Top and bottom 15 Dman who have played at least 50 minutes 5on5 in relative xGF% for October

[catnip]

This thread is a repeat offender. They are still relative numbers that haven't gotten any more meaningful since the last time this was tried. Back then it was the Oilers defenders hogging the top spots on both lists.

 

[tarheelhockey]

You don't understand how this stat works.

It's in comparison to the other dman on your team.

It favours guys who are used offensively, and guys who play on teams where the other pairings suck ass.

It punishes guys where the other pairings are elite (see Luke Hughes and Brett Pesce, who are both above 50% xgoals share), but since the rest of NJD has been a dominant 67% in the 4 games since they've been back, they have bad relative numbers

So yeah, I understand exactly how the stat works.

The stat is what’s problematic here, not my understanding of it.

 

[dgibb10]

So yeah, I understand exactly how the stat works.

You said
"Basically, god help a low-offense shutdown guy who ends up on a bad team or just in front of a bad goalie. He’ll be rated miles worse than a third pairing offensive specialist who isn’t producing but is on a good enough team that his GF/GA breaks even."

A) expected goals are goaltender independent

B) This 3rd pairing dman would be rated quite poorly considering he's being compared to the elite dmen in front of him. Example being luke hughes and brett pesce. Their xGF has broken even. Unfortunately, since the other 2 pairings are at 67% over the last 4 games, their relative numbers are bad.

The stat isn't good, but not for the reasons you think. It rewards guys who are "the least bad out of the dogshit" like Barron in this case, and can punish Mediocre when compare to elite

 

[tarheelhockey]

Yes, players used largely in pure defensive roles or against tough competition will be at a raw disadvantage by these metrics (there are better ones out there that try to adjust for usage, QoC, and QoT though). I know this and already acknowledged it.

Again though, as I said, there are guys who play tough minutes that are on the positive side of the ledger. K'Andre Miller plays the tough competition with Fox. Gavrikov plays with Anderson and they are the team's shutdown pair.

Good defensive players, even given their deployment, can still rate highly by these metrics. Last season, McDonagh, Alex Vlasic, Gudas, Nick Seeler were all near the top of the league by relative xG. They play the defensive minutes for their teams.

Again, defensive defenseman can outperform their usage - that is what defines an actually good defensive defenseman. If you're a defensive defenseman that's just getting shelled, you're not really doing your job. Ideally your designated shutdown guys can at least break even in tough usage, which basically subsidizes the rest of your team to get softer minutes. If they can't accomplish this, they're not really doing much for you.

I've named several players that do accomplish this.

You wanted the opposite side of the ledger? Last season, Tyson Barrie, Drysdale, and Calen Addison were some of the worst defensemen in the league by relative xG. They are guys that get soft, offensive deployment. Usage isn't going to completely obscure whether a guy is good or not.

Your team is terrible. Part of the reason for that is that you have two D-level defensemen who are lucky to have NHL jobs.

One is a defensive defenseman. He plays on the third pair, in front of a terrible goalie, with a bottom-pair partner. As a result he gets shelled.

The other is an offensive/skill defenseman. He plays on the second pair with a pretty good partner who is a pretty good shooter. As a result this guy gets a lot of o-zone starts against soft defenders, and picks up a lot of passive assists.

Does this metric rate these two players differently? If so, there’s a problem with the metric.

You said
"Basically, god help a low-offense shutdown guy who ends up on a bad team or just in front of a bad goalie. He’ll be rated miles worse than a third pairing offensive specialist who isn’t producing but is on a good enough team that his GF/GA breaks even."

A) expected goals are goaltender independent

B) This 3rd pairing dman would be rated quite poorly considering he's being compared to the elite dmen in front of him. Example being luke hughes and brett pesce. Their xGF has broken even. Unfortunately, since the other 2 pairings are at 67% over the last 4 games, their relative numbers are bad.

The stat isn't good, but not for the reasons you think. It rewards guys who are "the least bad out of the dogshit" like Barron in this case, and can punish Mediocre when compare to elite

See above.

 

[dgibb10]

Your team is terrible. Part of the reason for that is that you have two D-level defensemen who are lucky to have NHL jobs.

One is a defensive defenseman. He plays on the third pair, in front of a terrible goalie, with a bottom-pair partner. As a result he gets shelled.

The other is an offensive/skill defenseman. He plays on the second pair with a pretty good partner who is a pretty good shooter. As a result this guy gets a lot of o-zone starts against soft defenders, and picks up a lot of passive assists.

Does this metric rate these two players differently? If so, there’s a problem with the metric.



See above.

RE your scenario: Please find a metric that wouldn't rate them differently.

If you are looking for 1 metric that allows you to turn your brain off completely, you will not find it anywhere. (and one that would exist would be a combination of a BOATLOAD of different metrics, including expected goal share)

Also you mentioned 3-4 things that rang BIG alarm bells that you don't know what you're talking about.

A) shooter quality does not impact expected goal metrics (there are adjusted ones but we aren't talking about those here)
B) those 2 players would have the same goalie??? Also again, goalies don't impact expected goal metrics (Note: exception for a goalie who has terrible rebound control, although again, there are adjusted versions for this)
C) whether he gets an assist or not has no impact on the expected goal metrics.

 

[tarheelhockey]

RE your scenario: Please find a metric that wouldn't rate them differently.

Well that’s kind of the point, isn’t it? We can just use traditional stats or use the eye test if we want to be deceived about the quality of players. The last thing we need is a new way to be wrong.

 

[dgibb10]

Well that’s kind of the point, isn’t it? We can just use traditional stats or use the eye test if we want to be deceived about the quality of players. The last thing we need is a new way to be wrong.

It only deceives people who are either a) too lazy to think or b) not smart enough to understand what it tells you.

a stat doesn't need to be a 1 size fits all perfect description of a players talent to be valuable.

It is a piece of a puzzle telling you a specific thing. It's up to YOU as the fan to take all of the pieces of the puzzle and come to a conclusion. The more information out there, the better.