player; its tracked so that the layperson can follow hockey with some analysis.

Imagine if broadcasters started using corsi and trying to explain it to people? Does anybody even know the math formula to it? At least +/- is simple enough for a fan new to the game, or even a child, exploring a new analysis to hockey.

I've seen corsi statistics directly referred to on multiple networks already, actually. It has begun.

 

[MarkGio]

Corsi at its simplest is basically +/- but with shots, it's not complicated.

Sorry, I meant more the Corsi Rel QoC. Yeah a shooting tally for minus against isn't really an advance stat.

Quality competition? Is that based on ice time? Is context taken, such as team that takes 15-20 minutes in penalties?

Relative to the team? Is that how much more/less than the average? Is there weight?

I've never really seen the formula so I don't know.

 

[tarheelhockey]

Imagine if broadcasters started using corsi and trying to explain it to people? Does anybody even know the math formula to it?

I've seen corsi statistics directly referred to on multiple networks already, actually. It has begun.

Not only that, but how many sports fans can explain the math formula behind QB Rating or Slugging %? Yet these are "trading card" stats that people take at face value.

I know the Sabres broadcasts have been using Corsi, and I expect that will increase as fans continue to see it used as an alternative to +/-. They might not be able to explain the difference, but they will understand that it's a useful number.

 

[Ho Borvat]

Time to revisit this question

haha I nominate this for most awesome post ever.

 

[Yog S'loth]

Jack Johnson desperately wants you to believe that +/- is useless.

 

[JackJ]

My take:

Plus/Minus is the most misused stat in hockey.

Only useful when comparing players on the same team and players who perform the same function. Top 6 vs Top 6 not Top 6 vs Bottom Six Checker, otherwise it's useless.

In Yakopov's case:

Jordan Eberle: -5
Ales Hemsky: -10
Ryan Nugent-Hopkins: -10
Taylor Hall: -10
Sam Gagner: -16
Nail Yakupov: -24

He's -14 worse than his peers median. Average would be a bit more accurate. In this case he is playing a lot worse defensively.

Now that advanced stats are the norm its time to retire that stat completely.

 

[Exit Dose]

Corsi is definitely the better stat, as it factors more in. But +/- is a decent stat for the casual hockey fan as long as they keep a few things in mind (like the 3 I mentioned above).

This is a problem that isn't solved by increased sample size, though. There's no reason to think that the problem goes away because you've added results that didn't end in a goal to the process. Even a 30% rate of false positives is uncomfortably high.

 

[GKJ]

I don't think plus/minus is tracked so that professionals, GMs, and coaches alike can evaluate and analyse a player; its tracked so that the layperson can follow hockey with some analysis.

Imagine if broadcasters started using corsi and trying to explain it to people? Does anybody even know the math formula to it? At least +/- is simple enough for a fan new to the game, or even a child, exploring a new analysis to hockey.

To Corsi? Differential between all on-ice attempted shots for and against. It's actually easier to explain than +/- which excludes on-ice power play goals for and on-ice shorthanded goals against.


Jim Fox and the Sabres broadcasts have both referred to Corsi during broadcasts, so, it's out there.

 

[Feed Me A Stray Cat]

+/- isn't great, but I don't like how this study was done. There's more to how a goal is created than on the play itself directly.

A solid puck moving defenseman might not have a direct hand in a goal, but his swift clears and turning the puck up the ice will end up in creating more goals while also preventing many.

And obviously +/- needs to be adjusted for strength of team.

 

[Hardyvan123]

+/- is a mildly useful stat, but only with large helpings of context:

1. Effective power play specialists should be given some leeway since they can easily make up for bad +/- with power play production (which doesn't help their +/- stat).

2. Guys who log a lot of minutes for bottom feeders will likely have a very bad +/- even if they're good defensively.

3. They're most useful for comparisons between two players on the same team, not between two players on different teams.


Corsi is definitely the better stat, as it factors more in. But +/- is a decent stat for the casual hockey fan as long as they keep a few things in mind (like the 3 I mentioned above).

I pretty much agree with this, plus/minus isn't totally useless but it does give us an indication and direction most of the time when used in context.

 

[trudatman*]

I love +/-. I weigh it with time on ice. it's not a perfect formula, but it's very helpful for comparing players effectiveness within a team. "but zone starts and quality of stuff!" bah.