Stats for defense

[unknown33]

What are the most useful stat(s) to compare the defense of two players other than Norris/Selke voting or watching them play ?

As we all +/- on it's own is garbage ....

Click to read more...

 

[Buck Aki Berg]

Is it really a bad thing that we don't have finite stats to help us rank and compare defensemen? If anything, the subjective nature of the lack of finite stats makes the comparisons, and the arguements that ensue, more entertaining.

...though I suppose, since time-on-ice and plus/minus are both tabulated and readily available, you could do some legwork and make a GAA stat for defensemen.

 

[NOTENOUGHJTCGOALS]

I wouldnt place much judgement in any stat to give an idea how good a defensive ability a player has. I dont even know what stats are tracked but if they were I think some things that would give a good indication are:

Highest PK time on ice compared to teammates.

Lower number of defensive zone hooking/holding/interference penalties.

On the ice for most defensive zone faceoffs compared to teammates (of course with no changes on icing this wouldnt work anymore).

 

[TheDevilMadeMe]

For recent years, you have goals against per 60 minutes (GA/60). But quality of competition needs to be taken into account, and while there have been attempts to quantify this, I don't find them very satisfactory.

Overpass's adjusted plus/minus is a useful tool for comparing players in similar situations. (He can explain what it is more thoroughly).

Percentage of time spent on the penalty kill, combined with a good penalty kill % would indicate a good penalty killer.

One thing that i think is useful is to compare Norris voting and All-Star voting with points production. If a player gets a higher share of votes than his point production would indicate, that means something. Likewise if someone like Phil Housley racks up a lot of point, but with disproportionately lower share of Norris votes, that means something too.

And it's not so much a stat, but there is a stickied threat on this board that has coach's polls from over the years - one of the categories in these coach's polls is often "best defensive defenseman."

 

[Ogopogo*]

There are none.

The only way to evaluate defensive play is to watch the guy in action. There is no statistic that tells you anything about individual defensive play - no matter how much people want to believe it.

 

[overpass]

For recent years, you have goals against per 60 minutes (GA/60). But quality of competition needs to be taken into account, and while there have been attempts to quantify this, I don't find them very satisfactory.

Overpass's adjusted plus/minus is a useful tool for comparing players in similar situations. (He can explain what it is more thoroughly).

Percentage of time spent on the penalty kill, combined with a good penalty kill % would indicate a good penalty killer.

One thing that i think is useful is to compare Norris voting and All-Star voting with points production. If a player gets a higher share of votes than his point production would indicate, that means something. Likewise if someone like Phil Housley racks up a lot of point, but with disproportionately lower share of Norris votes, that means something too.

And it's not so much a stat, but there is a stickied threat on this board that has coach's polls from over the years - one of the categories in these coach's polls is often "best defensive defenseman."

I agree with this generally. The important thing for the pure statistical methods that don't take voting into account is to look at the numbers over several seasons. Don't expect precision. Don't expect the numbers to show much for players who aren't impact players, as the signal-to-noise ratio gets too low.

Adjusted plus-minus is simply an attempt to adjust plus-minus for the strength of the team and scoring level. But as a defensive measure, is has some of the same drawbacks as regular plus-minus, in that it includes offensive ability and it can't separate the contributions of linemates.

TheDevilMadeMe mentioned quality of competition. You could also look at zonestart, or the % of faceoffs taken in the defensive zone compared to the offensive zone, both available on behindthenet.ca for recent seasons. Neither one is a direct measure of defensive contribution, especially if you want a number that can be added to offensive contribution, but they let you know who the coaches consider to be their best defensive players. Unfortunately they aren't available beyond the last few seasons, so they don't give us any information from a hockey history standpoint.

 

[unknown33]

Adjusted plus-minus is simply an attempt to adjust plus-minus for the strength of the team and scoring level. But as a defensive measure, is has some of the same drawbacks as regular plus-minus, in that it includes offensive ability and it can't separate the contributions of linemates.

Great work.

Just curious: What was your source for the stats?

 

[seventieslord]

For recent years, you have goals against per 60 minutes (GA/60). But quality of competition needs to be taken into account, and while there have been attempts to quantify this, I don't find them very satisfactory.

Overpass's adjusted plus/minus is a useful tool for comparing players in similar situations. (He can explain what it is more thoroughly).

Percentage of time spent on the penalty kill, combined with a good penalty kill % would indicate a good penalty killer.

One thing that i think is useful is to compare Norris voting and All-Star voting with points production. If a player gets a higher share of votes than his point production would indicate, that means something. Likewise if someone like Phil Housley racks up a lot of point, but with disproportionately lower share of Norris votes, that means something too.

And it's not so much a stat, but there is a stickied threat on this board that has coach's polls from over the years - one of the categories in these coach's polls is often "best defensive defenseman."

This is pretty much what I would have said too.

There's no great way to do it, but there are a few decent ways to do it that, when used appropriately, can help to tell the story.

 

[overpass]

Great work.

Just curious: What was your source for the stats?

I downloaded the plus-minus data from here and worked with that. There's an excellent database with the data here. Having it all in one file made it possible to calculate the numbers for all players.

Hockey-reference.com also has all the stats necessary to calculate the numbers.