Prospect Info: 2018 NHL Draft

Status
Not open for further replies.

njx9

Registered User
Feb 1, 2016
2,161
340
Lashof & McIrath may be slow but they are not untalented.

Compared to what? The guys in Toledo? Beer leaguers? Neither is NHL-talented, in any way, shape or form, and given that, the team should be absolutely open to moving on from either or both the instant it has a quality prospect ready to take their minutes.

As it relates to the draft, it just gets back to the argument that there's plenty of room for prospects if the team prioritizes prospect development over giving as much time as possible to AHL-lifers.
 
  • Like
Reactions: Fire Ken Holland

Zetterberg4Captain

Registered User
Aug 11, 2009
13,730
2,136
Detroit
Who cares about pre-draft rankings? I'm quite sure best drafting organizations go with heavily different lists than same "averages of all averages".

Usually only TOP3 is pretty much in line.

I care

The wings are not hitting homeruns ever year(nor should they be).

They need to keep it simple

Those predraft lists are not just some basement gremlins thoughts or wishes.

Dont go reaching early.

At pick 33 take the guy everyone else has ranked from 25-40, not a guy presumed to be at 88

Once the cupboard is stocked full with enough talent then start reaching
 

waltdetroit

Registered User
Jul 20, 2010
2,649
526
Lashof & McIrath are good AHL players. Lashof & Renny formed one of the best AHL D lines last year and not just my opinion. If you think having good vets on the ice (like player coaches) for the young players is nonsense, you are entitled to your opinion. But all of the NHL & AHL teams operate that way for good reasons.
 

Rzombo4 prez

Registered User
May 17, 2012
5,980
2,698
I care

The wings are not hitting homeruns ever year(nor should they be).

They need to keep it simple

Those predraft lists are not just some basement gremlins thoughts or wishes.

Dont go reaching early.

At pick 33 take the guy everyone else has ranked from 25-40, not a guy presumed to be at 88

Once the cupboard is stocked full with enough talent then start reaching

Yeah, we should just fire our whole scouting department and just draft off of Bob's list or one of the other public scouting services.
 
  • Like
Reactions: lilidk and Ezekial

jkutswings

hot piss hockey
Jul 10, 2014
10,853
8,578
Yeah, we should just fire our whole scouting department and just draft off of Bob's list or one of the other public scouting services.
Or we could not jump to the drama of extremes, and acknowledge that somewhere in between "just rankings" and "completely outside the box" makes more sense.
 

Zetterberg4Captain

Registered User
Aug 11, 2009
13,730
2,136
Detroit
Yeah, we should just fire our whole scouting department and just draft off of Bob's list or one of the other public scouting services.

Is this an uneducated joke?

These lists compile lots of opinions from across the hockey spectrum.

To think the DRW scouting staff is the be all and end all is at best ignorant
 

Bench

3 is a good start
Aug 14, 2011
21,202
14,898
crease
Yeah, we should just fire our whole scouting department and just draft off of Bob's list or one of the other public scouting services.

Sarcasm aside...

I'd actually be interested to see how a computer learning algorithm could do drafting. Right now they have AI that can sort through medical records to diagnose rare cancers. Surely if you can train IBM's Watson to do that, you could input parameters for successful drafting outcomes. The computer could then sort through literally thousands of points of data and generate a list of prospects that meet criteria for most likely to succeed. It's like Moneyball but on steroids. And since scouting is such a crapshoot anyway, this could increase fidelity into your picks dramatically.

OK, so... who wants to start a new business with me?
 

TheMule93

On a mule rides the swindler
May 26, 2015
12,466
6,512
Ontario
Sarcasm aside...

I'd actually be interested to see how a computer learning algorithm could do drafting. Right now they have AI that can sort through medical records to diagnose rare cancers. Surely if you can train IBM's Watson to do that, you could input parameters for successful drafting outcomes. The computer could then sort through literally thousands of points of data and generate a list of prospects that meet criteria for most likely to succeed. It's like Moneyball but on steroids. And since scouting is such a crapshoot anyway, this could increase fidelity into your picks dramatically.

OK, so... who wants to start a new business with me?

What kind of hardware do you need for that
 
Last edited:

HisNoodliness

The Karate Kid and ASP Kai
Jun 29, 2014
3,652
2,022
Toronto
Sarcasm aside...

I'd actually be interested to see how a computer learning algorithm could do drafting. Right now they have AI that can sort through medical records to diagnose rare cancers. Surely if you can train IBM's Watson to do that, you could input parameters for successful drafting outcomes. The computer could then sort through literally thousands of points of data and generate a list of prospects that meet criteria for most likely to succeed. It's like Moneyball but on steroids. And since scouting is such a crapshoot anyway, this could increase fidelity into your picks dramatically.

OK, so... who wants to start a new business with me?
A year ago or so (I'm sorry, I'm not willing to go digging for it), An article highlighted that you'd be more successful than NHL teams at drafting if you drafted purely based on point totals. So if you were to put even an iota of work into actual analytics I'm sure you could greatly outdo NHL scouts. I don't even think it would be very hard. I however do think that you'd have a hard time convincing people without ten years of very successful predictions. GMs are mostly former hockey players. Chosen for their brawn and not brains.
 

Rzombo4 prez

Registered User
May 17, 2012
5,980
2,698
Sarcasm aside...

I'd actually be interested to see how a computer learning algorithm could do drafting. Right now they have AI that can sort through medical records to diagnose rare cancers. Surely if you can train IBM's Watson to do that, you could input parameters for successful drafting outcomes. The computer could then sort through literally thousands of points of data and generate a list of prospects that meet criteria for most likely to succeed. It's like Moneyball but on steroids. And since scouting is such a crapshoot anyway, this could increase fidelity into your picks dramatically.

OK, so... who wants to start a new business with me?

I am pretty sure that is what the Panthers were trying to do when they fired Tallon. I am sure you could develop something very workable if you really wanted to. It will be interesting to see how these initial models hold up long term as the game changes.

As for the sarcastic comment, we aren't going to out perform the market by following the crowd (which seems to be everyone's big hang up). Successful drafting requires a ton of subjective projection (at least until your AI project fully replaces it). Opinions of players are going to differ greatly, especially once you get past the first round. At that point, there is no consensus to even follow. I don't want our scouts to make popular choices, I want them to make the right pick more often than not, reach or not.
 

Flowah

Registered User
Nov 30, 2009
10,249
547
And since scouting is such a crapshoot anyway, this could increase fidelity into your picks dramatically.
I really think it is. And I think people have to get used to that possibility that scouts aren't worth much currently.

Over 1100 forwards were drafted from 2005-2015 from the 2nd round to the end of the draft. Since then, only 39 of those 1100 have both played in 41+ games and been over .5 PPG. Which in the grand scheme of things I think is a pretty low bar. For reference, Tatar is .543 and Nyquist is .586. Good enough players I guess.

And it's not even like the rest of the draft is pulling down rounds 2 and 3. Fewer than 9% of the 2nd rounders and 6% of the 3rd rounders taken at forward from 2005-2015 ever went above .5 PPG. These are some sad, sad numbers if you're asking me to believe the scouting departments of pro-teams are doing more than mostly guessing and getting lucky once in a while.

Sad fact of the day. Since 2005, the Wings have failed to draft a single defenseman in any round that met or outpaced that round's average PPG for defensemen. We're not bad at forward. We are better than league average. But on defense it would be hard for us to get any worse.
 

Bench

3 is a good start
Aug 14, 2011
21,202
14,898
crease
What kind of hardware do you need for that

It's a pretty extensive hardware setup. More than you'd want to create in your garage, for example. Multiple server racks. It would depend how quickly you want the information. But you could probably rent the server power required.

How deep you go down this rabbit hole also changes things quite a bit. If we're just talking some simple statistical cross referencing that you determine, it's probably relatively simple, and likely already being done by some franchises. What I'm suggesting would take significantly more work, in that you're training the computer to arrive at new conclusions, not only the conclusions you've input as desired.

So in a perfect system, the computer starts to recognize, independently, "Oh hey, players from league X with Y reach and Z points per game are performing above average for their draft position." That might seem simple, but what a computer can do is quickly run evaluations if these kinds of findings are statistically significant. One area that sports stats seems to lag behind science, and often due to sample sizes and lack of rigor, is checking for significance. Only then can you know if something is within variance. And for a human to sit down and run all these checks for every, tiny, little, piece of data... would be overwhelming. In the example above you have 3 points of data cross referenced with every prospect in every league with all known data. That's a lot. But imagine if you could find some kind of significance in 16 points of data. That's the level of insight you'd be hard pressed to have any human be able to sort.

It seems crazy to me NHL teams aren't developing these tools, or if they are, they are keeping it quiet. Of course, we read only a few years ago that Babcock and Holland only partially look at advanced stats (which I can be critical of due to the aforementioned lack of rigor). If they barely take those into account, what are the odds they'd dump millions in resources developing cutting edge statistical algorithms? It will take a team being a thought leader, and having success, before the old boys catch up.

This is all a fun thought experiment. I only brought it up because I know for a fact this will revolutionize medical diagnoses, so... why not sports projections?
 
  • Like
Reactions: jkutswings

odin1981

There can be only 1!
Mar 8, 2013
5,033
881
Canton Mi
Sarcasm aside...

I'd actually be interested to see how a computer learning algorithm could do drafting. Right now they have AI that can sort through medical records to diagnose rare cancers. Surely if you can train IBM's Watson to do that, you could input parameters for successful drafting outcomes. The computer could then sort through literally thousands of points of data and generate a list of prospects that meet criteria for most likely to succeed. It's like Moneyball but on steroids. And since scouting is such a crapshoot anyway, this could increase fidelity into your picks dramatically.

OK, so... who wants to start a new business with me?

I read in a article the other day how there are ai programs interpreting legal cases and such already. So basically ai is already up to phd level stuff atm. Wouldn't surprise me if in a few years people try that with drafting in pro sports either somehow.
 

Rzombo4 prez

Registered User
May 17, 2012
5,980
2,698
I really think it is. And I think people have to get used to that possibility that scouts aren't worth much currently.

Over 1100 forwards were drafted from 2005-2015 from the 2nd round to the end of the draft. Since then, only 39 of those 1100 have both played in 41+ games and been over .5 PPG. Which in the grand scheme of things I think is a pretty low bar. For reference, Tatar is .543 and Nyquist is .586. Good enough players I guess.

And it's not even like the rest of the draft is pulling down rounds 2 and 3. Fewer than 9% of the 2nd rounders and 6% of the 3rd rounders taken at forward from 2005-2015 ever went above .5 PPG. These are some sad, sad numbers if you're asking me to believe the scouting departments of pro-teams are doing more than mostly guessing and getting lucky once in a while.

Sad fact of the day. Since 2005, the Wings have failed to draft a single defenseman in any round that met or outpaced that round's average PPG for defensemen. We're not bad at forward. We are better than league average. But on defense it would be hard for us to get any worse.

Is that an indictment of scouting or of the overall lack of talent outside of the first round? I don't think anyone ever claimed that drafting 18-year-olds into a pro sports league was easy.
 
  • Like
Reactions: njx9
Jul 30, 2005
17,660
4,588
I mean, what is location, really
It's a pretty extensive hardware setup. More than you'd want to create in your garage, for example. Multiple server racks. It would depend how quickly you want the information. But you could probably rent the server power required.

How deep you go down this rabbit hole also changes things quite a bit. If we're just talking some simple statistical cross referencing that you determine, it's probably relatively simple, and likely already being done by some franchises. What I'm suggesting would take significantly more work, in that you're training the computer to arrive at new conclusions, not only the conclusions you've input as desired.

So in a perfect system, the computer starts to recognize, independently, "Oh hey, players from league X with Y reach and Z points per game are performing above average for their draft position." That might seem simple, but what a computer can do is quickly run evaluations if these kinds of findings are statistically significant. One area that sports stats seems to lag behind science, and often due to sample sizes and lack of rigor, is checking for significance. Only then can you know if something is within variance. And for a human to sit down and run all these checks for every, tiny, little, piece of data... would be overwhelming. In the example above you have 3 points of data cross referenced with every prospect in every league with all known data. That's a lot. But imagine if you could find some kind of significance in 16 points of data. That's the level of insight you'd be hard pressed to have any human be able to sort.

It seems crazy to me NHL teams aren't developing these tools, or if they are, they are keeping it quiet. Of course, we read only a few years ago that Babcock and Holland only partially look at advanced stats (which I can be critical of due to the aforementioned lack of rigor). If they barely take those into account, what are the odds they'd dump millions in resources developing cutting edge statistical algorithms? It will take a team being a thought leader, and having success, before the old boys catch up.

This is all a fun thought experiment. I only brought it up because I know for a fact this will revolutionize medical diagnoses, so... why not sports projections?
This is all stuff that would be fairly easy to do. Like, it's fairly surface-level data science/machine learning stuff that you could get a grad student to do, at least to get a working approximation. So the computation and data mining are not really the trouble. The big trouble is in getting data collection right. Lower-level leagues and teams just don't have the will or even ability to collect the kinds of data you need to get the kind of specificity you need. And we also have some pretty crude measurements that can't account for some of the more important aspects of the eye-test, like gap control.

But, also, if we had the data, we probably wouldn't need to do machine learning, anyway. We could just fire up a statistics package and go to town. It would be helpful to be able to use other methods, but I expect human models could get pretty close. And actually, there's a guy over in the prospects forum who is attempting to do this. He's got his own homebrew model that he uses to generate rankings. He tweaks it a little bit every year, but he isn't wowing anybody quite yet.

Medicine is a lot more complex, because so much of that is contextual, and then you have the same problem again: data collection can be hard. HIPAA throws a huge wrench into that whole process. Similar issues with law. But if a kid scored 20 ES goals, he just did. That's all.
 
Last edited:
Jul 30, 2005
17,660
4,588
I mean, what is location, really
As for the sarcastic comment, we aren't going to out perform the market by following the crowd (which seems to be everyone's big hang up). Successful drafting requires a ton of subjective projection (at least until your AI project fully replaces it).
I don't follow. Just because you aren't following the crowd doesn't mean you're doing something meaningful. There are plenty of totally wrong moves that don't follow the crowd, and we usually label those as aggressively stupid. So just because the Wings aren't following the crowd doesn't mean they're doing the right thing.

But on top of that, why are we starting at "out-perform the market"? The Wings are underperforming it right now. It might be asking a little much of an underachieving scouting group to become worldbeaters overnight. It seems to me that it would be a smarter move to follow the crowd, because at least you know you're going to get roughly average returns. It seems to me that your post assumes the crowd is doing it wrong, but that doesn't actually seem to be true. It doesn't take a lot of imagination to figure the crowd is doing what they're doing because they see it working in the NHL and try to draft more of that.

I mean, who's really doing it wrong? The team who drafts what is proven to work right now or the team who drafts for what could work in the future, depending on what direction the league goes? I know which one I'd pick. Drafting for a style of play that does not exist right now and may not exist in the future is idiotic.
 
  • Like
Reactions: Zetterberg4Captain

waltdetroit

Registered User
Jul 20, 2010
2,649
526
You may need some way to quantify aspects/factors such as work ethic, competitiveness, and (of course) character. Other items: quick tick muscle mass compared to strength muscle, family data to determine final growth, bone density, maybe even DNA analysis. FWIW I can see computers doing this and maybe be as successful as scouts, but just as unsuccessful too
 

Flowah

Registered User
Nov 30, 2009
10,249
547
Is that an indictment of scouting or of the overall lack of talent outside of the first round? I don't think anyone ever claimed that drafting 18-year-olds into a pro sports league was easy.
I think it's an indictment of scouting. You can say there's not enough talent out there to fill all 7 rounds of the draft every year and that's true. But clearly many excellent and even some elite players fall through the cracks and end up being picked in rounds 3-7. What is that if not an indictment of the scouting? How do 30, now 31, teams pass up elite talent? Because they didn't see it.

Of course it's not easy. If it were easy everyone would do it. The thing is really no one does it. Which calls into question the value of NHL scouts at least for me.

I wonder what the hit rate would be of some of the people on this board. Most of us have played hockey I'm guessing. We all watch an extraordinary amount of hockey. If you landed on just a couple NHL players that get .5 PPG out of 40 picks you'd be hitting close to league average. That's not a super high bar to hit. Or maybe it is.
 
  • Like
Reactions: Fire Ken Holland

Flowah

Registered User
Nov 30, 2009
10,249
547
This is all a fun thought experiment. I only brought it up because I know for a fact this will revolutionize medical diagnoses, so... why not sports projections?
Haven't there already been people who've calculated how well draft picks tend to do in the NHL based on which leagues they came from?

What else would a computer look at exactly?
 

jkutswings

hot piss hockey
Jul 10, 2014
10,853
8,578
What else would a computer look at exactly?
You could set up a complex correlation, looking at how all the possible combinations of fifty different variables influenced or predicted NHL success for a given position on the ice. Then pare that down to the 5-10 that most strongly influence the correlation, and use that as a model for drafting.

The statistics would be the easy part, though. As some have already said, it's
gathering all that data - when minor leagues don't even uniformly track half of it to begin with - that's the hard part.

You would almost be forced to have some sort of really arcane setup, like a standard camera suite at all the rinks, at exactly the same positions, tied to tracking software that measured all sorts of complicated things, like average positioning, reaction time, and passing efficiency.

Or they could do something much harder to extrapolate with any meaning, but MUCH easier on logistics, and adopt a combine like the NFL, to provide an easily standardized (yet extremely limited) environment for gathering data right before the draft.

Interesting thought experiment, though, like Bench said before.
 

Rzombo4 prez

Registered User
May 17, 2012
5,980
2,698
I don't follow. Just because you aren't following the crowd doesn't mean you're doing something meaningful. There are plenty of totally wrong moves that don't follow the crowd, and we usually label those as aggressively stupid. So just because the Wings aren't following the crowd doesn't mean they're doing the right thing.

But on top of that, why are we starting at "out-perform the market"? The Wings are underperforming it right now. It might be asking a little much of an underachieving scouting group to become worldbeaters overnight. It seems to me that it would be a smarter move to follow the crowd, because at least you know you're going to get roughly average returns. It seems to me that your post assumes the crowd is doing it wrong, but that doesn't actually seem to be true. It doesn't take a lot of imagination to figure the crowd is doing what they're doing because they see it working in the NHL and try to draft more of that.

I mean, who's really doing it wrong? The team who drafts what is proven to work right now or the team who drafts for what could work in the future, depending on what direction the league goes? I know which one I'd pick. Drafting for a style of play that does not exist right now and may not exist in the future is idiotic.

Don't confuse necessity for sufficiency (an age old issue here at HFboards for the record). I would never claim that a purely contrarian approach will always produce success in all instances. It won't, just like following the heard won't. It appears that I may have misunderstood your original point. It sounds as if you are simply satisfied with average. Most posters on here have been bitching about our inability to find elite talent outside of the first round (a result that is statistically very, very far from average). Something tells me that deep down inside you really don't want average either. Most claim that we do not swing for the fences enough (the great size vs. skill debate of 2017). It seems, however, you want us to do something much different all together.

My question to you is how far off of average are we in terms of drafting? Why are you convinced that we aren't already following the crowd? Is it because we took Lindstrom at 38 last year? That sort of pick gets made all of the time throughout the league. Do you have any other examples for us? So say we are not following the crowd and we decide we want to start following the crowd. How exactly do we go about doing that. Should we start polling scouts from other organizations throughout the season? Something tells me that we might have a hard time getting honest answers if we manage to get responses at all. Should we double the size of our scouting department and pit one group of scouts against the other? Should we compare them to the public scouting services? If so, should we require that the scouting services show a certain level of competency? Do we have any proof that public scouting services outperform most NHL scouts?

I guess I am just having a hard time figuring out what exactly you want the Wings to do differently apart from drafting better.
 

Zetterberg4Captain

Registered User
Aug 11, 2009
13,730
2,136
Detroit
It just seems ro me that the wings have been reaching a bit on players taken in the first four rounds

Thats not to mean had they taken player X that it would be assured that the pkayer would make it, i just dont think our current scouting dept is good enough to trust over the opinions of others.

I dnt know perhaps i am way iff base here but if I were in charge i would be playing KISS the next few drafts during the first three or four rounds

Simple does not mean low skilled
 
Status
Not open for further replies.

Ad

Upcoming events

Ad

Ad

-->