Winnipeg Jets select D Logan Stanley (1/18) Part II (Mod warning in OP)

[Whileee]

How is that a different discussion? The post you quote is my defense of that statement.

I could have said that Stanley should not have been drafted at all but passed over. I didn't. I said he is a long shot who should have been taken in the late rounds, like 5th or 6th. That is not 'overboard' at all. The typical success rate of players taken in those rounds matches what I think are the odds of Stanley becoming a successful pick.

Using at least one NHL draft pick value chart #18 and #79 are almost the same value as #22 and #36. Getting Luke Green at #79 puts further support to that, since I don't think he's much behind some of the D available at #36. I think it really comes down to what you think of Stanley. I would have been happy to pay that price to move up to #18 to get a player I liked. I just don't like Stanley.

 

[Grind]

How many very big junior defensemen in the past 15-20 years would have been first rounders based on projection models?

My concern with these models is that they are not very sensitive with large defensemen. Almost none of the best big defensemen in the NHL today would have been rated highly by production-based methods.

Has anyone actually gone back and used statistical methods to calculate the accuracy of the cohort-based projection methods for different classes of players using something like a kappa or J statistic? Seems to me that the method can't take credit for panning Boris Valabik and Jared Cowan without also owning missing out on almost every one of the top big D in the NHL (e.g. Myers and Weber and Chara).

Absolutely it would.

But these guys are so rare I don't think you COULD create a model that did predict them due to sample.

To test the sample you have to apply it against a different group then your control/set up.

So in other words you work backwards off the data available for the oldest big dman taken.

Then it in the future for the remaining ones.

We're talking about a sample of like 4 succesful players in the modern era (post 1995)

One of those gets burnt on your control test.

So now your down to something accurately finding 3 players.

That's simply too small of a grouping to get anything reliable.

Especially when the vast majority of 6'5+ dmen don't end up in the NHL anyways.

You could develop something that will correctly did find those 3 players using future looking data anyways (predictive not descriptive) but your still hanging your hat on three players.

That likelihood of that trend continuing is pretty rare.

But it would be very interesting to see and track going forward .

 

[truck]

Girard.

But I wouldn't have been super excited about either at 22.

The forward quality available is where the significant disconnect was, as evidenced by our similar pretty draft mock on this board.

I wasn't in love with any of the players that were available when they traded up. Some likes, no loves. The Jets had Stanley in theor top 10 though - I can understand their justification for trading up.

 

[truck]

No, no, no. Nobody says that. Of course there are exceptions. The trouble is there is no way to identify the exceptions. If there was they wouldn't be exceptions. They would be the expected results of some other rule.

The point, and the whole point is that the home run swing fails far too often relative to the successes to ever be a smart decision.

Stanley might be the next home run. He might be the next Chara. But they have no reason to believe that. It is just a blind guess and if it works out it will be blind luck. The worst thing is that if it works out it will lead them to make more bad decisions just like it because they will believe they were right.

Guys like those listed and like Stanley should be taken in the 5th or 6th round. Not the first and especially not trading up to get him. Good grief!

The bolded are flat out wrong. There are ways to identify exceptions - to what degree? I don't know, but I suspect it varies from case to case.

I am sure they do have a reason to believe that he will out-perform the basic statistical ranking. Is the reasoning sound? Not sure, but there is surely a reason.

As I said here:

They are pretty well all long shots outside of the top 15.

We are talking about the theoretical difference between 35%ish and 15%ish (pulled thise numbers out of my butt, but I think they are representative). This is based on a purely statistical model. We know we don't have a full data set. We don't what other variables were considered.

Would I have made this pick? No. I would tend to favor known odds, but I would also expect case by case variance. If the Jets believe his athleticism pushed his odds from 15 to 20 (or more) - that would be relevant.

We know that the Jets valued Stanley higher than many teams. We also know that we don't have all of the info as to why. A lack of info doesn't mean these inputs don't exist. It also doesn’t make the inputs invalid. Actual scouting adds value. Athleticism adds value. Were these things weighted too heavily? My guess is yes, but that doesn't mean theu weren't factors. It also doesn't mean they weren't valid.

 

[jetsfaneh]

I wasn't in love with any of the players that were available when they traded up. Some likes, no loves. The Jets had Stanley in theor top 10 though - I can understand their justification for trading up.

Goes to show how Winnipeg has a different scouting method than other clubs. Stanley top 10 probably wasn't that way on many draft boards.

There seems to be a premium put on improvement trajectory and what age that happens at. Perhaps their research shows that players that have a big upswing at a certain age is a good predictor of future development a la Scheifele (who's skating wasn't very good early on) where as others may prove to plateau and not exceed their draft positions.

Like Hillier said in his interview, if Stanley continues to develop at half the rate he did this past season they would be very happy with that. Lots of analytics being thrown around here which is great and all. But the Jets scouting department may have their own studies that they defer to which may make Stanley a prime candidate to potentially develop at a clip higher than most prospects.

There are lots of numbers and things that are considered with decision models. I fully expect that the Jets scouting department is no different here. And very much think that their decision model is what highlighted Logan Stanley as being a guy that they would fall in love with according to this model. Same way they did with Scheifele. Not predicting success here. But that pick worked out nicely.

 

[buggs]

Goes to show how Winnipeg has a different scouting method than other clubs. Stanley top 10 probably wasn't that way on many draft boards.

There seems to be a premium put on improvement trajectory and what age that happens at. Perhaps their research shows that players that have a big upswing at a certain age is a good predictor of future development a la Scheifele (who's skating wasn't very good early on) where as others may prove to plateau and not exceed their draft positions.

Like Hillier said in his interview, if Stanley continues to develop at half the rate he did this past season they would be very happy with that. Lots of analytics being thrown around here which is great and all. But the Jets scouting department may have their own studies that they defer to which may make Stanley a prime candidate to potentially develop at a clip higher than most prospects.

There are lots of numbers and things that are considered with decision models. I fully expect that the Jets scouting department is no different here. And very much think that their decision model is what highlighted Logan Stanley as being a guy that they would fall in love with according to this model. Same way they did with Scheifele. Not predicting success here. But that pick worked out nicely.

For the sake of discussion I'm not 100% convinced the Jets are fully into modeling anything. Just a gut feel but their tendencies at times seem to favor old school, Toe Blake - Eddie Shore style preferences. Stuart. Thorburn. Peluso. All active on a roster that has younger, better performing players seeing less icetime. Is that coaching? Is that philosophical to the organization? Something in between?

Scheifele is a great support to the trajectory argument. Mostly because it appears he has been a success for that argument. I wouldn't be surprised to find a similar view on other teams in the league and other kids that projected similarly to Scheifele but didn't work out quite the same way. As Garret has pointed out, everything aside from the trajectory argument points to Scheifele being in the tier he was drafted in anyway, so the result is less surprising. Stanley on the other hand fits the trajectory argument well, but wasn't in a complementary tier to his cohorts based on where he was drafted. That means the Stanley is a riskier pick than Scheifele even though they both fit a presumed development trajectory model (which isn't really a model I suspect but more of a mindset - models have data you can input and outcomes you can generate).

Individuals are just that, individuals. There are guys that work harder than Scheifele in the league. Perhaps not many but there will be some. Stanley will fall on the spectrum somewhere along the line. His current tendencies show that he's closer to the Scheifele line than the bottom and that's likely why the Jets drafted him. Fair enough. But what none of that shows is what his actual talent and more importantly limitations are. He has limited mobility and his skating is limited. True, both can be worked on. What's not true however is that hard work overcomes everything. It's a nice notion and great to tell kids but sometimes physiological factors can provide serious limitations to success. No idea if Stanley has them or not. But sometimes big kids don't turn out to be good, they're just big. It's a competitive advantage at younger ages and against lower levels of competition. It's much less of an advantage when the cohort at the next level is bigger and more skilled plus probably works as hard as you.

Again, really hope the pick works out and the naysayers are proven wrong, mostly because I think Stanley will face a lot of pressure just because of where he was drafted and that some comparables are not complementary (Valabik). Not really fair to the kid but welcome to the big leagues, fluffy bunny time is over.

 

[Gm0ney]

Depends on your definition of appeal to authority: assuming that a Doctor of Neurology can speak on world economic issues with some reliability simply because he's great in his field is an appeal to authority. Assuming that he can speak to issues revolving around Neurology is not necessarily an appeal to authority: he is an expert.

So IMO: it comes down to your assumption regarding the expert nature of either the pundits or the scouts. Personally, I take all opinions with a grain of salt, but it's not necessarily an appeal to authority if you assume that a professional who is paid to render their opinion within their field of expertise could be correct.

Scouts make a lot of mistakes, evidently. When the originators of PCS discovered they could outdraft half the teams in the league just by drafting the highest scoring CHL forward on the board, it really makes me question the expertise of scouting staffs and GMs around the league.

Additionally, unlike a neurosurgeon, who's had more than a decade of education and training - and who would've been academically top-tier even before devoting all that time to their craft - scouts are a collection of...dudes who played some hockey and spend a lot of time around the game? I might value a hockey loving neurosurgeon's opinion on Logan Stanley over a random NHL scout...

 

[Gm0ney]

To be clear, I'm not making reference to anyone in particular, nor especially to anyone who's actually watched him. If you have, well good for you. Smear away.

But just as many are willing to go along with whatever the Jets' scouting staff will do--about that you're very right--there are others who will blindly follow whatever the non-proprietary adjusted-scoring model du jour is.

It's analogous.

(For the record, yes--you beat me. I haven't watched him one bit. But I also haven't hazarded a guess about his ceiling or floor. Truthfully, I'm just curious as to why a team who've drafted high value scoring-model players over the last couple of years like Stanley so much. Very curious.)

I think it's because they truly value a tough stay-at-home defenseman in the mold of Mark Stuart. I think they believe there's a call for that kind of player.

I'm down with that style of D as long as he isn't a gong show in other aspects of the game. Those are rare finds, though. Maybe they really Stanley's the guy. Seems like a bad bet, but we'll find out...

 

[Whileee]

I wasn't in love with any of the players that were available when they traded up. Some likes, no loves. The Jets had Stanley in theor top 10 though - I can understand their justification for trading up.

I don't think they had Stanley in their top 10. This was just a throw-away comment from Lawless. Hillier and Chevy seemed to indicate that Stanley was more in the 15-17 range in their list.

 

[Whileee]

The bolded are flat out wrong. There are ways to identify exceptions - to what degree? I don't know, but I suspect it varies from case to case.

I am sure they do have a reason to believe that he will out-perform the basic statistical ranking. Is the reasoning sound? Not sure, but there is surely a reason.

As I said here:

We know that the Jets valued Stanley higher than many teams. We also know that we don't have all of the info as to why. A lack of info doesn't mean these inputs don't exist. It also doesn’t make the inputs invalid. Actual scouting adds value. Athleticism adds value. Were these things weighted too heavily? My guess is yes, but that doesn't mean theu weren't factors. It also doesn't mean they weren't valid.

I completely agree with this, truck. Well said.

 

[Whileee]

Absolutely it would.

But these guys are so rare I don't think you COULD create a model that did predict them due to sample.

To test the sample you have to apply it against a different group then your control/set up.

So in other words you work backwards off the data available for the oldest big dman taken.

Then it in the future for the remaining ones.

We're talking about a sample of like 4 succesful players in the modern era (post 1995)

One of those gets burnt on your control test.

So now your down to something accurately finding 3 players.

That's simply too small of a grouping to get anything reliable.

Especially when the vast majority of 6'5+ dmen don't end up in the NHL anyways.

You could develop something that will correctly did find those 3 players using future looking data anyways (predictive not descriptive) but your still hanging your hat on three players.

That likelihood of that trend continuing is pretty rare.

But it would be very interesting to see and track going forward .

How many 5'9 defensemen have been productive NHLers? What does that say for the accuracy of projection of a player like Sam Girard?

 

[Gm0ney]

How many 5'9 defensemen have been productive NHLers? What does that say for the accuracy of projection of a player like Sam Girard?

How many ever get a chance? Not saying that's the difference, but you've seen the hatred of Enstrom by many Jets fans - even though he's a very effective defenseman by any measure, people can't get over his size. Meanwhile 6'3" Cam Barker got 310 NHL games in and chance after chance...

 

[Whileee]

How many ever get a chance? Not saying that's the difference, but you've seen the hatred of Enstrom by many Jets fans - even though he's a very effective defenseman by any measure, people can't get over his size. Meanwhile 6'3" Cam Barker got 310 NHL games in and chance after chance...

Just wondering about the stability of the projection models at the margins of size.

 

[Evil Little]

The bolded are flat out wrong. There are ways to identify exceptions - to what degree? I don't know, but I suspect it varies from case to case.

I am sure they do have a reason to believe that he will out-perform the basic statistical ranking. Is the reasoning sound? Not sure, but there is surely a reason.

As I said here:



We know that the Jets valued Stanley higher than many teams. We also know that we don't have all of the info as to why. A lack of info doesn't mean these inputs don't exist. It also doesn’t make the inputs invalid. Actual scouting adds value. Athleticism adds value. Were these things weighted too heavily? My guess is yes, but that doesn't mean theu weren't factors. It also doesn't mean they weren't valid.

I think it's because they truly value a tough stay-at-home defenseman in the mold of Mark Stuart. I think they believe there's a call for that kind of player.

I'm down with that style of D as long as he isn't a gong show in other aspects of the game. Those are rare finds, though. Maybe they really Stanley's the guy. Seems like a bad bet, but we'll find out...

The bolded sum up my feelings, even if I'm a little more... cautiously interested.

 

[Whileee]

Absolutely it would.

But these guys are so rare I don't think you COULD create a model that did predict them due to sample.

To test the sample you have to apply it against a different group then your control/set up.

So in other words you work backwards off the data available for the oldest big dman taken.

Then it in the future for the remaining ones.

We're talking about a sample of like 4 succesful players in the modern era (post 1995)

One of those gets burnt on your control test.

So now your down to something accurately finding 3 players.

That's simply too small of a grouping to get anything reliable.

Especially when the vast majority of 6'5+ dmen don't end up in the NHL anyways.

You could develop something that will correctly did find those 3 players using future looking data anyways (predictive not descriptive) but your still hanging your hat on three players.

That likelihood of that trend continuing is pretty rare.

But it would be very interesting to see and track going forward .

I think the point is that the model has constraints at the upper size limits because so few make it, and the ones that do usually had very poor productivity at age 17. So the model isn't sensitive enough to pick up most of those that do make it.

I think that with Stanley, we'll be better able to project him after another year or two of development. I'm not optimistic, but I'm hopeful.

 

[Grind]

Just wondering about the stability of the projection models at the margins of size.

Obviously not great.

The difference is expanding your criteria.

For Girard players that are his size don't o an out frequently (though there are more then you may think)

But players that score like him do (regardless if size)

For guys like Stanley players like him based on size alone don't pan out a whole lot more then players based on size like Girard.(more but not significantly more)

As opposed to players that score like Stanley pan out dramatically dramatically less.

Your trying to project the proverbial needle in the haystack.

Your giving me headaches with your odd approach to this. I do t know if it's the communication or what but I have trouble trying to understand what your attempting to theorize other then trying to make some weird perfect confirmation biasing device?

 

[YWGinYYZ]

Scouts make a lot of mistakes, evidently. When the originators of PCS discovered they could outdraft half the teams in the league just by drafting the highest scoring CHL forward on the board, it really makes me question the expertise of scouting staffs and GMs around the league.

Additionally, unlike a neurosurgeon, who's had more than a decade of education and training - and who would've been academically top-tier even before devoting all that time to their craft - scouts are a collection of...dudes who played some hockey and spend a lot of time around the game? I might value a hockey loving neurosurgeon's opinion on Logan Stanley over a random NHL scout...

Oh I know. And I've looked at all the thought processes behind PCS, Sham Sharron, etc, and heavily slant my thinking towards a more progressive look at players. That said, I just find it lazy to blame someone's position on "appeal to authority", especially when it doesn't fit the classic mold.

I would definitely trust some neurosurgeon's to outdraft some scouts though.

 

[Puckatron 3000]

It seems to me Whileee is (successfully) arguing the established models for predicting NHL success should be treated with skepticism when applied to Stanley. Lack of data for players of his size.

In other words:
- People are mad because analytics give Stanley a bad chance to make the NHL.
- The analytics suck in this case.
- So, people are freaking out for no (or the wrong) reason.

 

[YWGinYYZ]

It seems to me Whileee is (successfully) arguing the established models for predicting NHL success should be treated with skepticism when applied to Stanley. Lack of data for players of his size.

In other words:
- People are mad because analytics give Stanley a bad chance to make the NHL.
- The analytics suck in this case.
- So, people are freaking out for no (or the wrong) reason.

I do think they have legitimate reasons to be concerned, but like truck, there were obviously reasons they felt justified in moving up to grab Stanley.

I will now grab my and cheer one of our new prospects on, my feelings on him be damned.

 

[Puckatron 3000]

I do think they have legitimate reasons to be concerned, but like truck, there were obviously reasons they felt justified in moving up to grab Stanley.

I will now grab my and cheer one of our new prospects on, my feelings on him be damned.

I agree with that.

It's reasonable to dislike Stanley because he can't score. Or because you distrust scouts' infatuation with super huge defensemen. Or because you liked other guys better.

It may be unreasonable to dislike Stanley because of his projected chances of making the NHL.

Interesting conversation, anyway. Reading this thread is a full-time job.

 

[garret9]

How many very big junior defensemen in the past 15-20 years would have been first rounders based on projection models?

My concern with these models is that they are not very sensitive with large defensemen. Almost none of the best big defensemen in the NHL today would have been rated highly by production-based methods.

Has anyone actually gone back and used statistical methods to calculate the accuracy of the cohort-based projection methods for different classes of players using something like a kappa or J statistic? Seems to me that the method can't take credit for panning Boris Valabik and Jared Cowan without also owning missing out on almost every one of the top big D in the NHL (e.g. Myers and Weber and Chara).

They are sensitive and the percentages land where the trend would be expected to fall if you didn't have information on that group. When a curve keeps falling by 2 y's for every x and after (6,5) you get (7,3) on the last data point, you don't feel as nervous about the numbers as you would if it didn't.

The n is not relatively small. Most small ns are guys who are exceptional and big like Laine. Or super young for a league like Werenski.

Most big men were lower scorers because big and high scoring is rare. They basically make it 100% but they rarely exist.

The issue with the models is they won't tell you who is the 15% vs 85% because luck and other factors play a role.

We know that scouts have some ability to identify the other factors, but we also know they overvalue those factors.

 

[garret9]

I don't think they had Stanley in their top 10. This was just a throw-away comment from Lawless. Hillier and Chevy seemed to indicate that Stanley was more in the 15-17 range in their list.

Hillier and Zinger did and were the loudest proponents for Stanley and Cederholm.

 

[pucka lucka]

An appeal to authority (argument from authority) is well defined. It's exactly what is happening here. An proclaimed expert isn't right because he's an expert. He's right if he's actually right, based on evidence. i can't believe we are actually trying to redefine logical fallacies now to support the management team.

 

[garret9]

Again, I also add that
1. Stanley has bad numbers aside from scoring
2. my opinion is that his eye test isn't good either

 

[YWGinYYZ]

An appeal to authority (argument from authority) is well defined. It's exactly what is happening here. An proclaimed expert isn't right because he's an expert. He's right if he's actually right, based on evidence. i can't believe we are actually trying to redefine logical fallacies now to support the management team.

Two things:

1) Who's doing this? I'm certainly not: I don't "support" the management team, other than to give credit or fling dung if they do well or do something stupid.
2) The definition of appeal to authority is more nuanced then you're espousing here. A proclaimed expert isn't necessarily an expert in any field other than the one they're an expert in.

You've obviously misread the point I'm trying to make.