Player Discussion Jake Sanderson (D) PART 3

Micklebot

Moderator
Apr 27, 2010
53,820
31,025
New thread

Continued from here



Sanderson is a tough prospect to judge with those kind of evaluations.

They typically underrate defensive dmen, which for the first few years of development, Sanderson very much looked like when only looking at stats. He's since added at lot of offense to his game but the skill set was always there.

I think he's going to be a stud for us, in the Slavin mold. I think he has more offensive upside, but you never know how those traits will translate.
 

Tragedy

Registered User
Jan 10, 2013
1,324
767
Regina, SK
I have no clue how that guy gets his rankings… must be from his ass
The system has a weight on more than just 1 year so Sanderson's lower offensive numbers earlier in his development impact the prediction. Basically it's heavily biased against late bloomers or defensive minded players who don't put up big numbers. It doesn't mean anything though because as an example here is Cale Makar's D0, D1 and D2 numbers which are pretty comparable

1649363253595.png
 

Ouroboros

There is no armour against Fate
Feb 3, 2008
15,028
10,308
Byron Bader still thinks Erik Brannstrom is going to be a star player. Judge accordingly.
 
  • Like
Reactions: aragorn

dumbdick

Galactic Defender
May 31, 2008
11,342
3,766
This guy is guilty of a really common academic trap. You build an opaque model that produces data that is not representative of reality, and then you fall in love with the thing and start analyzing the derived data coming out of it at par with direct observations of evidence. Results like this should be a big red flag to say "this model clearly isn't working here". But instead it's further down the rabbit hole.

At this point you're not doing analysis anymore, you're just playing with a toy you built.
 

Ice-Tray

Registered User
Jan 31, 2006
16,361
8,162
Victoria
This guy is guilty of a really common academic trap. You build an opaque model that produces data that is not representative of reality, and then you fall in love with the thing and start analyzing the derived data coming out of it at par with direct observations of evidence. Results like this should be a big red flag to say "this model clearly isn't working here". But instead it's further down the rabbit hole.

At this point you're not doing analysis anymore, you're just playing with a toy you built.
Really well put, had me laughing out loud on that one. Definitely a common trap!
 

Sun God Nika

Palestine <3.
Apr 22, 2013
19,922
8,283
Byron Bader still thinks Erik Brannstrom is going to be a star player. Judge accordingly.

That’s fine if he wants to get paid like a 3rd pairing dman which he would be here I’m fine with keeping him. In a ideal world him and Gus and some other high but not top quality piece get packaged out for a elite player
 

Micklebot

Moderator
Apr 27, 2010
53,820
31,025
This guy is guilty of a really common academic trap. You build an opaque model that produces data that is not representative of reality, and then you fall in love with the thing and start analyzing the derived data coming out of it at par with direct observations of evidence. Results like this should be a big red flag to say "this model clearly isn't working here". But instead it's further down the rabbit hole.

At this point you're not doing analysis anymore, you're just playing with a toy you built.
To be fair, he does recognize some of the blindspots of the model such as missing on defensive wizards, I don't think he buys into Sanderson's offense, which that model is biased towards. Heck, he basically says anything less than Slavin would be a terrible miss with a pick that high which to me suggests he undervalued Slavin.
 

JD1

Registered User
Sep 12, 2005
16,129
9,700
To be fair, he does recognize some of the blindspots of the model such as missing on defensive wizards, I don't think he buys into Sanderson's offense, which that model is biased towards. Heck, he basically says anything less than Slavin would be a terrible miss with a pick that high which to me suggests he undervalued Slavin.
To be fairer still, I don't think data scientist modelling geeks can sit in front of a computer and develop models with the ability to accurately project the future of teenagers.
 

Burrowsaurus

Registered User
Mar 20, 2013
42,431
16,053
This guy is guilty of a really common academic trap. You build an opaque model that produces data that is not representative of reality, and then you fall in love with the thing and start analyzing the derived data coming out of it at par with direct observations of evidence. Results like this should be a big red flag to say "this model clearly isn't working here". But instead it's further down the rabbit hole.

At this point you're not doing analysis anymore, you're just playing with a toy you built.
Mm no his model has been ok in the past. So he gotten some validation. He’s admitted there are some blind spots on it.

It’s just a guy on Twitter who is an hockey fan having some fun with stats and charting
 

The Devilish Buffoon

🇵🇸 viva 🇵🇸 free 🇵🇸
Dec 24, 2018
12,162
10,951
To be fair, he does recognize some of the blindspots of the model such as missing on defensive wizards, I don't think he buys into Sanderson's offense, which that model is biased towards. Heck, he basically says anything less than Slavin would be a terrible miss with a pick that high which to me suggests he undervalued Slavin.

Yeah, this is a good piece of context. Between 2010-2015, there's probably an average of around two guys per year who is on par with/better than Slavin per year between the 3rd & 10th slot... and most of them are F.

Marner, Draisaitl, Rantanen, Huberdeau, Werenski, Ehlers, Jones, Nurse, Lindholm, Scheifele, Couturier, Hamilton are the guys who some might consider better/on par with Slavin but personally I would take Slavin over about half of them. Only clearly superior players IMO are the first 4 forwards. When you raise the cutoff to 5, you could eliminate almost half of those guys and are left with Rantanen as the only clearly better player (imo).

If Sanderson becomes Slavin.... holy shit man, we're set. And I think most would acknowledge this, as Slavin has been between 5th-20th in Norris voting every year of his career outside of his rookie season.
 
  • Like
Reactions: NB613

ijif

Registered User
Dec 20, 2018
749
733
This guy is guilty of a really common academic trap. You build an opaque model that produces data that is not representative of reality, and then you fall in love with the thing and start analyzing the derived data coming out of it at par with direct observations of evidence. Results like this should be a big red flag to say "this model clearly isn't working here". But instead it's further down the rabbit hole.

At this point you're not doing analysis anymore, you're just playing with a toy you built.

More documentation would be better, but it's still pretty clear what is going on in his model.

Nobody is treating statistical models as on par with direct observation, and let's not act like direct observation in this context is anything scientific. NHL teams directly observed Panarin not getting drafted. They also directly observed drafting Yakupov first overall.
 

Micklebot

Moderator
Apr 27, 2010
53,820
31,025
To be fairer still, I don't think data scientist modelling geeks can sit in front of a computer and develop models with the ability to accurately project the future of teenagers.
No doubt. If they could, teams would hire them instead of scouting.
 

Ad

Upcoming events

Ad

Ad