Devils 2017-18 team discussion (player news and notes) - season begins!

[Emperoreddy]

Sample size this bugs me. Like in baseball. I don’t pay attention to batting averages in April. They just don’t mean much because one odd bad or good game can send an average to a weird place that isn’t at all representave. There is too much noise. By August you have so many ABs that an outlier game doesn’t throw the numbers off and you can better judge who is hitting well and who isn’t (and not just that season but compared to multiple seasons). Actual trends emerge.

At what point can you say that here? How many games is enough to cancel out outliers and statistical noise to say without a doubt that this is a good corsi team and this isn’t?

Same with any stat really. Shooting percentage as well.

 

[devilsblood]

Hell NO I'm not only using it only 5v5! I'm using it for all situations.

If there's any predictive team value to this metric to begin with, why the hell would I eliminate all powerplay and penalty kill minutes, which is further weakening my data set?!?!?!?! By a lot.

That makes absolutely zero sense.

If it proves to be a much better predictor of success why would you not use 5v5?

Typically when people talk Corsi, they are talking 5v5.

 

[Jets012]

So yea that's why I get on Santini's case so much. When you literally are DEAD LAST in the whole league in the REL CF% I think that's a massive problem and a major indictment that you aren't ready for the minutes you're getting. He's also dead last in xG+/- too throughout the whole league. And it shows that he's near the bottom of our team in goal differential on the 5 on 5.

 

[devilsblood]

Sample size this bugs me. Like in baseball. I don’t pay attention to batting averages in April. They just don’t mean much because one odd bad or good game can send an average to a weird place that isn’t at all representave. There is too much noise. By August you have so many ABs that an outlier game doesn’t throw the numbers off and you can better judge who is hitting well and who isn’t (and not just that season but compared to multiple seasons). Actual trends emerge.

At what point can you say that here? How many games is enough to cancel out outliers and statistical noise to say without a doubt that this is a good corsi team and this isn’t?

Same with any stat really. Shooting percentage as well.

Well April is what 10%, while August 75%. So maybe 75%?

 

[Zippy316]

Sample size this bugs me. Like in baseball. I don’t pay attention to batting averages in April. They just don’t mean much because one odd bad or good game can send an average to a weird place that isn’t at all representave. There is too much noise. By August you have so many ABs that an outlier game doesn’t throw the numbers off and you can better judge who is hitting well and who isn’t (and not just that season but compared to multiple seasons). Actual trends emerge.

At what point can you say that here? How many games is enough to cancel out outliers and statistical noise to say without a doubt that this is a good corsi team and this isn’t?

Same with any stat really. Shooting percentage as well.

From what I've seen/read, 20 game benchmarks are usually a good start, but still small. Over a full season, almost all stats end up normalizing across the board.

I would say that goes for just about any stat.

 

[BenedictGomez]

I’ve watched every game this year and I feel confident in saying we are not a lotto team no batter what our Corsi is. I’ve watched a number of Edmonton games and I feel confident in saying they are not a good team and their good Corsi does not mean they are unlucky. They are just a bad team.

I'm guessing you're also not impressed that Montreal has the 3rd best Corsi in the entire NHL right now?

Behind Edmonton, the #2 best Corsi of course.

Both behind Carolina, the #1 ranked NHL Corsi team this year, owners of a 7-10 record.

 

[Devils731]

Hell NO I'm not only using it only 5v5! I'm using it for all situations.

If there's any predictive team value to this metric to begin with, why the hell would I eliminate all powerplay and penalty kill minutes, which is further weakening my data set?!?!?!?! By a lot.

That makes absolutely zero sense.

So you want to say a number has bad predictive value and you’ll prove that by using it in a way that people don’t use it? I mean come on.

This would be like me arguing about a players plus/minus but I also give players a plus on the power play and minus while penalty killing when nobody else looks at it that way.

 

[Zippy316]

So you want to say a number has bad predictive value and you’ll prove that by using it in a way that people don’t use it? I mean come on.

This would be like me arguing about a players plus/minus but I also give players a plus on the power play and minus while penalty killing when nobody else looks at it that way.

He's clearly incapable of looking past his ignorance at this point.

I wish I had better access to some of the stats I've seen showing CF% relation to team success.

 

[Emperoreddy]

I just don’t like seeing absolute claims made unless you got some serious concrete trends, and ironclad data to back it up.

 

[BenedictGomez]

If it proves to be a much better predictor of success why would you not use 5v5?

That's not what I said. You need to use it for ALL situations.

Either the base concept for the statistic works or the base concept for the statistic doesn't work. This is a logical test.

If you believe in this metric, it makes zero sense to discard all the minutes played on the PP or PK. This isn't something I should have to explain.

 

[BenedictGomez]

He's clearly incapable of looking past his ignorance at this point.

I wish I had better access to some of the stats I've seen showing CF% relation to team success.

Yes, it's working great so far this year.

Ohhhh...... I'm sorry, I forgot. ALL outliers!!!!!

I'm guessing you're also not impressed that Montreal has the 3rd best Corsi in the entire NHL right now?

Behind Edmonton, the #2 best Corsi of course.

Both behind Carolina, the #1 ranked NHL Corsi team this year, owners of a 7-10 record.

 

[Jets012]

I just don’t like seeing absolute claims made unless you got some serious concrete trends, and ironclad data to back it up.

What absolute claims have been made?

 

[devilsblood]

That's not even an "argument", it's a fact.

If you're going to buy into this crap, Fenwick without a doubt is better than Corsi, because it at least corrects a glaring Corsi error.

Which is why I say that the "Corsi people" often just want to sound smart, because if they really understood what they were talking about they'd be using Fenwick instead, yet almost nobody does.

I'm cool with corsi, I think the shot outweighs the block. Especially when you often see guys block a ton of shots because they can't clear the puck.

But if you want to judge defensive d-men using fenwick, then I can see that, but I think for most players corsi is better.

 

[Emperoreddy]

I'm guessing you're also not impressed that Montreal has the 3rd best Corsi in the entire NHL right now?

Behind Edmonton, the #2 best Corsi of course.

Both behind Carolina, the #1 ranked NHL Corsi team this year, owners of a 7-10 record.

Haha not in the slightest.

I imagine some of this evens out by the end of the year, but the problem is no one looks back at game 20 and their predictive articles claiming X team is do for a break out because corsi and admit they were wrong and said team isn’t even a top corsi team in the end.

 

[NJDevs26]

That's not what I said. You need to use it for ALL situations.

Either the base concept for the statistic works or the base concept for the statistic doesn't work. This is a logical test.

If you believe in this metric, it makes zero sense to discard all the minutes played on the PP or PK. This isn't something I should have to explain.

Well in fairness it's a LOT easier to have good CF/whatever stats when you're on the PP and a lot harder when on the PK. That isn't elementary. Adding those stats in doesn't really factor in how often we're on the PP and PK compared to other teams cause that differs by the team. The more accurate comparison would be comparing our average PP/PK advanced stat percentages to other team's average PP/PK percentages.

 

[devilsblood]

That's not what I said. You need to use it for ALL situations.

Either the base concept for the statistic works or the base concept for the statistic doesn't work. This is a logical test.

If you believe in this metric, it makes zero sense to discard all the minutes played on the PP or PK. This isn't something I should have to explain.

But 5v5 is what people use. Why do they use it? Because it's proven to be good predictor of success.

But if you want to use the stat that is not as good, and then bitch about it, then OK.

 

[NJDevs26]

I'm cool with corsi, I think the shot outweighs the block. Especially when you often see guys block a ton of shots because they can't clear the puck.

But if you want to judge defensive d-men using fenwick, then I can see that, but I think for most players corsi is better.

That's another problem I have with the advanced stats, when shot blocks have automatically become a BAD thing. It's not always indiciative of being unable to clear the puck. Sometimes puck entry's gained, a shot's blocked and it leads to a clear. Or just cancels out a good scoring opportunity before the next faceoff. Would people prefer not being able to block shots and having goalies see more rubber? I mean any shot block by definition cancels a scoring chance but it's come to be seen as weakness which is mind-boggling.

 

[devilsblood]

I just don’t like seeing absolute claims made unless you got some serious concrete trends, and ironclad data to back it up.

80% over the last 3 years is pretty good.

 

[BenedictGomez]

I'm guessing you're also not impressed that Montreal has the 3rd best Corsi in the entire NHL right now?

Behind Edmonton, the #2 best Corsi of course.

Both behind Carolina, the #1 ranked NHL Corsi team this year, owners of a 7-10 record.

The inversion of this is fun too right now.

The 3 WORST Team Corsi in the NHL right now?


#29 New Jersey Devils - #3 best record in the entire NHL (tied with Winnipeg)
#30 Colorado Avalanche - Right in the middle of the NHL pack
#31 Winnipeg Jets - #3 best record in the entire NHL (tied with NJ)


Statistical outliers EVERYWHERE!!!!

It's winning the PowerBall and MegaMillions odds crazy I tell you!!!!

 

[devilsblood]

That's another problem I have with the advanced stats, when shot blocks have automatically become a BAD thing. It's not always indiciative of being unable to clear the puck. Sometimes puck entry's gained, a shot's blocked and it leads to a clear. Or just cancels out a good scoring opportunity.

I don't think people are saying they are automatically bad.

But when you see a Volchenkov chugging around unable to clear the puck and racking up blocks? Welp maybe it's not the greatest stat.

 

[devilsblood]

The inversion of this is fun too right now.

The 3 WORST Team Corsi in the NHL right now?


#29 New Jersey Devils - #3 best record in the entire NHL (tied with Winnipeg)
#30 Colorado Avalanche - Right in the middle of the NHL pack
#31 Winnipeg Jets - #3 best record in the entire NHL (tied with NJ)


Statistical outliers EVERYWHERE!!!! It's PowerBall and MegaMillions crazy I tell you!!!!

Why are you continuing to roll out the small sample examples?

We've played 17 games. Gibbons is our leading goal scorer.

 

[Zippy316]

I just don’t like seeing absolute claims made unless you got some serious concrete trends, and ironclad data to back it up.

I don't have much time to compare, but it seems like last year might be changing the trend a bit.

A few teams at the top had high CF% but awful PDO which is likely why they didn't make the playoffs. Carolina was befallen by terrible goaltending and low shooting %. Rangers had a bottom five CF% but an extremely high PDO which makes sense. Anyone that watched them last year could tell they were a counter attack team that was opportunistic, same goes for Ottawa.

It should be interesting to track this year to see if teams are starting to take on the Ottawa/Pittsburgh way of playing a speed based game off the counterattack/rush and generating higher quality chances, but historically, CF% has been a pretty good indicator of team success. I can remember some ridiculous stats back from 2012-2014 where almost every team from the top 15 aside from 1/2 were in the playoffs.

Here is last years which seems to invalidate the trend a bit:

Team Advanced Stats Finder | Hockey-Reference.com

Here is 2012-13 which validates the trend outside of NJ but the Devils PDO was the worst in the league that year:

Team Advanced Stats Finder | Hockey-Reference.com

 

[NJDevs26]

I don't think people are saying they are automatically bad.

But when you see a Volchenkov chugging around unable to clear the puck and racking up blocks? Welp maybe it's not the greatest stat.

It always gets cited as being bad, and it's factored in as a negative in advanced stats. How many 'Volchenkovs' really exist in the NHL nowadays? I mean Santini and Larsson are sort of that kind of player but they can move a little better at least. You also have to be smart to block shots, especially if you're slow as **** and can't move laterally like Volch. But the slow guys are pretty much getting phased off most D's in the league anyway.

 

[Emperoreddy]

I’m more interested in figuring out how Pitt is tied for first in the division with a -15 GD.

 

[Devils731]

That's another problem I have with the advanced stats, when shot blocks have automatically become a BAD thing. It's not always indiciative of being unable to clear the puck. Sometimes puck entry's gained, a shot's blocked and it leads to a clear. Or just cancels out a good scoring opportunity before the next faceoff. Would people prefer not being able to block shots and having goalies see more rubber? I mean any shot block by definition cancels a scoring chance but it's come to be seen as weakness which is mind-boggling.

Shot blocks are bad if they pile up and I’m also giving up lots of shots. If I’m not giving up lots of shots and blocking a bunch of pucks that’s not so bad. If I’m not giving up many shots and not having to block many shots, that’s great.

Time my team spends giving up shots or blocking shots is time not spent in my teams offensive zone.

Also, bad breaks against my team can happen while blocking shots much more often than a good break for my team.