Speculation: Jets General Rumour, Trade, Free Agent and Waiver Speculation 16-17 Part XVII

[puck stoppa]

He's a key part of the NHL's 28th-worst PK unit!

I'm okay with chiarot as our #7 next year. But we can't have him and Stu killing penalties. Hope morrissey, trouba, enstrom, Myers do more of that next year.

 

[Mortimer Snerd]

Chiarot was a defensive stalwart in 2015/16. No wonder the Jets penciled him into the top-6 for this season.

What does Beats Per Minute have to do with hockey?

OK, seriously what does BPM mean, O or D?

 

[pucka lucka]

I thought this was interesting:



You can sift through the Tweets explaining the Star rating but it's basically what defensemen get sent out to face the best players on the opposition teams. Enstrom at 20. Postma dead last (and Chiarot not much ahead of him in the "Bottom 20") of about 200-ish players.

So this is purely quantitative? bleh.

 

[pucka lucka]

What does Beats Per Minute have to do with hockey?

OK, seriously what does BPM mean, O or D?

Bungles Per Minute

 

[Mortimer Snerd]

I thought this was interesting:

You can sift through the Tweets explaining the Star rating but it's basically what defensemen get sent out to face the best players on the opposition teams. Enstrom at 20. Postma dead last (and Chiarot not much ahead of him in the "Bottom 20") of about 200-ish players.

So basically it is a measure of coaches' confidence levels in the various players?

 

[Mortimer Snerd]

Bungles Per Minute

OK, makes sense.

 

[Gm0ney]

So this is purely quantitative? bleh.

Are you one of The Binning People?

 

[Gm0ney]

So basically it is a measure of coaches' confidence levels in the various players?

Yeah. Dellow explains it pretty well in that Twitter thread if you read through it. Maurice thinks highly of Enstrom - he's getting sent out there vs. the other teams' star players. And he thinks Postma and Chiarot need to be sheltered mightily.

Maurice tends to lean hard on his favoured players though - Top 3 lines get 90% of the ES ice time. Competent defensemen are out there for 25 minutes per game...so it might make things look a little more extreme. But Postma generally gets really low minutes as it is - and Maurice is sure trying to keep him away from other teams' top lines...so it's not surprising he's dead last.

 

[puck stoppa]

Yeah. Dellow explains it pretty well in that Twitter thread if you read through it. Maurice thinks highly of Enstrom - he's getting sent out there vs. the other teams' star players. And he thinks Postma and Chiarot need to be sheltered mightily.

Maurice tends to lean hard on his favoured players though - Top 3 lines get 90% of the ES ice time. Competent defensemen are out there for 25 minutes per game...so it might make things look a little more extreme. But Postma generally gets really low minutes as it is - and Maurice is sure trying to keep him away from other teams' top lines...so it's not surprising he's dead last.

I'm convinced no matter who we have at the bottom, Pmo won't use them for 10-11 min, ever! That's his style.

 

[garret9]

Chiarot was a defensive stalwart in 2015/16. No wonder the Jets penciled him into the top-6 for this season.

You should keep reading where defensive BPM isn't even used in the WAR model because it adds no value.

BPM only is used for offensive side of WAR in both EV and PP.

As was hinted in the BPM write-up, the defensive version of the stat has basically zero value when compared to the information provided by XPM.

 

[Gabe Kupari]

I'm convinced no matter who we have at the bottom, Pmo won't use them for 10-11 min, ever! That's his style.

And possibly one of the reasons why at times the guys he plays, overplays look sluggish out there for stretches, people talk about Scheif and Wheeler and Buff this season at times saying whats up there, why are they struggling, part of it is probably because they are on the ice way 2 much. Our 4th line plays 5 minutes a night MAYBE...

Wheeler and Scheif are something like 2 of the top 10 or so forwards in the league in ice time, Buff leads the league in ice time doesn't he? per game? or darn close.

Maurice overplays guys, he does it with the goalies also, running a goalie 15 16 games in a row is crazy stuff. Teams just don't do that anymore.

 

[CaptainChef]

I'm convinced no matter who we have at the bottom, Pmo won't use them for 10-11 min, ever! That's his style.

That and having specified roles for third & fourth liners. Might have worked back in the day -- today it SHOULD get a coach a quick ticket to the exit ramp (but it likely won't)

 

[garret9]

What does Beats Per Minute have to do with hockey?

OK, seriously what does BPM mean, O or D?

DTMAH made two models: Boxscore Plus-Minus (BPM) and Expected Plus-Minus (XPM)

BPM looks at a player's boxscore statistics and places a value to them. For example: the even strength offensive values in boxscore stats weigh a player's goals, primary assists, secondary assists, unblocked shots, expected shooting percentage, and quality of teammate.

XPM is like a super Corsi. It adjusts for shot quality factors like expected goals but also adjusts for usage (all 9 other players on the ice for each event, coach of both teams, score of game, schedule (back-to-backs), and zone starts). The actual rate of events matters as well, and it will also heavily reward/penalize for good/bad shifts.

Putting an O or D in front of either model means you are only looking at the offensive side of things.


DTMAH combines these numbers to create his WAR.
EV Offense is a blend of OBPM and OXPM, while EV Defense is DXPM exclusively since the DBPM seems to cary no real value.
PP Offense is currently a blend of both OBPM and OXPM but weighted more heavily towards the former.


DTMAH then added some additional factors, with face-offs, penalties taken, and penalties drawn.

 

[Mortimer Snerd]

DTMAH made two models: Boxscore Plus-Minus (BPM) and Expected Plus-Minus (XPM)

BPM looks at a player's boxscore statistics and places a value to them. For example: the even strength offensive values in boxscore stats weigh a player's goals, primary assists, secondary assists, unblocked shots, expected shooting percentage, and quality of teammate.

XPM is like a super Corsi. It adjusts for shot quality factors like expected goals but also adjusts for usage (all 9 other players on the ice for each event, coach of both teams, score of game, schedule (back-to-backs), and zone starts). The actual rate of events matters as well, and it will also heavily reward/penalize for good/bad shifts.

Putting an O or D in front of either model means you are only looking at the offensive side of things.


DTMAH combines these numbers to create his WAR.
EV Offense is a blend of OBPM and OXPM, while EV Defense is DXPM exclusively since the DBPM seems to cary no real value.
PP Offense is currently a blend of both OBPM and OXPM but weighted more heavily towards the former.


DTMAH then added some additional factors, with face-offs, penalties taken, and penalties drawn.

Thanks Garret.
Couldn't decipher the terms.
XPM in particular sounds interesting but WAR is still the top aggregate stat?

Coming up with a single stat that accurately describes a player's value has to be the 'holy grail' for analysts. When I see these new models being used I always have to wonder how well they are actually performing. We know Corsi has been demonstrated effective/predictive to a certain level because it has a history of performing. How well verified is WAR?

 

[garret9]

Thanks Garret.
Couldn't decipher the terms.
XPM in particular sounds interesting but WAR is still the top aggregate stat?

Coming up with a single stat that accurately describes a player's value has to be the 'holy grail' for analysts. When I see these new models being used I always have to wonder how well they are actually performing. We know Corsi has been demonstrated effective/predictive to a certain level because it has a history of performing. How well verified is WAR?

DTM's WAR is used combining XPM and OBPM in different ways depending on a few things:
1) whether even strength or special teams
2) whether forward or defender being evaluated
3) whether offense or defense

Testing for WAR and its components can be found here: https://hockey-graphs.com/2016/10/28/testing-and-final-remarks/

 

[Mortimer Snerd]

DTM's WAR is used combining XPM and OBPM in different ways depending on a few things:
1) whether even strength or special teams
2) whether forward or defender being evaluated
3) whether offense or defense

Testing for WAR and its components can be found here: https://hockey-graphs.com/2016/10/28/testing-and-final-remarks/

Thanks. Keep up the good work here Garret. Much appreciated.

 

[Whileee]

What does Beats Per Minute have to do with hockey?

OK, seriously what does BPM mean, O or D?

BPM = "Box Plus-Minus"

Box Plus-Minus (BPM) is a box score-based metric for evaluating a hockey player’s quality and contribution to the team. It is very different than an Expected Plus-Minus type model, which is a play-by-play regression metric. BPM relies on a player’s box score information to estimate a player’s performance relative to replacement level.

The BPM metric is comprised of a few variables that have been selected by a machine learning modeling exercise as predictive of offensive and defensive outputs, expressed as "plus-minus" outcomes.

In my post above, I have shared the charts that show the variables and weights for the offensive and defensive BPM (OBPM and DBPM). As noted above, the DBPM estimates are based on several measures like take-aways and give-aways, hits taken, hits, blocks, etc. Several of the parameters are adjusted by possession stats.

 

[Whileee]

You should keep reading where defensive BPM isn't even used in the WAR model because it adds no value.

BPM only is used for offensive side of WAR in both EV and PP.

Yeah, DBPM doesn't have much value, but interesting to see how it gets constructed.

I found this interesting, with respect to the XPM metric.

Through the use of ridge regression and Bayesian techniques, Expected Plus-Minus is a step forward into analyzing shot attempt numbers. While they do not allow us to isolate the cause of each player’s rating, whether that be through their teammates or competition, they do provide a more accurate representation of a player’s value than more commonly used methods.

If neither XPM nor OBPM "adjust" for quality of teammate (in fact OBPM actually boosts players who have high QoT), are there methods that try to adjust for QoT, or is it assumed that players who play with higher quality of teammates are better players (i.e. coaches make rational decisions about usage), which is implicit in the OBPM, and perhaps the XPM?

 

[Flair Hay]

He's a key part of the NHL's 28th-worst PK unit!

Just saying.

When a guy is top 1/3 of the league in all shots against and goals against categories...

...On the 28th ranked PK...

He's not Stuart, who sucks on PK too. Maybe he's not facing the first power plays? I dunno. But he's been our best PK defenseman this year.

 

[Whileee]

DTMAH made two models: Boxscore Plus-Minus (BPM) and Expected Plus-Minus (XPM)

BPM looks at a player's boxscore statistics and places a value to them. For example: the even strength offensive values in boxscore stats weigh a player's goals, primary assists, secondary assists, unblocked shots, expected shooting percentage, and quality of teammate.

XPM is like a super Corsi. It adjusts for shot quality factors like expected goals but also adjusts for usage (all 9 other players on the ice for each event, coach of both teams, score of game, schedule (back-to-backs), and zone starts). The actual rate of events matters as well, and it will also heavily reward/penalize for good/bad shifts.

Putting an O or D in front of either model means you are only looking at the offensive side of things.


DTMAH combines these numbers to create his WAR.
EV Offense is a blend of OBPM and OXPM, while EV Defense is DXPM exclusively since the DBPM seems to cary no real value.
PP Offense is currently a blend of both OBPM and OXPM but weighted more heavily towards the former.


DTMAH then added some additional factors, with face-offs, penalties taken, and penalties drawn.

A couple of times in this thread you've mentioned that the OBPM and XPM "adjusts" for the quality of linemates. I haven't found as much information about the XPM model, but the note in the article I cited above seems to indicate that it doesn't adjust for QoT or QoC.

Through the use of ridge regression and Bayesian techniques, Expected Plus-Minus is a step forward into analyzing shot attempt numbers. While they do not allow us to isolate the cause of each player’s rating, whether that be through their teammates or competition, they do provide a more accurate representation of a player’s value than more commonly used methods.

The OBPM doesn't seem to adjust for QoT, but rather uses it as an instrumental variable in computing the OBPM. That means that if you play with higher QoT, your OBPM is higher than if you play with lower QoT. That is the opposite of "adjusting" for QoT.

Can you help to clarify?

 

[bigbuff33]

What do you guys think about acquiring orlov as our 2 pair lhd?

 

[Mortimer Snerd]

What do you guys think about acquiring orlov as our 2 pair lhd?

He is a left shot but plays RHD for Caps. His numbers look very good. He is an arb eligible RFA. No idea what it would take to sign him. He might not be too happy about leaving a contender to come here. Then there is the price Caps would want for him.

I'd be all in favour of getting him but I'm not sure it would work out or be affordable.

 

[puck stoppa]

Not sure where to put this but with Strait going down that means that we have to recalls left from the Moose for the rest of the year? Hope it Connor Roslo for the last two games.

 

[10Ducky10]

The roster is unlimited so why could we not call both of them up and the goalie?

 

[ginner classic]

My one issue with Schneider is his age. He's now 31 and is also signed to a very long term big money deal (He's worth every penny now). I think we need to ask does his age mean that he will still be an elite goalie through much of our window or are we buying him at the end of his peak and he declines through his mid 30s as we are contending.

I think he's a terrific goalie but trading our first this year for him is a bit risky due to his age. I'd just as soon grab another A prospect and look to FA or a younger guy like Ratna in a trade.

Canuck fan....

Schneider is easily worth your 1st this year. The Canucks were backed into a corner and made a last minute decision to trade Schneider instead of Luongo. New Jersey got a steal. If goalfending is the number 1 issue and the prospect pipeline is in good shape....yeah...absolutely it's worth it. Schneider was also painted as an elite backup that had not proven anything.

New Jersey would want more IMO.