WreckingCrew
Registered User
- Feb 4, 2015
- 12,604
- 38,802
More recently he specifically said he didn't want to be a YouTube scout much longer and would gladly take any and all NHL calls.
I like his work and listen to his podcasts. My guess is that folks are waiting on a few years of data to see if he has insights other don't get to using their methods. Which is likely going to be a problem, because he does seem to hedge bets based on the consensus. For instance last season he said that his data clearly showed that Brett Berard had the best transitions numbers of any players he tracked. Yet in his analysis he wondered if his style would translate to the NHL. I found that problematic when on a similarly sized player in Khusnutdinov with less impressive stats by Scouch's own measures, he said the numbers indicated future success. By the time he came to that conclusion most services had Khusnutdinov in the 25-40 range, so Scouch really didn't add much.
More recently he specifically said he didn't want to be a YouTube scout much longer and would gladly take any and all NHL calls.
I know you are a big raw data PSF fan. A metric known to you alone. But raw data can only go so far.
with one of Samoskevich, Bolduc, or maybe Koivunen.
Still, it begs the question why isn’t he well known in his community? Clearly we involved fans know him and he is well enough known to make money on YT and via patreon, so he must be on real scouting services radar and this team radars. So why doesn’t he have a job? Are the scouting groups to dumb to find him, to set in their ways to hire from outside, or is there something flawed in his opinions that make him someone they actively avoid?
Not saying any of us can know the answers to those questions, but it makes me wonder.
I like Scouch and I could imagine hiring him for his opinion but he's not that serious.
A lot of his data work is really skimpy. It's a few games here and there in different leagues with no data work for a player's teammates to know how they're doing in context. Basically low context low sample-size data isn't worth a whole lot. If he was hired full time he could help build a full database but teams usually already have that internally. Scouch is just doing a cheap-o free version of it. So it's not really a market advantage for him.
And his opinions are sometimes really not going to impress teams. He doesn't like the playoffs. He prefers regular season hockey, and he considers the Maple Leafs a great success. He likes a certain style of play more so than winning when most GMs (and fans) think matters the most. If the Ottawa Senators ever win anything with this current corps, it is going to be interesting, because Scouch has dumped all over many of their high profile picks. He was not a Brady Tkachuk fan and that take isn't aging well. He also had a strange moment about Jake Sanderson last year, where he acknowledged that this is shaping up to be a superb all situations D-man with a great transition game, and basically said "who cares" and ranked him much lower without any adequate explanation. More power to Scouch, he has built a good brand, but he isn't trying to do what teams are trying to do.
Yes I don't give all the PSF information out. If I am on to something, then it has value which I want to protect. It would be great if I could figure it out for forwards, but haven't found anything that makes sense.
As far as raw data versus "informed" viewing, I think this scene takes only slight license with the way scouts (probably more so with hockey) think/work.
There are still many who will argue that Teräväinen is not among the best in the league defensively despite everything from simple stats (GA/60) to deeper analytics (xGA) indicating that he is a defensive force. The thing about raw data is that in its truest form it lacks almost all bias--with the caveat that any metric can be created with a bias.
Koivunen would be a good pick, even though few have him ranked quite this high.
Like bleed has been saying for a while, this low in the draft I don’t really recognize anybody. It’s just a list of names.
So my logic is to pick the best name. And since Chaz Lucius is off the board, Stankoven wins by default.
For me if I have two players with similar profiles and similar transition data and I think that one is a better skater who attacks the dangerous areas of the ice more then I would be higher on the better skater.
Chaz Lucius isn’t a good hockey name. I could see it as an overhyped QB though.
He’s the name you give to the asshole pretty boy character in a late 90s high school rom com.
Good point and more to consider in an ongoing discussion that I thoroughly enjoy.
This strategy doesn't always improve on raw data. Because "better skater" is highly subjective.
Two drafts ago Pronman expressed doubt about Connor McMichael due to his speed. I responded to one of his Athletic Q&As that McMichael was in the top 3 in the CHL prospect showcase for speed. Pronman's response was that a straight-line speed test had "too much noise." That doesn't make any sense and sounds quite like the "Moneyball" quote about the ball exploding off the bat of a player who can't really hit. Pronman saw a slow skater and dismissed contradictory data as not applicable. Now I fully accept that similar to the NHL combine where a 40-yard dash number could hide an inability to make quick cuts, the straight-line skating speed doesn't mean McMichael is an exceptional skater. But Pronman didn't see that saying he lacked speed was objectively wrong--due to his bias.
Sometimes raw data is almost unacceptable, so we construct reasons to not accept it. I work in insurance underwriting. Some of the most experienced underwriters were hardest to convince as analytics became more prevalent. Let me provide the classic example that everyone from the general public to CEOs found difficult to believe. Being "in" an accident was almost as good an indicator of future losses as "causing" an accident. Now we all like to think that being the victim of an accident doesn't make you a poorer driver--and that is true. However, the majority of accidents are a combination of poor driving and external circumstances such as poor weather, poor road conditions, heavy traffic, etc. Often when someone is not-at-fault they are still driving in situations where multiple factors that lead to accidents are present. So while we all want to believe that being hit by another vehicle doesn't indicate anything about our driving, it often indicates someone driving where there are numerous external conditions that lead to more accidents. Therefore, that driver is more likely than a driver who hasn't been hit to be at-fault in a future accident.
Now before folks who read this get up in arms, no insurance commission would allow an insurer to increase rates based on being a victim in an accident. What insurers have discovered is how to track the underlying conditions (think micro-data in hockey and other sports) by using telematics to measure the number of quick stops, swerving to avoid potholes or wildlife, etc.
My guess is many teams are figuring out how to be more precise in what they measure and less subjective in dismissing the data that doesn't fit their scouting team's biases.