2022-2023 Blues Multi-Purpose Thread

[BlueDream]

Our web site now lists a director of hockey analytics, so we are more forthcoming this year.

After reading this I just went and looked at the full list of front office and other employees on that website, and I am laughing at how there's a section for "Blues Ambassadors." There's only one, and it's Brett Hull. They literally gave him his own title when the only thing his job entails is drinking beer and drunkenly yelling "Let's Go Blues."

 

[bleedblue1223]

Our web site now lists a director of hockey analytics, so we are more forthcoming this year.

And he's been with the team since 18/19, starting as a data analyst.

 

[Itsnotatrap]

Dom isn’t sharing daily win percentages and calling out where there are edges anymore this season, and he shared that part of the rationale is his downswing from last year. On net he squeaked out an edge, but not to the levels of previous seasons. He said having others lose in that downswing with him was part of his reason for that, and mentioned a possibility it wasn’t EV anymore.

Dom gets into silly spats with Blues fans and does some things that annoy me quite a bit. To say he doesn’t have any introspection is way too uncharitable. He has a lot more integrity about his model than your average gambling tout out there in the universe. Pulling it back because he either isn’t sure if he is missing something big and drifting towards -EV, or is sure he is missing something but hasn’t built a suitable supplement to the process, is way more honorable than what most in that industry do.

 

[Itsnotatrap]

 

[STL fan in MN]

Our web site now lists a director of hockey analytics, so we are more forthcoming this year.

Mikey’s been with the team a number of years now. Years ago, I was told Ryan Miller (not the goalie) did analytics for the Blues. So I’d assume he did some basic stuff before they brought on Perelman.

Here’s a quick blurb about him.
Meet Michael Perelman, the "numbers guy" for the St. Louis Blues -

 

[joe galiba]

Our web site now lists a director of hockey analytics, so we are more forthcoming this year.

now they will have to find and eliminate anyone who looked at this super secret information

 

[joe galiba]

how dare he make me actually want to possibly like him, at least a little

 

[Bye Bye Blueston]

Mikey’s been with the team a number of years now. Years ago, I was told Ryan Miller (not the goalie) did analytics for the Blues. So I’d assume he did some basic stuff before they brought on Perelman.

Here’s a quick blurb about him.
Meet Michael Perelman, the "numbers guy" for the St. Louis Blues -

Thanks. Always nice to see a fellow Jew in hockey!

 

[Stupendous Yappi]

OK, here's some food for thought:

Thesis: The Blues understand something that the public and probably most of the league don't understand with respect to some proprietary advanced metrics. This is tied into their team strategy and the way the offense is predicated. The result is that they score goals better than 'expected goals' would predict.

We've made comments along these lines off and on for a couple years now, often in connection with why Dom (and other) models which use publicly available metrics underestimate St Louis. (My thesis is disproven if the Blues just 'got lucky' last season.)

If you assume this is true, that the Blues are onto something, I have a couple conclusions. The rest of the league will catch on quickly and copy it. Its what happens in pro sports. But the more interesting thing I wonder is: do the Blues' proprietary metrics inform their NFL scouting in a way that also allows them to identify players who are undervalued by standard metrics? Can it go that far?

 

[Bye Bye Blueston]

OK, here's some food for thought:

Thesis: The Blues understand something that the public and probably most of the league don't understand with respect to some proprietary advanced metrics. This is tied into their team strategy and the way the offense is predicated. The result is that they score goals better than 'expected goals' would predict.

We've made comments along these lines off and on for a couple years now, often in connection with why Dom (and other) models which use publicly available metrics underestimate St Louis. (My thesis is disproven if the Blues just 'got lucky' last season.)

If you assume this is true, that the Blues are onto something, I have a couple conclusions. The rest of the league will catch on quickly and copy it. Its what happens in pro sports. But the more interesting thing I wonder is: do the Blues' proprietary metrics inform their NFL scouting in a way that also allows them to identify players who are undervalued by standard metrics? Can it go that far?

When teams have identified a secret sauce, they can use that in multiple ways. Astros (trash cans aside) have been great about that, identifying spin rate and other areas where they think guys can be much better than their ordinary numbers suggest (with some tweaks).

So if we have identified ways that we can generate offense that is not generally understood, then it's reasonable to assume that we seek to identify advantages elsewhere. I've theorized that we value specific traits in (at least high) draft picks that lead to us disproportinately drafting wingers in 1st round, as I don't think we downgrade them viz-a-vis centers as much as other teams do. And I think we believe that D are far riskier in 2nd half of first round so we stay away from them.

And circling back to your question, if we have gained a specific advantage that allowed us to outperform that advantage won't last as other teams will catch on and catch up, but the thought process that enabled us to find that advantage will help us find other ones.

 

[Zezel’s Pretzels]

do the Blues' proprietary metrics inform their NFL scouting in a way that also allows them to identify players who are undervalued by standard metrics? Can it go that far?

Well, I hope their NFL scouting is minimal at most.

 

[BlueDream]

Well, I hope their NFL scouting is minimal at most.

Patrick Mahomes could be our answer at LHD.

We'll be really ahead of other NHL teams now.

 

[Brian39]

I'm not sure that our analytics people have identified a secret sauce in the sense that they have 'new' or 'different' data. I think it is more likely that they are looking at the same data as everyone else, comparing it to eye tests of our actual games and concluding that the data isn't accurately measuring what we are doing. I don't think you need a different xG formula to know that Robert Thomas sets up a ludicrous amount of tap in goals, but does so at the expense of passing up some shots that he takes. We have a lot of highly skilled playmakers and there is a clear coaching directive to think pass-first. Tarasenko doesn't have the same shot that he used to, but last season was his best passing season by a long shot.

Maybe this directive is something from the analytics department, but I'd guess it is more likely from the eye test group saying "I don't give a damn that the analytics say we aren't generating enough chances. My eyes are telling me that we aren't scoring lucky goals."

I could be 100% incorrect because at the end of the day it is pure speculation.

 

[PocketNines]

Maybe this directive is something from the analytics department, but I'd guess it is more likely from the eye test group saying "I don't give a damn that the analytics say we aren't generating enough chances. My eyes are telling me that we aren't scoring lucky goals."

Intuitively it seems difficult to measure "tap-in chances generated." But if that were a stat it feels like the Blues would be leaders in it. Might account for what appears to be an unsustainable scoring rate to current models.

 

[bleedblue1223]

I'm not sure that our analytics people have identified a secret sauce in the sense that they have 'new' or 'different' data. I think it is more likely that they are looking at the same data as everyone else, comparing it to eye tests of our actual games and concluding that the data isn't accurately measuring what we are doing. I don't think you need a different xG formula to know that Robert Thomas sets up a ludicrous amount of tap in goals, but does so at the expense of passing up some shots that he takes. We have a lot of highly skilled playmakers and there is a clear coaching directive to think pass-first. Tarasenko doesn't have the same shot that he used to, but last season was his best passing season by a long shot.

Maybe this directive is something from the analytics department, but I'd guess it is more likely from the eye test group saying "I don't give a damn that the analytics say we aren't generating enough chances. My eyes are telling me that we aren't scoring lucky goals."

I could be 100% incorrect because at the end of the day it is pure speculation.

Yeah, I think it's a blend of both. Analytics are likely being used as supportive evidence for certain tweaks we've made that also incorporate the more old-school approach. Teams simply can't replicate a focus on successful passes to the slot because if you don't have a playmaker like Thomas, then you aren't going to be able replicate it. I imagine we are using a combo of eye test and analytics to dictate certain aspects of a system or how players get deployed.

From the past, we were told when Backes played more wing at the end of his time here, it was because the analytics showed we'd benefit more from him causing turnovers as a winger on the forecheck as opposed to playing the more traditional center role during a forecheck. I imagine a combo of eye test and analytics drove the decision to play Parayko in the shutdown role and let Petro play the more hybrid role, and that also played a role in the Krug/Faulk pairing getting even more offensive minutes.

We've seen some clubs that people would say are heavy analytics and get praise for being cutting edge, but really they were just slaves to numbers. Being analytical is simply gathering as much data as you can and reporting on it in meaningful ways and then determining how it can be applied.

Intuitively it seems difficult to measure "tap-in chances generated." But if that were a stat it feels like the Blues would be leaders in it. Might account for what appears to be an unsustainable scoring rate to current models.

I think he also had a tweet around the similar time that had it for the players, and Thomas dominated in this area.




Here's the assists off of high danger passes, but want to find the one for completed high danger passes to see where Thomas ranked.

 

[BlueDream]

I'm not sure that our analytics people have identified a secret sauce in the sense that they have 'new' or 'different' data. I think it is more likely that they are looking at the same data as everyone else, comparing it to eye tests of our actual games and concluding that the data isn't accurately measuring what we are doing. I don't think you need a different xG formula to know that Robert Thomas sets up a ludicrous amount of tap in goals, but does so at the expense of passing up some shots that he takes. We have a lot of highly skilled playmakers and there is a clear coaching directive to think pass-first. Tarasenko doesn't have the same shot that he used to, but last season was his best passing season by a long shot.

Maybe this directive is something from the analytics department, but I'd guess it is more likely from the eye test group saying "I don't give a damn that the analytics say we aren't generating enough chances. My eyes are telling me that we aren't scoring lucky goals."

I could be 100% incorrect because at the end of the day it is pure speculation.

I agree.

I think Berube is extremely good at adapting to the roster he has, and he doesn't get enough credit for that. He picks up on what we are good at and essentially builds his system around that. We've played two completely different styles when you compare the 2019 team to last year's team, and both had pretty good success.

 

[bleedblue1223]

I agree.

I think Berube is extremely good at adapting to the roster he has, and he doesn't get enough credit for that. He picks up on what we are good at and essentially builds his system around that. We've played two completely different styles when you compare the 2019 team to last year's team, and both had pretty good success.

Assistants under him have also thrived. Sort of a case, where I don't think we can really hone in one person being the most responsible for, but just an overall well-run organization and everyone is performing their role well.

 

[bleedblue1223]

Isn't really saying much here, but if would be interesting if Brown proves doubters like myself wrong and can be a decent #3 C. We've only seen Schenn line up at C during training camp so far, but he's obviously flexible enough where that doesn't mean a whole lot.

 

[PocketNines]

Berube knows who the top 7 are. Barbashev is probably #8 but he's (probably) in a lame duck season so there is opportunity for guys on lesser contracts to seize third line spots.

I believe Toropchenko intends to make a big push to force himself into that conversation. So if a few players outperform him, you're looking at a great problem. It's great to see Brown and Neighbors off to good camp starts. The internal surprise is always one of the best parts of the regular season.

 

[Reality Czech]

Isn't really saying much here, but if would be interesting if Brown proves doubters like myself wrong and can be a decent #3 C. We've only seen Schenn line up at C during training camp so far, but he's obviously flexible enough where that doesn't mean a whole lot.

Berube mentioned in an interview the other day that Schenn is the ultimate team-first guy and will play anywhere he's asked, which is obviously a great thing to have. It opens up a lot of options for the rest of the lineup.

In case anyone missed it, Strickland did a radio interview with Kyrou and the kid said all the right things. I like the fact that he's not caught up in stats and he doesn't want to label himself as a passer, shooter or any specific type of player. He recognizes that he's got a lot of room for improvement and sounds like he is willing to put in the work to become a better all around player.



As a side note, I do wonder what will happen if Kyrou doesn't click with ROR. Tarasenko and ROR have never been that effective together either, so I imagine it would mean Buchnevich moves up to play with ROR and we go with Kyrou-Thomas-Tarasenko as the #1 scoring line.

 

[execwrite1]

Faulk-Perunovich
Parayko-Leddy
Bortuzzo-Krug

 

[PocketNines]

I would love for Perunovich to prove himself up to that kind of 20+ minute job. I am skeptical based on what I've seen but the chances are nonzero. And if it happened that's pretty much the best case scenario for the Blues defense this season and therefore the team's ultimate chances, so I want it to happen. Just, doubtful.

 

[Thallis]

I guess I have a problem with the acceptance of the idea of "expected numbers", especially as the analytics have developed in hockey in general, but in particular with Dom's "model". One of the hallmarks of any analytical process (and any individual standing behind said process) is the thirst for knowledge to explain the deviations from the "expected" outcomes and the willingness to adjust the modeling to incorporate data to better predict the "expected".

The reason Dom is so vilified around these parts, aside from his unprofessional disdain for our beloved franchise, is he has looked at the Blues' outperformance to the "expected" for several years now and can only offer "they got lucky" as an explanation instead of searching for better data to create a more realistic expectation. For every squad like the Blues that overperforms relative to his flawed model, there is likely an equal an opposite data point that underperforms to the "expected" and his approach to that is likely met with "they got unlucky" as the explanation. If you're passion is data analytics and you have no curiosity as to why certain group over- or under-perform against your projections, you're probably in the wrong field.

Thus ends my rant against Dom and his lazy analytics.

This is just not true. When the Blues outperformed the model to win the cup, he adjusted the model to better capture defensive play. In 2019-2020, the Blues were considered a top team by the model and performed as expected. In 2020-2021, the Blues were considered a top team by the model and strongly underperformed. Last year they were considered middling and overperformed.

It's normal for models based on percentages to fluctuate in accuracy, especially at the individual level. You are correct that it's his job to determine when and if the projections are wrong and adjust based on it, but the model hasn't been nearly as consistently wrong as you're suggesting.

Intuitively it seems difficult to measure "tap-in chances generated." But if that were a stat it feels like the Blues would be leaders in it. Might account for what appears to be an unsustainable scoring rate to current models.

McCurdy on HockeyViz is trying to measure that with his new "Setting" stat that we measure up pretty well in. It's brand new, and McCurdy's predictive model has been historically pretty flimsy, but this might be pretty useful.

 

[PocketNines]

Combining it with my own eye test and gut feeling, I've made a good profit betting hockey the last couple years.

What categories are you finding fruitful? Season futures, individual games, over/unders & scoring props, player props, playoff series/Cup betting, etc.

 

[PJJJP]

Anyone see the reports that the cap will jump a decent amount in the 2024-2025 season?