Devils team discussion (news, notes and speculation) - part VI

[Devs3cups]

Was Bader the one who said Nemec broke his model in a good way? I lose track of all the fancy looking charts that get posted all over the internet.

Yeah it was him!

 

[Devils731]

I believe Bader was also earlier than most at moving Wright down his draft board, which was “crazy” at the time.

 

[ChicksDigTheTrap]

Was Bader the one who said Nemec broke his model in a good way? I lose track of all the fancy looking charts that get posted all over the internet.

I was just about to post that we better hope he was very right on Nemec

 

[Rhodes 81]

Should mention calling Bader a f***ing moron has less to do with the conclusions he comes to and more to do with the way he acts like a petulant child.

 

[Saugus]

Should mention calling Bader a f***ing moron has less to do with the conclusions he comes to and more to do with the way he acts like a petulant child.

For me it's because he commits the cardinal sin for a data scientist: using data without understanding the context behind it, thus using it incorrectly.

 

[Oneiro]

Dimitri talks to Novo about the Timo deal.

 

[Poppy Whoa Sonnet]

For me it's because he commits the cardinal sin for a data scientist: using data without understanding the context behind it, thus using it incorrectly.

There was a guy very similar to Bader that had a better model imo named Jeremy Davis: NextGen Hockey - Analyzing Hockey's Future Stars

He tried to monetize it but failed to get enough customers, so he gave up. Ultimately Bader's brand of snark is what wins in the internet attention economy not how good your model is, cause people are paying for entertainment not insights.

 

[MartyOwns]

If Schmid hasn't done enough to make this team full time next year, I would feel so deflated and hopeless if I were him.

If Blackwood did enough in 18-19 to be penciled in for a full time NHL slot in 19-20 (and he definitely did, I'm not saying he didn't) then Schmid should have this year.

It's pretty crazy how he's come up huge with 2 wins to end a close game, either tied or up by one goal. Last night and Ottawa game we won in OT, where he had to kill a penalty. That was his first action of the year.

And I know he didn't get the win last night.

schmid has been a very pleasant surprise. thanks shero!

 

[NjDevsRR]

Love me a good analytics argument

schmid has been a very pleasant surprise. thanks shero!

I love his positioning and rebound control. Very composed as well for such a young goalie. Coming in cold and doing a fantastic job against the Defending Champs on the road was impressive.

 

[Nubmer6]

I love his positioning and rebound control. Very composed as well for such a young goalie.

I like his calmness. He just doesn't seem to get flustered. No unnecessary movements. No panic. He just feels solid.

 

[Poppy Whoa Sonnet]

A guy like Eyssimont would be nice, I don't understand why he was traded for 50% retained Namestinkov but I'd like to add a guy like him to replace Zetterlund.

 

[Bleedred]

I don't wanna see Reimer ever play another game for the Sharks and I don't even want them to win games and he's GREAT for Bedard. So that should make where I stand on seeing him play for the Devils perfectly clear.

I'll shit if we acquire him today. My guess is they might want more for him than we'd ever consider giving them, or else Vegas would have taken him over Quick. Unless they wanted to charge more to a division rival.

It's hard to tell whether Reimer is just cooked or if he's doing his usual every other year thing and he was good last year, so this was to be expected. But he is 35 years old or about to turn 35.

 

[Saugus]

I don't wanna see Reimer ever play another game for the Sharks and I don't even want them to win games and he's GREAT for Bedard. So that should make where I stand on seeing him play for the Devils perfectly clear.

I'll shit if we acquire him today. My guess is they might want more for him than we'd ever consider giving them, or else Vegas would have taken him over Quick. Unless they wanted to charge more to a division rival.

It's hard to tell whether Reimer is just cooked or if he's doing his usual every other year thing and he was good last year, so this was to be expected. But he is 35 years old or about to turn 35.

Tell me, are there any goalies in the NHL you do like?

 

[Bleedred]

Tell me, are there any goalies in the NHL you do like?

Plenty (although the average and bad ones still get massively overrated relative to average and bad skaters) but Reimer is shot.

Yeah, the Sharks are bad but he’s been Jon Gillies level probably since December.

Keep him the f*** away from here.

Just go with who we got. I don't wanna see the decrepit #29 here ever again, but at least he's already under contract to play this year and property of ours.

Reimer isn't worth more than Hammond was last year.

I mean, he is, he's not that f***in bad, but I wouldn't pay more for him now than Hammond last year.

 

[Bleedred]

Tell me, are there any goalies in the NHL you do like?

And just to expand on this, the goalie position is in a worse position than it's been in a long time.

It's why Craig Anderson is the only 40-something year old player in the league this year, it's why Korpisalo was even re-signed and had an opportunity to be traded for Jonathan Quick and why so many of these shit bags like Martin Jones still get signed annually to one year deals.

Because the state of goaltending is about as ordinary and average as it's been in a LONG time.

All the elite goalies from not even a decade ago got old very quick and declined at young ages, outside of Luongo and Lundqvist, who were older already when they started to decline. The rest either had injuries (Schneider, Price, Holtby, Quick, Bob, Rask) that destroyed them and caused them to either LTIR or outright retire or be reduced to the AHL or wind up so bad they're completely unserviceable and should retire any game now Quick, Fleury is getting there but he's late 30's.

 

[Devs3cups]

Plenty (although the average and bad ones still get massively overrated relative to average and bad skaters) but Reimer is shot.

Yeah, the Sharks are bad but he’s been Jon Gillies level probably since December.

Keep him the f*** away from here.

Just go with who we got. I don't wanna see the decrepit #29 here ever again, but at least he's already under contract to play this year and property of ours.

Reimer isn't worth more than Hammond was last year.

I mean, he is, he's not that f***in bad, but I wouldn't pay more for him now than Hammond last year.

Plenty? Name 3. No cheating.

 

[Bleedred]

Plenty? Name 3. No cheating.

View attachment 662116

That was easy

Hellebuyck, Sorokin, Shesterkin, Vasilevskiy, Saros.

Look, I even named five!

Those are just the only REALLY GOOD ones though. Everybody else is either average/above average or BAD.

There's really no top-10 goalies list anymore, like there definitely could have been in 2015.

 

[Guttersniped]

I'm not sure I agree with the notion that jfresh just regurgitates his charts. I think it was just this week that he stated his model is probably not an accurate representation of Patrick Kane, for example. I honestly don't remember him calling B. Tkachuck a 3rd liner and I don't subscribe to him so I can't just pull up his card, but I would guess he would have said something similar then.

The purpose of developing analytical models is not to be clairvoyant or 100% accurate. It is to develop a baseline for evaluating which inputs most commonly contribute to a desired output. In hockey, there are far too many inputs (actions a player takes) that can have too random of an outcome to do this completely accurately. If what you are doing is accurate 90% of the time, that is a success in my mind. Jfresh may be wrong on some players, but it is not more than 10%. Comparatively, if an eye-test scout was correct about a player 90% of the time, they would be an extremely valuable commodity. Few scouts are this accurate. Given that, a good team will weigh their scouts against their analytics department and look for opportunities where they line up with each other.

As others have said, you have to use one tool to corroborate another. It is not the fault of the many intelligent people working in sports analytics if fans use their information to come to incorrect conclusions, no more so than it is the fault of the people that maintain the counting stats if someone assumes some bum player is good because they had 20 goals one year. Fans are just going to do that because what else are we on this message board for?

Bader is a f***ing moron.

Bader is hawking the results of his NHLe model without sharing it so he has to make the process more mysterious and the results more impressively less infallible.

So unlike other public stat guys there’s no discussion about his process that allows for more him to be open about how it works or more self-reflective about it’s flaws/limits.

Instead it all about how teams would have been better off following his amazing NHLe model.

Now mistakes are definitely made when drafting, but teams can’t just live on scoring wingers and undersized PP QBs alone. (Though their AHL teams would be lit.)

The kicker for me was when he did a top prospect list he tweaked it using prospects rankings lists.

The Hockey Prospecting Top 32 (2022)

Introducing the final Hockey Prospecting rankings of the top 32 for the 2022 draft! The rankings factor in the probabilities from the model, but also age, runway left to improve (i.e., how much tim…

hockeyprospecting.com

So it’s is his model, just with some helpful tweaks lol. (Because it’s just a NHLe model so it needs that.)

As a counter example, Dom had Nick Suzuki 7.9m contract as 9th on his 10 Worst Contracts list because that’s where the model had him.

His entry for that article was explanation why Suzuki had that result now (previous results, terrible current team), and how while there’s risk, he still has a lot of upside as a young player.

He could have tweaked the numbers to avoid the abuse he took for that (he’s a Leafs fan and that didn’t help).

A ton of people just make judgements about lists like that based on the names, and don’t read the articles, and that includes the comments in the Athletic.(Bob McKenzie is another guy people never read but discuss and his stuff is free.) So tweaking it to clean it up would make it more appealing but kind of defeats the purpose.

Luckily for him Habs fans generally are an even keeled, reasonable bunch when it comes to any perceived criticism of their players.

 

[StevenToddIves]

I believe Bader was also earlier than most at moving Wright down his draft board, which was “crazy” at the time.

No, McCagg was the first to do this. Bader had Wright at #1. Ultimately, his rankings were weak, as usual. Lane Hutson (8) ahead of David Jiricek (who might go 1st overall if the 2022 draft were held now), Lekkerimaki (11) over Slafkovsky (16!), Geekie (10) and Firkus (20) over Gauthier (22) and Kasper (25) Ty Nelson at #27 while Lian Bichsel was unranked. The dude just embarrasses himself year in and year out.

 

[Devs3cups]

That was easy

Hellebuyck, Sorokin, Shesterkin, Vasilevskiy, Saros.

Look, I even named five!

Those are just the only REALLY GOOD ones though. Everybody else is either average/above average or BAD.

There's really no top-10 goalies list anymore, like there definitely could have been in 2015.

That’s a good list! Was just messing around!

Always appreciate your goalie insights!

 

[StevenToddIves]

I was just about to post that we better hope he was very right on Nemec

Bader had Nemec at #2, which wasn't so revolutionary because most people had Nemec somewhere in the #3-#5 range. Admittedly, Bader was more bullish on Nemec than most.

 

[My3Sons]

Bader had Nemec at #2, which wasn't so revolutionary because most people had Nemec somewhere in the #3-#5 range. Admittedly, Bader was more bullish on Nemec than most.

Hope all is well with your writing and business endeavors. You picked the right year to stop draft writing for the NJ content consumers.

 

[StevenToddIves]

This is entirely backwards. If you build a model and it gets a single player 'wrong', redoing the model so that player is 'good' is the exact wrong thing to do. Maybe that player isn't actually good and your prior judgement is bad. JFresh's model isn't calling Tkachuk a third-liner, it is saying that he contributes as much to winning as a third liner. Brady Tkachuk is an exceptionally hard player to model because he doesn't play defense and he gets a million shots from in close but scores on them less than pretty much anyone in the NHL. Understand that a model is not perfect and the goal should not be perfection.

For the record, you are not open to new methods of evaluation at all. You are for methods that support your priors and nothing else. Hey, I love those methods too, we all do. Doing any sort of analytical work means shelving those notions.

Well, you missed my point a bit so I'll explain it again.

Let's just hypothetically say I come up with a new statistical analytic to measure hockey players. When I feed all the players into my algorithm, it says that Connor McDavid is actually a mediocre hockey player. The correct thing to do would be to admit that my model is extremely flawed and go back to the drawing board. The problem is that people like Bader and JFresh make a living by gullible people believing their models are incredibly accurate -- so they would respond to this same conundrum by tweeting out a dozen times that "Connor McDavid is not really as good as people think he is" or whatever. And that is why those guys are not very good at what they do. They're not giving us an alternate way to measure hockey performance -- no, they're selling their own s**t and trying to make a buck.

Ultimately, Brady Tkachuk is not an exceptionally hard player to model, he's just hard to model under a restrictive and limited algorithm which ignores several of the qualities which make an NHL forward effective. The fact is Brady Tkachuk is a PPG+ player who is also one of the top 5 forwards in the league when it comes to deflecting pucks, hunting rebounds, forechecking, screening goalies, winning battles down low and overall physical play. But the same way a lazy analytics-writer would prefer to insult a player than to second guess his algorithm, these people would also rather dismiss these extremely valuable qualities as unimportant than try to factor them into their algorithm.

It's lazy, and it's bad science.

But people are gullible, and they're willing to part with their money if they're manipulated correctly. And ultimately, that's what the Baders and JFreshes of the world are really good at, and that's why I speak out against it.

 

[Zajacs Bowl Cut]

its just such a strawman, because there is no model that would say Connor McDavis is a mediocre player.

Brady Tkachuk may be good at all those things you listed, but he is abhorrent defensively. Which the models show.

 

[Hisch13r]

For all the talk about Tatar's finishing being shot he's now nearly back at his career average for GPG and is at his career average SH%