Confirmed Signing with Link: [ARI] Clayton Keller re-signs with Coyotes (8 Years, $7.15M AAV)

[Ziggdiezan]

Zone starts have a relatively trivial impact on raw possession statistics, considering most shifts start on the fly. Surely you know this?

Yup on the fly on like 70% of the total starts. Maybe more? However if a player never starts in the defensive zone, do you think it will be more likely that a player gets sent over the boards when their team is heading up ice or when they are chasing into their own zone?

Like I said for extreme cases (when some players are getting 70% oZS% while others are getting 40% oZS%) that difference in the non on the fly stats alone would be fairly significant for shot differential based stats.

 

[Kairi Zaide]

The effect of zone starts on player performance is extremely overrated. Furthermore, Keller posted very poor underlying metrics last season despite 74.51% offensive zone starts, so it’s not as if a bump up to 83.65% is going to move mountains for him. Just looking at every piece of evidence that we have, it seems very clear that Keller’s improvement by the underlying metrics is his own doing. And as mentioned above, his on-ice shooting percentage is very low, and is due for a rebound.

What is your evidence to suggest that RAPM does not do well at accounting for zone starts in extreme cases?

It's not just for zone starts, but any other extreme cases of factors that play into RAPM. Conclusions based on statistical linear regressions, when the population has a normal distribution, generally become somewhat flawed the further you are from the center of the distribution, because the linearity assumption is only fully valid close to the average.

 

[WhiskeyYerTheDevils]

Yup on the fly on like 70% of the total starts. Maybe more? However if a player never starts in the defensive zone, do you think it will be more likely that a player gets sent over the boards when their team is heading up ice or when they are chasing into their own zone?

Like I said for extreme cases (when some players are getting 70% oZS% while others are getting 40% oZS%) that difference in the non on the fly stats alone would be fairly significant for shot differential based stats.

1) that's not how line changes work lol. They don't decide which players are coming over the boards based on which direction the play is going lol.

2) Of the ~30% shifts that start with a faceoff, about 1/3 of those are neutral zone draws. So that offense zone % is only relevant for about 20% of a players shifts.

So even a 30% delta in zone starts between players really only means they get 3% more shifts starting in offensive zone and 3% fewer shifts starting in defensive zone.

 

[Ziggdiezan]

The effect of zone starts on player performance is extremely overrated. Furthermore, Keller posted very poor underlying metrics last season despite 74.51% offensive zone starts, so it’s not as if a bump up to 83.65% is going to move mountains for him. Just looking at every piece of evidence that we have, it seems very clear that Keller’s improvement by the underlying metrics is his own doing. And as mentioned above, his on-ice shooting percentage is very low, and is due for a rebound.

What is your evidence to suggest that RAPM does not do well at accounting for zone starts in extreme cases?

I'm talking purely about shot differential based stats largely because of the relative probability of a shot for vs shot against when starting in the offensive vs defensive zone.

All I was saying is comparing him to his peers with a standard deviation style evaluation for stats like CA/60 is not the best method considering his unique zone usage. I'm not saying he hasnt improved, he likely has.

As I said previously when I looked at players on the same team (so a lot of variables/modifiers they use in their regression analysis would be constant) the modifies applied for teammates with very high vs very low oZS% etc seemed rather insignificant for shot based stats (corsi). I think I came across it when comparing leaf defenders and how little the adjusted models modified corsi stats for defenders of opposite usage (Dermott vs Zaitsev).

 

[Black Kevin]

Say what you say about Keller, I would gladly take a chance on him. Even with this contract. I can see him live up to those dollars one day. Still a young guy and really talented.

 

[Ziggdiezan]

1) that's not how line changes work lol. They don't decide which players are coming over the boards based on which direction the play is going lol.

2) Of the ~30% shifts that start with a faceoff, about 1/3 of those are neutral zone draws. So that offense zone % is only relevant for about 20% of a players shifts.

So even a 30% delta in zone starts between players really only means they get 3% more shifts starting in offensive zone and 3% fewer shifts starting in defensive zone.

You don't think if a player is incredibly sheltered zone wise by the coach that it would influence even a little what players he sends over when a rush is developing. Maybe I'm a pessimist.

A difference of 30% is quite low for the extreme cases (percent difference of like 55% between guys getting 40% and 70% usage) assuming I'm understanding your use of delta which likely should be normalized.

Even using those numbers 3% more offensive and 3% less defensive zone starts would have a pretty decent impact when your comparing corsi values that are represented using standard deviations of small relative differences in value. These differences in the corsi values being compared are likely much lower than a percent difference of 15%.

Personally I prefer to look at unadjusted stats and get all the context of zone usage, QoC etc my self rather than looking at these regression heavy models. Lots and lots of assumptions built in.

 

[WhiskeyYerTheDevils]

You don't think if a player is incredibly sheltered zone wise by the coach that it would influence even a little what players he sends over when a rush is developing. Maybe I'm a pessimist.

A difference of 30% is quite low for the extreme cases (percent difference of like 55% between guys getting 40% and 70% usage) assuming I'm understanding your use of delta which likely should be normalized.

Even using those numbers 3% more offensive and 3% less defensive zone starts would have a pretty decent impact when your comparing corsi values that are represented using standard deviations of small relative differences in value. These differences in the corsi values being compared are likely much lower than a percent difference of 15%.

Personally I prefer to look at unadjusted stats and get all the context of zone usage, QoC etc my self rather than looking at these regression heavy models. Lots and lots of assumptions built in.

Assume 100 shifts:
Player 1 w/ 70% offensive zone starts:
70 shifts start on the fly
10 shifts start off neutral with zone faceoffs
14 shifts start off with offensive zone faceoffs
6 shifts start off with defensive zone faceoffs

Player 2 w/ 40% offensive zone starts:
70 shifts start on the fly
10 shifts start off neutral with zone faceoffs
8 shifts start off with offensive zone faceoffs
12 shifts start off with defensive zone faceoffs

That's about 6-7 games worth shifts, where player 1 gets all of 6 more offense zone starts.

That's not going to have a meaningful impact on possession statistics.

 

[sabrebuild]

So much zone talk.

NERDS!!!!

 

[Cor]

47 points last season.

On pace for 46 points this season

 

[Carlzner]

I don’t think the criticism is unwarranted. He’s been very, very underwhelming. I’m disappointed.

He may look underwhelming to you, a coyotes fan who watches every game. But JoeThorntonsRooster is here to provide you with graphs to prove he is in fact good, not bad.

 

[WhiskeyYerTheDevils]

On pace for 46 points this season

He led the coyotes in scoring last year, and is currently 2nd on the team in scoring this year, how is this being seen as a negative?

He'd be close to a point per game player if he played on a team like Toronto. But Arizona seems to know the value of defense. And it looks like it's working as they just won again.

 

[TomasHertlsRooster]

He may look underwhelming to you, a coyotes fan who watches every game. But JoeThorntonsRooster is here to provide you with graphs to prove he is in fact good, not bad.

There are some people whose eye tests are so horrible that they can’t even tell which players are on the ice when their team wins the Stanley Cup. That’s why it’s always good to use stats.

 

[Menzinger]

He led the coyotes in scoring last year, and is currently 2nd on the team in scoring this year, how is this being seen as a negative?

He'd be close to a point per game player if he played on a team like Toronto. But Arizona seems to know the value of defense. And it looks like it's working as they just won again.

Thats a fairly bold an unsubstantiated claim.

I personally like Keller quite a bit, but his offensive ability may just be lower than we thought

 

[WhiskeyYerTheDevils]

Thats a fairly bold an unsubstantiated claim.

I personally like Keller quite a bit, but his offensive ability may just be lower than we thought

Well he's currently scoring at the same rate as Kessel who was a point per game player in Pittsburgh last year.

Arizona doesn't play the type of system that produces high scoring forwards.

 

[Frolov 6'3]

Well he's currently scoring at the same rate as Kessel who was a point per game player in Pittsburgh last year.

Arizona doesn't play the type of system that produces high scoring forwards.

True, very much Kings hockey, I agree. People rarely attach importance to that.

But he doesnt separate himself from the rest either (you said he was their scoring leader). Derek Stepan and Vinny Hinostraza missed games last season. Now other players lead the Coyotes. That is why this is being seen as a negative. You have to look at the whole picture.

Soon Keller is getting paid $7M. As fan or team I would like to see more than 15 goals and 30 assists next year.

 

[Dekes For Days]

He'd be close to a point per game player if he played on a team like Toronto.

What a ridiculous statement. There is absolutely zero truth to this.

But Arizona seems to know the value of defense.

Since the beginning of last season:

CA/60: 24th
FA/60: 21st
xGA/60: 13th
SCA/60: 19th
HDCA/60: 23rd

What a defensive powerhouse.

For the record, Keller is 2nd last on the team in xGA/60 at ES and one of the worst in xGF% despite the most favourable deployment.

Well he's currently scoring at the same rate as Kessel who was a point per game player in Pittsburgh last year.

While Kessel likely will pick up the pace a bit, Kessel's raw production was boosted over the last couple years by PP production and playing with Malkin and Crosby. A decline in production was expected.

 

[Cotton]

I think the fact he signed for 7.1Mx8 says that he knows where his game is at and where he expects it to go.

7.1M is a reflection of what he'll become, not what he is. He's in the same talent pool as Nylander and Ehlers, who all make around the same money and who all produce about the same results.

 

[42]

Seems to be doing better offensively lately. 8 points in last 10 games. I haven't watched him so I'm just going by the numbers. His TOI is on the low side, though, at around 17 mins. For a player that is arguably the most offensively talented one on the team, one would think he'd be getting played more.

 

[Scintillating10]

Would Yotes trade Keller?

 

[42]

Would Yotes trade Keller?

Not likely at this time. With his contract and current level of play, his stock is low.

 

[Skinnyjimmy08]

I hope he bought his agent something real nice with that contract he got him.

I just never see him becoming a dominating offensive player

 

[Peasy]

Not even on pace for 50 points this year.

 

[slimbob8]

Not even on pace for 50 points this year.

I find that so shocking. I thought for sure this guy was gonna be a ppg player. He's still young so he probably has another level in him, but you can't help but be disappointed in his offensive numbers so far, especially after a 65 point rookie year.

 

[Qward]

Would Yotes trade Keller?

No need to necro bump. Just create a new thread. All you did was create confusion.

 

[Scintillating10]

No need to necro bump. Just create a new thread. All you did was create confusion.

Want as few threads as possible. Mods would just put them together. Would they trade Keller?