Advanced stats and assessing defenders.

[MacDonald4MVP]

Is there a way to actually evaluate defense?

Click to read more...

 

[Filthy Dangles]

Shot Suppression metrics?

 

[Breakers]

Shot Suppression metrics?

Yeah shot suppression, so"CA60" & "Relative CA60"
You'd look at ones penalty killing numbers

You'd look at high high dangerous scoring chances. while said player is one the ice.

WAR numbers and Puck impact as well

 

[BigRangy]

If you ask oilerbear there's definitely a way.

Watching how the Flames have been playing for the last few seasons I'm almost actually inclined to think he's onto something with his thinking.

 

[LEAFANFORLIFE23]

Advanced stats can and do lie at times.

Exhibit A Martin Marincin, he has had good advanced stats in the past and yet he's awful I have watched this guy lose games by himself, last year for example he's the reason the Leafs lost the season opener to the sens despite Matthews getting 4 goals he was responsible for both the 3-3 and 4-4 goals and both happened because he quit on the plays, not because he got beat, because he didn't try, he quit.

That's what he does, he quits and that's why he's in the AHL

 

[Breakers]

Marincin has average defensive numbers though.

 

[LEAFANFORLIFE23]

Marincin has average defensive numbers though.

yet he's not good enough to make this team, a team who for all their strengths offensively isn't the best defensively, yet he can't help them.

What does that tell you?

 

[Doctor No]

I realize that it's super easy with "advanced stats" to just say "so and so has great advanced stats, but he sucks nerds LOL." Because there are lots of "advanced stats", including ones that don't have context applied.

And so no one can pick apart your argument because you haven't actually offered up what you're attacking.

With that said, what are the "advanced stats" that suggest that Marincin is good (per your first post above)?

 

[LEAFANFORLIFE23]

I realize that it's super easy with "advanced stats" to just say "so and so has great advanced stats, but he sucks nerds LOL." Because there are lots of "advanced stats", including ones that don't have context applied.

And so no one can pick apart your argument because you haven't actually offered up what you're attacking.

With that said, what are the "advanced stats" that suggest that Marincin is good (per your first post above)?

One of the things I keep hearing about is how he has, or has had good shot suppression numbers, even if that were true if he treats the puck like it's a bomb that is about to explode every time he has the puck it doesn't matter, it also doesn't matter if he quits on plays and he does, way WAY to much

 

[Doctor No]

One of the things I keep hearing about is how he has, or has had good shot suppression numbers

Then show us the numbers that support that case.

Coming in with a "I heard such-and-such about advanced statistics, but Marincin's in the minor leagues" doesn't help advance the argument. We don't know what numbers you're relying upon, so how can anyone critique your argument?

 

[Johnny Engine]

Here are a few links on Marincin's statistical results.

Most of them acknowledge what Marincin is good and bad at. It's extremely difficult to get around him at the blue line because he covers so much space. Furthermore, that's exactly the kind of skill that's likely to drive shot-attempt metric results - The Hockey News has a primer on that here.

While I can't prove a negative, I'm not sure that there are any effective statistical models that can prove how detrimental "whoops" moments are, right around your own net, and man, does Marincin ever serve up some doozies. I'd understand why the coaching and management of the Leafs don't want to tolerate that kind of play from Marincin on their roster, zone entries be damned.

Anyway, the lesson I get from this is that no stat is going to measure anything as big and amorphous as "defense", but it can give you indicators as to specific areas where a player is or isn't effective - especially if you have some eye-test familiarity to back it up.

 

[Doctor No]

Agreed - a lot of people are hoping that "advanced stats" will one day result in people pouring a pile of data into a grinder, turning the crank a few times, and out the end will pop out a ranked list of the BEST HOCKEY PLAYERS. (Alternatively, they're claiming that that's what the "stats" guys are hoping for).

It's a fallacy. It's never an either/or - numbers need experience, and experience needs numbers. Both make the other better. The eye test works - it's worked for decades. That doesn't mean that things can't be better.

 

[morehockeystats]

Agreed - a lot of people are hoping that "advanced stats" will one day result in people pouring a pile of data into a grinder, turning the crank a few times, and out the end will pop out a ranked list of the BEST HOCKEY PLAYERS. (Alternatively, they're claiming that that's what the "stats" guys are hoping for).

It's a fallacy. It's never an either/or - numbers need experience, and experience needs numbers. Both make the other better. The eye test works - it's worked for decades. That doesn't mean that things can't be better.

The advanced stats, or any stats, should not work like that grinder. They should work like a DSS (Decision Support System) input. When a person is trying to make a decision between two alternatives, they should display the factors the decision-maker deems important, or help him or her formulate these factors.
Otherwise, computer could run the sports team, and we see how well it's going in Arizona...