2021-2022 S Blues Multi-Purpose Thread Part 3

  • Xenforo Cloud will be upgrading us to version 2.3.5 on March 3rd at 12 AM GMT. This version has increased stability and fixes several bugs. We expect downtime for the duration of the update. The admin team will continue to work on existing issues, templates and upgrade all necessary available addons to minimize impact of this new version. Click Here for Updates
Status
Not open for further replies.
Hopefully we can get his extension signed before season starts, bc if he puts up 95 points gonna be $$$$.

I think Kyrou could easily pot 100+ if he had a killer instinct but most of this last season he looked like he was coasting. Coasting and still put up a PPG.
 
That's my gripe with some of hockey's advanced statisticians. They want hockey to be like baseball, but the sport doesn't dictate that it can be broken down in the same way because of the lack of data points.
 
Hopefully we can get his extension signed before season starts, bc if he puts up 95 points gonna be $$$$.
I was actually wondering the other day if this is why we've heard nothing but crickets on his extension since the Thomas extension was announced. He seems like the type of guy that would relish betting on himself and if he feels like he's going to get more ice time with Perron gone I can't say that I blame him.

Kyrou had 19 PP points last season and Perron had 26. I could see Kyrou getting closer to Perron's total this season if he is promoted to PP1 and stays healthy. If he's also getting more ES minutes and keeps scoring at close to the same pace, I could see him with 60-65 ES points versus the 56 he had last season. He could easily gain $8MM ($1MM per year) by betting on himself versus signing now, but I don't see there being that much downside risk unless he completely craps the bed this season or sustains a serious injury.
 
I was actually wondering the other day if this is why we've heard nothing but crickets on his extension since the Thomas extension was announced. He seems like the type of guy that would relish betting on himself and if he feels like he's going to get more ice time with Perron gone I can't say that I blame him.

Kyrou had 19 PP points last season and Perron had 26. I could see Kyrou getting closer to Perron's total this season if he is promoted to PP1 and stays healthy. If he's also getting more ES minutes and keeps scoring at close to the same pace, I could see him with 60-65 ES points versus the 56 he had last season. He could easily gain $8MM ($1MM per year) by betting on himself versus signing now, but I don't see there being that much downside risk unless he completely craps the bed this season or sustains a serious injury.
Could also just be Army doesn't leak things and Contract talks, especially for RFAs with one year left don't always happen overnight. It sounded like the Thomas talks lasted anywhere from two weeks to a month.
 
In the final 38 games of his season, he had 17 points and was a +36. And if you want to look at naturalhattrick.com for some of the deeper advanced stats, he's every bit in line with a guy like Pelech who is probably a good barometer for elite defensive play.
Clearly one of the six worst contracts in the league. Dom is a farce
 
  • Like
Reactions: BlueMed
Statistics are a lot more developed for tracking and modeling offensive output. Defensive impact is less studied, and Dom's model is a good example where this deficiency gets pretty obvious. Teams like the Blues and Islanders the past few years have been grossly underappreciated by the model.

It takes a special kind of arrogance to publish a Top 10 worst contract list like that and display the model's holes so unapologetically.
 
That's my gripe with some of hockey's advanced statisticians. They want hockey to be like baseball, but the sport doesn't dictate that it can be broken down in the same way because of the lack of data points.

Baseball and golf are static sports. Hockey is a dynamic sport integrating both quantitative and qualitative data. Theoretically it can be done, but it would take very complex models. An experienced hockey player/fan with a good understanding of the game and whats going on in the league provides much more context and value than a simple ratio or two, even at the expensive of some subjectivity.
 
Baseball and golf are static sports. Hockey is a dynamic sport integrating both quantitative and qualitative data. Theoretically it can be done, but it would take very complex models. An experienced hockey player/fan with a good understanding of the game and whats going on in the league provides much more context and value than a simple ratio or two, even at the expensive of some subjectivity.
Plus the NHL has even more data it only shares with the teams. The public models don’t have all the data that is available to the teams themselves
 
  • Like
Reactions: Stupendous Yappi
I just saw the Twitter graphic for the article and he is showing $31.2 million in surplus value for the deal. That means his model values Parayko as a $2.6 million per year player.

Right.
Yeah that’s crazy. No team in the league would come close to getting him for that. Parayko, at his absolute WORST, would still be worth at least 4 mil per season.

Not to mention that he has always seemed to have a grudge against the Blues in general and go out of his way to denigrate our players and prospects.
It’s because Blues fans give him shit and make fun of him on Twitter and he has openly admitted he hates our fans the most. He cannot handle it.
 
That's my gripe with some of hockey's advanced statisticians. They want hockey to be like baseball, but the sport doesn't dictate that it can be broken down in the same way because of the lack of data points.
Could you elaborate on what this means? Because I don’t really understand what you’re trying to say here. Most statisticians for hockey fully understand the nature of hockey and the statistics we have now (and the vast majority we don’t have access to) reflect that. Data is being collected every game from a multitude of different angles. Data points even as far down as the paths players take and how long they hold onto the puck every shift are being recorded. The point I’m making is that I don’t understand the connection between baseball and hockey from lack of data points because there is a ton of data one can collect if one put in the effort. The problem is no one, or very few, are willing to do that unless incentivized to.

I’m not judging, of course, because I’m about as lazy as they come. Why record data when streaming exists?
 
I just saw the Twitter graphic for the article and he is showing $31.2 million in surplus value for the deal. That means his model values Parayko as a $2.6 million per year player.

Right.
In all fairness, it’s hard to say a guy is good defensively when his metrics look so poor. It’s really hard to quantify much positive value from defensive defensemen when many metrics will show negative connotations.

That doesn’t excuse the analyst from looking at the numbers and determining with the eye test that “something doesn’t quite add up here, so I will place some caveats or alternatives analysis into the discussion”.

Looking at the salary discrepancy is also telling. It begs the question, how does a guy so supposedly grossly overpaid get so many minutes against high quality competition? If he is that bad and on a competing team, how come he is used the way he is?
 
Could you elaborate on what this means? Because I don’t really understand what you’re trying to say here. Most statisticians for hockey fully understand the nature of hockey and the statistics we have now (and the vast majority we don’t have access to) reflect that. Data is being collected every game from a multitude of different angles. Data points even as far down as the paths players take and how long they hold onto the puck every shift are being recorded. The point I’m making is that I don’t understand the connection between baseball and hockey from lack of data points because there is a ton of data one can collect if one put in the effort. The problem is no one, or very few, are willing to do that unless incentivized to.

I’m not judging, of course, because I’m about as lazy as they come. Why record data when streaming exists?
I know you didn’t ask me, but baseball data is clean. Game is broken down into non-continuous plays that yield discreet data points. Hockey data is messier. The continuous nature of play and multiple inputs on any action make it more difficult to accurately construct model to value players. There is so much data and you need to really understand what you are seeing to know what is valuable. This isn’t to say it’s impossible, just that someone like dom can have model that in some ways seems helpful but is in many ways incomplete and misleading.

Teams like the blues that don’t fit his model maybe don’t require a little adjustment. Maybe he is missing the thread that ties it all together. Like how at some level you can do physics without calculus but when you look at small enough objects it all breaks down. And that break is on team level. On player level, especially for defense, it seems like his model has nearly no value.
 
It's also easier with the eye test to look at a sniper and conclude his scoring is down because he's playing with bottom 6 grinders. It's easy to quantify it, compare it to previous seasons, etc. With defenseman, especially defensive defensemen, it's harder to eye test and quantify how much their play changes when they are paired with either an inferior partner or a partner that doesn't complement their style.
 
And one example for how advanced stats are used internally and externally. In Backes' last season or 2 here, he transitioned from a traditional center to a forechecking center, where Steen or Stastny would handle the traditional center role when we were on defense. This was done because the stats the Blues tracked showed that we'd benefit more from Backes being on the forecheck and creating turnovers.

I don't think I've seen a public stat for creating turnovers off the forecheck. Analytics 100% have a place in the game, and I don't agree with those that knock it, it's more so the people online that don't fully understand them.
 
Last edited:
  • Like
Reactions: Bye Bye Blueston
I completely agree with the opinion that Dom's model (and advanced stats in general) don't measure defensive defenseman even remotely well.

I vehemently disagree that Parayko is currently or has been performing like a #2/3 tweener D man. Comparisons to Seabrook's prime to support that notion are wild to me. That's like calling a guy a #2/3 center because he isn't performing like Malkin. Seabrook in his prime would have been a #1 D man on a lot of NHL teams and was the undisputed #2 on a borderline dynasty that had one of this generations clear cut Hall of Famers as their #1. Seabrook played the 11th most minutes of all NHL D from 2009/10-2015/16. 30th in minutes per game and 27th in even strength minutes per game over that stretch. He was a clear cut #1/2 tweener on a team that had a top 5 D man in the league as the #1. Seabrook got $5.8M a year as an RFA in 2011 when the cap was $59.4M. That is the percentage equivalent of $7.95M against the cap as it stood when Parayko signed his extension. Seabrook's first UFA contract was for $6.875M in 2015, which is equivalent to $7.85M when Parayko signed his extension. It takes him to 38 years old and had a full NMC for the first 6 years.

Parayko's contract absolutely doesn't set an expectation that he should be as good as Brent Seabrook in his prime.

He hasn't remotely been deployed as a #2/3 D man since Petro left. He led the Blues in TOI per game and was 26th league-wide. He led the team in even strength TOI per game (by 1:27 a night) and was 3rd league-wide. Roman Josi is the only player in the league who played more even strength minutes than Parayko last year and they were ludicrously easier minutes than Parayko got. Pretty much all of his minutes were hard minutes. He didn't pad the total TOI numbers with PP usage. 37 D men played 23+ minutes a night last year and Parayko's 17 seconds of PP TOI per game is the lowest of all of them. The next lowest is 31 seconds a night and there are only 3 of these 37 D men who got less than a minute a night on the PP. Parayko was consistently matched up against the opponent's top line, his O zone start rate was below 40% and he had a rotating door of middling partners all season. His most frequent partner last year was a guy that is universally considered overpaid at $3.275M. His 2nd most frequent partner was a guy who just signed at $1.9M rather than going to arbitration.

There is a very legitimate argument that Parayko's assignment/workload was the most difficult in the NHL last season. Despite that, he netted a respectable 35 points and was +16 (which was 7th on a team that had a very good +69 goal differential). The possession metrics are ugly, which is true of basically every guy around the league who gets deployed in a shut down role. I haven't seen any analysis that is critical of him that remotely acknowledges how difficult his usage was. It is always 'he gets a slight bump for his usage' without going into any detail about how no D man in the league was asked to do as much as he did.

Parayko is not an all situations clear cut #1D who can drag around a #4/5 level D man against top competition and still look like a stud for 23+ minutes an night. But I'm not sure that even 10 of those guys exist in today's NHL. There certainly aren't 10 guys who did that last year. The fact that he isn't that guy does not mean that he is suddenly a #2/3 D man.
 
Last edited:
Could you elaborate on what this means? Because I don’t really understand what you’re trying to say here. Most statisticians for hockey fully understand the nature of hockey and the statistics we have now (and the vast majority we don’t have access to) reflect that. Data is being collected every game from a multitude of different angles. Data points even as far down as the paths players take and how long they hold onto the puck every shift are being recorded. The point I’m making is that I don’t understand the connection between baseball and hockey from lack of data points because there is a ton of data one can collect if one put in the effort. The problem is no one, or very few, are willing to do that unless incentivized to.

I’m not judging, of course, because I’m about as lazy as they come. Why record data when streaming exists?
I'm saying there are a number of people who put too much stock into these models which are still obviously flawed. In baseball there are hundreds of data points each game that carry extreme significance. Because of this design, there's so much more meaningful data that can be taken from a single pitch. Much of baseball comes down to a one-on-one interaction between a pitcher v. a batter.

In sports like hockey there are so many extremely important things that happen without possession of the puck and, as far as I know, nobody has even come close to quantifying in a meaningful way in terms of determining player value. There are some who will treat these advanced statistics as the gospel, rather than admitting that these models have significant limitations when it comes to play without possession.

Dom naming Parayko as one of the 10 worst contracts in hockey is a perfect example of this. I doubt there are very many people who have watched Parayko actually play the game, can understand the value he brings to the ice, and still make that claim. If the Blues made Parayko available, teams would be lining up to take on that contract.
 
I always find it interesting where there's always a point when the season finishes and people sort of wipe their minds clean. Did everyone forget how well the Blues did of slowing down Mack and Makar. The O'Reilly line and Leddy/Parayko did a ton of work. Besides Mack's hat trick game, that we ended up winning, he was pretty much a non-factor. Only 1 team held him pointless in multiple games, and only 1 team held his goal scoring to 1 game.
 
Could you elaborate on what this means? Because I don’t really understand what you’re trying to say here. Most statisticians for hockey fully understand the nature of hockey and the statistics we have now (and the vast majority we don’t have access to) reflect that. Data is being collected every game from a multitude of different angles. Data points even as far down as the paths players take and how long they hold onto the puck every shift are being recorded. The point I’m making is that I don’t understand the connection between baseball and hockey from lack of data points because there is a ton of data one can collect if one put in the effort. The problem is no one, or very few, are willing to do that unless incentivized to.

I’m not judging, of course, because I’m about as lazy as they come. Why record data when streaming exists?

I'm saying there are a number of people who put too much stock into these models which are still obviously flawed. In baseball there are hundreds of data points each game that carry extreme significance. Because of this design, there's so much more meaningful data that can be taken from a single pitch. Much of baseball comes down to a one-on-one interaction between a pitcher v. a batter.

In sports like hockey there are so many extremely important things that happen without possession of the puck and, as far as I know, nobody has even come close to quantifying in a meaningful way in terms of determining player value. There are some who will treat these advanced statistics as the gospel, rather than admitting that these models have significant limitations when it comes to play without possession.

Dom naming Parayko as one of the 10 worst contracts in hockey is a perfect example of this. I doubt there are very many people who have watched Parayko actually play the game, can understand the value he brings to the ice, and still make that claim. If the Blues made Parayko available, teams would be lining up to take on that contract.


My problem with statistical models in hockey is that they are often times black boxes. A whole host of statistical information is inserted into the black box, and it spits out a single easily expressed number. But the whole process of how that number was achieved is shrouded, either through being too complex or being proprietary.

I prefer simple advanced stats. Corsi is just the number of shot attempts either for or against. It doesn't express as much as a full statistical model, but I know what it means. I know its strengths and weaknesses. It doesn't lie, it just doesn't attempt to tell the full story. Since I inherently understand what it means, I can combine it with other factors (eye test, deployment stats, QoC stats, etc) to better understand the full picture myself.

With a full model, I don't know which of those factors the model weighted and how heavily, because its all in a black box. I can get a sense that the model heavily favors X while ignoring Y by looking at how it rates certain players. But that's a bit of work to just understand the model. I don't think understanding the model is the point. I think the point of a model is for people to substitute its opinion for their own. That doesn't work in a place where discussion can get as granular as it can here (ideally). We end up discussing the model more than the player, and that is inherently less interesting to me.
 
Last edited:
My problem with statistical models in hockey is that they are often times black boxes. A whole host of statistical information is inserted into the black box, and it spits out a single easily expressed number. But the whole process of how that number was achieved is shrouded, either through being too complex or being proprietary.

I prefer simple advanced stats. Corsi is just the number of shot attempts either for or against. It doesn't express as much as a full statistical model, but I know what it means. I know its strengths and weaknesses. It doesn't lie, it just doesn't attempt to tell the full story. Since I inherently understand what it means, I can combine it with other factors (eye test, deployment stats, QoC stats, etc) to better understand the full picture myself.

With a full model, I don't know which of those factors the model weighted and how heavily, because its all in a black box. I can get a sense that the model heavily favors X while ignoring Y by looking at how it rates certain players. But that's a bit of work to just understand the model. I don't think understanding the model is the point. I think the point of a model is or people to substitute its opinion for their own. That doesn't work in a place where discussion can get as granular as it can here (ideally for me). We end up discussing the model more than the player, and that is inherently less interesting to me.
Yes. It’s great to see detailed advanced stats too, like zone entry and exit numbers. These tell you much more than basic stats but do so in a way that is clear about what they are.
 
My problem with statistical models in hockey is that they are often times black boxes. A whole host of statistical information is inserted into the black box, and it spits out a single easily expressed number. But the whole process of how that number was achieved is shrouded, either through being too complex or being proprietary.

I prefer simple advanced stats. Corsi is just the number of shot attempts either for or against. It doesn't express as much as a full statistical model, but I know what it means. I know its strengths and weaknesses. It doesn't lie, it just doesn't attempt to tell the full story. Since I inherently understand what it means, I can combine it with other factors (eye test, deployment stats, QoC stats, etc) to better understand the full picture myself.

With a full model, I don't know which of those factors the model weighted and how heavily, because its all in a black box. I can get a sense that the model heavily favors X while ignoring Y by looking at how it rates certain players. But that's a bit of work to just understand the model. I don't think understanding the model is the point. I think the point of a model is for people to substitute its opinion for their own. That doesn't work in a place where discussion can get as granular as it can here (ideally). We end up discussing the model more than the player, and that is inherently less interesting to me.
Amen.

I would assume that the more and more advanced stat models are meant to simply things with a simply player X is better because his whatever score is higher but to me it just makes it more cloudy.

Like, a stat like GAR (Goals Above Replcement) is super enticing to look at as it’s supposed to show how many goals above or below a replacement level player someone should contribute based on the combination of a ton of other fancy stats…but when you start combining fancy stats, each with their own flaws and outliers, I would think that the error rate would go up and correlation to the actual value a player contributes would go down.

As others have said, I think hockey is simply a harder sport to do advanced stats for. With baseball, there’s a set play, mostly a pitcher vs a batter, and then the play is over. All sorts of things can be measured. Then there’s another play. I would assume football would be similar because of how the plays are broken down but I don’t pay attention to that sport much so I’m just guessing there. Hockey is so much more fluid with one play flowing into another that breaking it all apart and separating the signal from the noise would have to be a lot harder.
 
I agree with most of what's being said here. Advanced stats should be used to support what our eyes see but a lot of people want these numbers to tell the whole story. Hockey just doesn't work in that way. It's lazy.
 
  • Like
Reactions: BlueMed
Status
Not open for further replies.

Ad

Upcoming events

Ad

Please disable your adblocker on HFBoards.com

It looks like your adblocker is attempting to interfere with the intended operation of this site. Support us by allowing our site to function as we intended. Please disable your adblocker and add us to your allowlist.

Frequently Asked Questions
I'm not using a blocker. Why am I seeing this message?
You're likely seeing this message because an app or extension on your computer is blocking ads. The app or extension may be a "privacy" or "malware" blocker, or a VPN.

I disabled my blocker. Why am I still seeing this message?
It's common to have two or more adblocking extensions running at the same time. See the question above.