Prospect Info: F Isak Rosen (14th Overall, 2021), Assigned to Rochester 12/10

[vcv]

This is a lot of words to blow off a data scientist asking for some counter evidence to the way his models tend to work.

Pro evaluators also think Risto was worth more than Reinhart or Buch so... yeah. And if Pro Evaluators think a player is good, why are they only playing him 5 minutes a game?

Kopitar had 0 points in 5 games his daft year, but destroyed the U20 WJC

Other people replying provided a bunch of good examples. Add in COVID last year mucking things up and it's hard to use last year to project Rosen's chance of success.

 

[jc17]

I'm not sure why he doesn't just link NHL stats to his program. I'm sure it can't be that difficult to add NHL stats and then figure out who has decent stats without good draft year numbers

 

[joshjull]

This is a lot of words to blow off a data scientist asking for some counter evidence to the way his models tend to work.

Pro evaluators also think Risto was worth more than Reinhart or Buch so... yeah. And if Pro Evaluators think a player is good, why are they only playing him 5 minutes a game?

Kopitar had 0 points in 5 games his daft year, but destroyed the U20 WJC

Read the thread on Twitter. He was given a lot of counter examples of players in similar situations . Guys like Saku and Mikko Koivu, Milan Michalek, Erickson Ek, Kucherov, etc.

It was also pointed out many players ended up in leagues this past year due to Covid not because the were ready or should have been.


I love using NHLe but its not perfect.

 

[SnuggaRUDE]

I'm not sure why he doesn't just link NHL stats to his program. I'm sure it can't be that difficult to add NHL stats and then figure out who has decent stats without good draft year numbers

I'm pretty sure that's what he does to determine NHLer and NHL star. But as you noticed early on, his tableau is awful for filtering leagues.

 

[jc17]

I'm pretty sure that's what he does to determine NHLer and NHL star. But as you noticed early on, his tableau is awful for filtering leagues.

It also looks like guys get dropped once they are full time NHLers. I downloaded the data, to try to look at draft year vs d+5 but most of the NHLers are no longer appearing

 

[Panthaz89]

When you run the numbers but forget to do basic research.

Yeah an underaged player getting basically no time in the SHL isn't anything new especially someone who needs to grow a lot. Charts and analytics are for someone who has time to actually make an impact on the game.

 

[SnuggaRUDE]

It also looks like guys get dropped once they are full time NHLers. I downloaded the data, to try to look at draft year vs d+5 but most of the NHLers are no longer appearing

They drop out after 5 years or whenever they're no longer a prospect. I think D+3 should get almost everyone.

 

[SnuggaRUDE]

Read the thread on Twitter. He was given a lot of counter examples of players in similar situations . Guys like Saku and Mikko Koivu, Milan Michalek, Erickson Ek, Kucherov, etc.

It was also pointed out many players ended up in leagues this past year due to Covid not because the were ready or should have been.


I love using NHLe but its not perfect.

Most of these counter examples did significantly better in their D SHL seasons. Their U20 SuperElit seasons were typically better (although not exclusively).

The COVID thing could be real. I think the eye test here is that an analytics guy saying "this doesn't fit the previous profiles of success", and the rebuttal of "he only got 5min/game" is answered with "the previous profiles of success earned more than 5min/game."

 

[Jim Bob]

Most of these counter examples did significantly better in their D SHL seasons. Their U20 SuperElit seasons were typically better (although not exclusively).

The COVID thing could be real. I think the eye test here is that an analytics guy saying "this doesn't fit the previous profiles of success", and the rebuttal of "he only got 5min/game" is answered with "the previous profiles of success earned more than 5min/game."

The question for the NHLe guys are what do their models say if you exclude his minutes in the SHL?

What does what he did in lower levels project?

 

[joshjull]

 

[KeyserSoze81]

Move over Tage Thompson, there is a new king of toe drags in the organization.

 

[jc17]

The question for the NHLe guys are what do their models say if you exclude his minutes in the SHL?

What does what he did in lower levels project?

his u17 season in Swedens U20 doesn't seem great in comparison to others. From what I see only Fiala, Kempe, and Loui Eriksson scored less and still had solid NHL careers. In his draft year 12 points in 12 games isn't a ton to go off of, but is still on the lower end of things.

I don't think the lack of NHLe dooms him, but I do think I would rather see a player with a better number in the first round or a player with a different style. If they believed that Rosen was far and away the BPA at that point, I guess you go with your gut, but if they had other guys near him like Coronato/Lucius, another D-man, or someone like L'Heureux whose game isn't primarily scoring, I'd probably use NHLe to lean away from Rosen

 

[Ace]

not bad for the 217th ranked prospect in the draft

 

[joshjull]

This is a lot of words to blow off a data scientist asking for some counter evidence to the way his models tend to work.

Pro evaluators also think Risto was worth more than Reinhart or Buch so... yeah. And if Pro Evaluators think a player is good, why are they only playing him 5 minutes a game?

Kopitar had 0 points in 5 games his daft year, but destroyed the U20 WJC

I had to come back to this post since I didn’t really address the bulk of it.

1) The tweet I posted was talking about ……. “We found back in the day that merely appearing in an SHL game indicated a much higher chance of success than similarly scoring players in SWE J20, especially for true U18s”.

And this……
“When prospects appear in elite-level pro in Sweden, where there aren’t mandates like in Czech league for example, there are selection biases at play that give you information on the player
That’s telling you that pro evaluators *already* believe they’re capable of playing pro”

He’s talking about how the SHL handles young players and what its meant in the past for their future. He’s also pointing out there are differences from league to league, like the Czech league having mandates. Not 100% sure what they are (ice time? Games played?) Since Bacon’s tweet was about all European leagues.

2) Kopitar had zero points in 15 games (10 playoff games). Its somewhat ironic that you’re trying to defend a data scientist’s model with a 5 game sample size from World juniors. That’s a meaningless sample size and the World Juniors itself is problematic as an example of anything.

3) To address your other post. There are players from various other leagues mentioned. Bacon made the point about production in any European league. It wasn’t specific to the SHL.



He wasn’t blowing off the data. He was giving it broader context. He wasn’t the only one taking issues with how Bacon did his work. Most were suggesting he should have used his U20 numbers.

 

[joshjull]

his u17 season in Swedens U20 doesn't seem great in comparison to others. From what I see only Fiala, Kempe, and Loui Eriksson scored less and still had solid NHL careers. In his draft year 12 points in 12 games isn't a ton to go off of, but is still on the lower end of things.

I don't think the lack of NHLe dooms him, but I do think I would rather see a player with a better number in the first round or a player with a different style. If they believed that Rosen was far and away the BPA at that point, I guess you go with your gut, but if they had other guys near him like Coronato/Lucius, another D-man, or someone like L'Heureux whose game isn't primarily scoring, I'd probably use NHLe to lean away from Rosen

He was definitely picked at the upper end or higher than where most lists had him. He needs to show us something this coming season.

 

[SnuggaRUDE]

I had to come back to this post since I didn’t really address the bulk of it.

1) The tweet I posted was talking about ……. “We found back in the day that merely appearing in an SHL game indicated a much higher chance of success than similarly scoring players in SWE J20, especially for true U18s”.

And this……
“When prospects appear in elite-level pro in Sweden, where there aren’t mandates like in Czech league for example, there are selection biases at play that give you information on the player
That’s telling you that pro evaluators *already* believe they’re capable of playing pro”

He’s talking about how the SHL handles young players and what its meant in the past for their future. He’s also pointing out there are differences from league to league, like the Czech league having mandates. Not 100% sure what they are (ice time? Games played?) Since Bacon’s tweet was about all European leagues.

2) Kopitar had zero points in 15 games (10 playoff games). Its somewhat ironic that you’re trying to defend a data scientist’s model with a 5 game sample size from World juniors. That’s a meaningless sample size and the World Juniors itself is problematic as an example of anything.

3) To address your other post. There are players from various other leagues mentioned. Bacon made the point about production in any European league. It wasn’t specific to the SHL.



He wasn’t blowing off the data. He was giving it broader context. He wasn’t the only one taking issues with how Bacon did his work. Most were suggesting he should have used his U20 numbers.

My original message was poorly attributed let me try again:

Bacon asks for information on prospects who score very little in men's leagues but still go on to do well in the NHL. Rhys says that they've found an U18 playing at the top level is predictive of success even compared to prospects scoring at similar paces in the U20 league. [For the purposes of Rosen that would be SHL and U20 Superelit). Basically saying hockey guys know it when they see it; without actually giving any examples.

From there several other tweets included names of people who might be similar, like the Kovu brothers, Ek, Nylander, Kopitar. The problem with that comparison is that Rosen vastly underperformed those players in his limited appearance at the senior mens level; and generally also under performed them in SuperElit.

That was my general umbrage with Rhys' tweet. I thought it was answering a critique of the way "hockey guys" do business by saying "this is how hockey guys do business".

Your additions are all make a lot of sense, and I agree.

I'd add that Bacon does use U20 NHLe when he does his projections. They're rolled into the the NHLe number. He had 6.9 D-1, and 5.76 D. Only about 0.5 of that came from his 1 point in the SHL.

Bacon's tableau doesn't have numbers from before the 2007 draft class so Kopitar and the Kovu brothers aren't on there; frankly in the nearly 30 years since then the NHLe is likely changed. Nylander was the top prospect per his model from 2014 with a D-1 13.21 D 20.00. So these players are suspect comparables from an analytic standpoint.

Rosen might be a great pick; I certainly hope he is.

Just for fun, if Buffalo would have picked off Bacon's list since 2011 this is what the team would look like:

Buffalo's 1st round picks since 2012 here's what they'd have picked blindly going off Star %:

2012: Hertl, Girgensons
2013: Lehkonen, Taylor Cammarata
2014: Nylander (and Bradyen Point grr)
2015: Eichel
2016: Debrincat
2017: Suzuki
2018: Dahlin
2019: Kaliyev, Mikko Kokkonen
2020: Rossi (with Lundell right at his heels)

 

[Asymmetric Solution]

Our Swedish Pasta, Rosey.

 

[jc17]

My original message was poorly attributed let me try again:

Bacon asks for information on prospects who score very little in men's leagues but still go on to do well in the NHL. Rhys says that they've found an U18 playing at the top level is predictive of success even compared to prospects scoring at similar paces in the U20 league. [For the purposes of Rosen that would be SHL and U20 Superelit). Basically saying hockey guys know it when they see it; without actually giving any examples.

From there several other tweets included names of people who might be similar, like the Kovu brothers, Ek, Nylander, Kopitar. The problem with that comparison is that Rosen vastly underperformed those players in his limited appearance at the senior mens level; and generally also under performed them in SuperElit.

That was my general umbrage with Rhys' tweet. I thought it was answering a critique of the way "hockey guys" do business by saying "this is how hockey guys do business".

Your additions are all make a lot of sense, and I agree.

I'd add that Bacon does use U20 NHLe when he does his projections. They're rolled into the the NHLe number. He had 6.9 D-1, and 5.76 D. Only about 0.5 of that came from his 1 point in the SHL.

Bacon's tableau doesn't have numbers from before the 2007 draft class so Kopitar and the Kovu brothers aren't on there; frankly in the nearly 30 years since then the NHLe is likely changed. Nylander was the top prospect per his model from 2014 with a D-1 13.21 D 20.00. So these players are suspect comparables from an analytic standpoint.

Rosen might be a great pick; I certainly hope he is.

Just for fun, if Buffalo would have picked off Bacon's list since 2011 this is what the team would look like:

Buffalo's 1st round picks since 2012 here's what they'd have picked blindly going off Star %:

2012: Hertl, Girgensons
2013: Lehkonen, Taylor Cammarata
2014: Nylander (and Bradyen Point grr)
2015: Eichel
2016: Debrincat
2017: Suzuki
2018: Dahlin
2019: Kaliyev, Mikko Kokkonen
2020: Rossi (with Lundell right at his heels)

These seem to be the active players that fit the bill. Most have a higher NHL probability and NHLe than Rosen, but the highlighted guys seem most comparable. A lot of them played in Allsvenskan, so I'm not sure if that fits.

 

[SnuggaRUDE]

View attachment 456196

These seem to be the active players that fit the bill. Most have a higher NHL probability and NHLe than Rosen, but the highlighted guys seem most comparable. A lot of them played in Allsvenskan, so I'm not sure if that fits.

This is nice work. The 4 players highlighted might be the closest active players; how many similar players were selected that didn't pan out? If there's very few of those maybe I'm being pessimistic for nothing.

 

[UnleashRasmus]

Move over Tage Thompson, there is a new king of toe drags in the organization.

Well yeah, a card board cut out is an improvement over Tage.

 

[jc17]

This is nice work. The 4 players highlighted might be the closest active players; how many similar players were selected that didn't pan out? If there's very few of those maybe I'm being pessimistic for nothing.

From what I see for SHL and Allavenskan there were 80 players with NHLe under 10, and 8 became what we'd consider top 6 players. I think I did it right, but not 100% confident it sorted correctly.

One thing that should be noted though is that the successful guys seemed to mostly be picked in the first 2 rounds. Again it's difficult because the tool doesn't list pick position but it seems like the scouts listing a guy as a first/ second rounder does give more certainty than the general 8/80 number which includes late round/undrafted players.

 

[SnuggaRUDE]

From what I see for SHL and Allavenskan there were 80 players with NHLe under 10, and 8 became what we'd consider top 6 players. I think I did it right, but not 100% confident it sorted correctly.

One thing that should be noted though is that the successful guys seemed to mostly be picked in the first 2 rounds. Again it's difficult because the tool doesn't list pick position but it seems like the scouts listing a guy as a first/ second rounder does give more certainty than the general 8/80 number which includes late round/undrafted players.

80 that became top 6 players since 2007? Any idea how many were selected?

 

[jc17]

80 that became top 6 players since 2007? Any idea how many were selected?

80 total players in his model, 8 top 6. 10% success rate. But again, it does look like the guys picked early outperformed the guys not picked or picked later

 

[TheMistyStranger]

not bad for the 217th ranked prospect in the draft

I was happy to see so many people back me on shutting that nonsense down on Bakes' twitter feed lol

 

[jc17]

I was happy to see so many people back me on shutting that nonsense down on Bakes' twitter feed lol

Who had him 217?