The Advanced Stats Thread Episode VIII:

[Irishguy42]

Side note: did the collapse bbcode break?

 

[East Coast Bias]

Don't recall if this was posted here already

Tableau Public

Paper: In Search of Similarity: Finding Comparable NHL Players

1.5 cheaper. Step was so underrated here.

 

[Ola]

Arsenal appoint former Candy Crush guru Mikhail Zhilkin as their new 'data scientist'

 

[Ola]

How big is analytics in the NBA, can’t Dolan build up a cross border division within MSG?

 

[East Coast Bias]

Vesey's #1 defensive comparable is none other than.........JT Miller

 

[Mac n Gs]

@silverfish, Sean Tierney backing up what you’ve been saying about :



SA16 too about giving more TOI to these players.

 

[Irishguy42]

The thread on the main baord is killing me softly.

"The most advanced stat is watching the game."

K

People do realize that watching the game is how advanced stats are made, right? Corey Sznajder is underrated.

 

[silverfish]

@silverfish, Sean Tierney backing up what you’ve been saying about :



SA16 too about giving more TOI to these players.

It sucks that it has to be breaking news to not spend money on fourth-liners. Like, the NHL is so prehistoric that teams waste millions of dollars a year on a 'type' of player that has no real utility. So annoying.

 

[SA16]

Somehow it's shocking to some that the best strategy is to simply play your best players the most. They're not so fragile that they are going to wear down by playing 22 mins/game instead of 18. There are plenty of top forwards that get huge ice time and succeed (Barkov/Trocheck/Kopitar/Couturier/McDavid/ROR to name several).

Florida was able to seriously limit the damage their bottom two lines did against themselves by playing Barkov/Trocheck 44 mins a game.

And that is why, sadly, I have argued in the past that I think it's POSSIBLE a team is better off playing McLeod for 5 minutes than [insert random 4th liner here] 10 minutes a game. Because that 5 minute difference will be extra ice time for the 1st/2nd line wings. (And of course they'd be better if the play [random 4th liner] 5 minutes instead of McLeod but we all know they wont do that.

@silverfish Any idea if it's possible to run a study on how teams perform when they run 11F/7D as opposed to 12 and 6? In my experience when teams do that generally the 7D barely plays and they just bump up the ice time for the good forwards.

 

[YoSoyLalo]

One of the changes I’m hoping for from Quinn is willingness to play his best players.

I want Kreider and Zibanejad playing 20+ a night, Shattenkrik playing 25+, no reason for them to play less.

 

[silverfish]

Somehow it's shocking to some that the best strategy is to simply play your best players the most. They're not so fragile that they are going to wear down by playing 22 mins/game instead of 18. There are plenty of top forwards that get huge ice time and succeed (Barkov/Trocheck/Kopitar/Couturier/McDavid/ROR to name several).

Florida was able to seriously limit the damage their bottom two lines did against themselves by playing Barkov/Trocheck 44 mins a game.

And that is why, sadly, I have argued in the past that I think it's POSSIBLE a team is better off playing McLeod for 5 minutes than [insert random 4th liner here] 10 minutes a game. Because that 5 minute difference will be extra ice time for the 1st/2nd line wings. (And of course they'd be better if the play [random 4th liner] 5 minutes instead of McLeod but we all know they wont do that.

@silverfish Any idea if it's possible to run a study on how teams perform when they run 11F/7D as opposed to 12 and 6? In my experience when teams do that generally the 7D barely plays and they just bump up the ice time for the good forwards.

Hm. I don't know if there's an easy way to identify that via the PBP scrape that I get because I don't get position, only who is on the ice. The json pages probably have that description and could probably work it from there... I think everyone on this forum in particular has a really bad taste in their mouth from game 7 against TB in the ECF where AV yolo'd himself by playing 7D, scratching Sheppard, and playing Hunwick 6 shifts.

 

[Mac n Gs]

Yeah. There’s that weird argument that all shifts matter equally, which would logically lead to the conclusion to play your better players more frequently.

Good depth is obviously important, but the focus should be not having players that negatively impact your team. These guys can be replaced every summer, which is why it’s better to allocate money towards players with higher impacts. Seems obvious, but people still think +/- and GAA are two solid metrics to quantify defense.

@silverfish I think there won’t be enough of a meaningful sample size to look into the 11F 7D argument. If you could do it, that would be cool.

 

[SA16]

Hm. I don't know if there's an easy way to identify that via the PBP scrape that I get because I don't get position, only who is on the ice. The json pages probably have that description and could probably work it from there... I think everyone on this forum in particular has a really bad taste in their mouth from game 7 against TB in the ECF where AV yolo'd himself by playing 7D, scratching Sheppard, and playing Hunwick 6 shifts.

Yea well that was just AV being AV and doing something they never do. The Devils did it a bunch this year and each time Hall/Palmieri's ice time spiked.

 

[silverfish]

Yea well that was just AV being AV and doing something they never do. The Devils did it a bunch this year and each time Hall/Palmieri's ice time spiked.

There might be a lot of noise that drowns this out, though. For example, some person who likes to pretend he is smarter than he is on Twitter posted something about how the Leafs record in games when JvR plays 15+ minutes sucks. But like, JvR is probably eating minutes in those games because the Leafs are already trailing and that's a simple hypothesis that checks out easily. JvR's ice time spikes when the Leafs needed a goal.

So I'm a bit uncertain of the utility here, but this is absolutely the type of shit that needs to start being explored. It's like, okay, we know what makes a good player or bad player even though people still don't want to admit that for whatever reason, but now how can we make this tactical? How can we use what we know to improve the team's record strategically? Like, baseball had the data, and now they're doing all sorts of crazy lineup optimizations and defensive shifts and all that. Obviously hockey is not baseball, but you get what I mean.

Like the random question my friend had asked me the other day. Does it make sense to play 4v6 instead of 5v6 so you can take free shots at the empty net? The answer was no, but that could provide utility and is not hard to learn one way or the other.

@silverfish I think there won’t be enough of a meaningful sample size to look into the 11F 7D argument. If you could do it, that would be cool.

With my current set up, it'd be a pretty massive undertaking because I'd essentially have to rebuild the scraper that I use to not use the nhlscrapr package but do something far more custom. That said, if for whatever reason nhlscrapr becomes redundant or stops working, I am f***ed.

 

[SA16]

There might be a lot of noise that drowns this out, though. For example, some person who likes to pretend he is smarter than he is on Twitter posted something about how the Leafs record in games when JvR plays 15+ minutes sucks. But like, JvR is probably eating minutes in those games because the Leafs are already trailing and that's a simple hypothesis that checks out easily. JvR's ice time spikes when the Leafs needed a goal.

So I'm a bit uncertain of the utility here, but this is absolutely the type of **** that needs to start being explored. It's like, okay, we know what makes a good player or bad player even though people still don't want to admit that for whatever reason, but now how can we make this tactical? How can we use what we know to improve the team's record strategically? Like, baseball had the data, and now they're doing all sorts of crazy lineup optimizations and defensive shifts and all that. Obviously hockey is not baseball, but you get what I mean.

Like the random question my friend had asked me the other day. Does it make sense to play 4v6 instead of 5v6 so you can take free shots at the empty net? The answer was no, but that could provide utility and is not hard to learn one way or the other.



With my current set up, it'd be a pretty massive undertaking because I'd essentially have to rebuild the scraper that I use to not use the nhlscrapr package but do something far more custom. That said, if for whatever reason nhlscrapr becomes redundant or stops working, I am ****ed.

Well yea the JVR thing is like the NFL stats you get where teams are like 30-1 or whatever when the RB gets at least 30 carries in a game...because obviously when you're winning you just run out the clock and when you're losing you pass.

I just figured looking at 11/7 could be a good proxy for playing your best forwards the most since you're forced into it. For example why is hockey played with 12F/6D? The only reason is really that that's how it's always been done. Who is to say it wouldn't be better to do something like 11/6 and then have a guy sitting on the bench who is purely a shootout specialist or something for example and he never plays unless it goes there? Or just rolling 3 lines and that's it and the other 3 "forwards" get purely used for penalty killing or if one of your 9 regulars gets hurt/ejected or the game turns into a blow out.

 

[Blue Blooded]

This goes well with what I've previously found which is that 1LDs are a lot more scarce than 1RDs, contrary to the common opinion.

While LDs are more abundant than RDs, the elite defencemen are almost exclusively right handed nowadays. My hot take is that a super talented kid that shoots left is put at center, but someone who shoots right is more commonly shifted to defence where handedness is more important.

 

[Mac n Gs]

This goes well with what I've previously found which is that 1LDs are a lot more scarce than 1RDs, contrary to the common opinion.

While LDs are more abundant than RDs, the elite defencemen are almost exclusively right handed nowadays. My hot take is that a super talented kid that shoots left is put at center, but someone who shoots right is more commonly shifted to defence where handedness is more important.

Do you have a list for this or is it just personal observation?

It’s awesome that you’re posting on these boards again.

 

[Blue Blooded]

Do you have a list for this or is it just personal observation?

It’s awesome that you’re posting on these boards again.

On my phone right now so I cannot datapost, but after developing metrics to predict future performance I noticed that RDs were vastly overrepresented at the top (it was something ridiculous like 16/20 for 16/17 IIRC).

 

[Mac n Gs]

Is Jason Zucker one of the most underrated wingers in the lead? 30-30 guy last season and has great underlying stats the past two years too.

Minny doesn’t have a lot of cap space to sign him long term. It could also be an opportunity to take back salary and add assets.

 

[silverfish]

Agree with Mac n Gs, hope you're sticking around @Blue Blooded

xG update:
Kind of shit so far, but at least progress is being made. Still have some data cleansing to do with the NHL PBP data where they record shots with a distance > 100 ft. Hoping cleaning that up will get that to a usable spot, but right now it's basically complete crap

Fitting 15-17 xG to the 17-18 season:

This needs to be way, way tighter.

Is Jason Zucker one of the most underrated wingers in the lead? 30-30 guy last season and has great underlying stats the past two years too.

Minny doesn’t have a lot of cap space to sign him long term. It could also be an opportunity to take back salary and add assets.

Eric Staal effect? What a colossal failure from AV that was.

Not as much as I expected... for what it's worth, I'm set up to do this for any player if anyone has requests. Might just loop all NYR in a minute and post

NYR:

 

[silverfish]

Is Chytil a volume shooter or is my sample too small? Well, I know my sample is too small, but in his 7 games at the end of the year, he led the team in 5v5 iCF60 by over 2. 17.06 vs Zibanejad next at 14.96 [Natural Stat Trick]. Over 2 shots on goal per game at the AHL level would suggest a willingness to shoot the puck.

This thought sparked by a guy in my mentions on Twitter who usually is a troll but had a good thought about how we replace Nash's shooting volume next year. Another 'recreate him in the aggregate' situation like with losing Stepan. Chytil is a darkhorse here, if he gets the minutes, to help lighten the load on the guys who should be carrying it, Zibanejad, Kreider, and Buchnevich.

 

[Mac n Gs]

Is Chytil a volume shooter or is my sample too small? Well, I know my sample is too small, but in his 7 games at the end of the year, he led the team in 5v5 iCF60 by over 2. 17.06 vs Zibanejad next at 14.96 [Natural Stat Trick]. Over 2 shots on goal per game at the AHL level would suggest a willingness to shoot the puck.

This thought sparked by a guy in my mentions on Twitter who usually is a troll but had a good thought about how we replace Nash's shooting volume next year. Another 'recreate him in the aggregate' situation like with losing Stepan. Chytil is a darkhorse here, if he gets the minutes, to help lighten the load on the guys who should be carrying it, Zibanejad, Kreider, and Buchnevich.

Chytil, Lias, and Pionk were a few of the most frequent shooters in Hartford, so therr may be a chance here. I think those resulte are possibly skewed a bit, but I donthink the potential to mitigate his loss is there.
It’s kind of silly how much of this season is reliant on Chytil and Andersson taking another step in their development because of how poor our depth is. Rebuild!

 

[silverfish]

Chytil, Lias, and Pionk were a few of the most frequent shooters in Hartford, so therr may be a chance here. I think those resulte are possibly skewed a bit, but I donthink the potential to mitigate his loss is there.
It’s kind of silly how much of this season is reliant on Chytil and Andersson taking another step in their development because of how poor our depth is. Rebuild!

Hard to keep this from going OT but yeah, totally. The off-season is still young, but as it stands right now and the moves this FO has made, I really have no idea what 'the vision' actually is.

Again, it's only July, so we'll see, but

 

[Mac n Gs]

Hard to keep this from going OT but yeah, totally. The off-season is still young, but as it stands right now and the moves this FO has made, I really have no idea what 'the vision' actually is.

Again, it's only July, so we'll see, but

Best guess? Instill a new style of hockey that’s heavily influenced by the current style used by the USDP based on the hires of Quinn and Brown. Oliver also has ties to US hockey through Joe Sacco and Quinn.

That north-south style is heavily driven by structure and pressuring the puck in the defensive zone and having an active defense drive transition up ice in 5-man units. As to the offensive zone, that’s where they get iffy as it’s a mix of cycling and freewheeling on the fly. It works when there’s talent in place to showcase skill, but I don’t think we have the firepower to do this for 4 lines.

With that in place, integrate prospects in waves so they can get softer usage and acclimate to the NHL game.

I honestly think improving the style of play is the main goal for this season. How many times have we seen AV’s system trashed by Brooks and Carp recently?

 

[silverfish]

Best guess? Instill a new style of hockey that’s heavily influenced by the current style used by the USDP based on the hires of Quinn and Brown. Oliver also has ties to US hockey through Joe Sacco and Quinn.

That north-south style is heavily driven by structure and pressuring the puck in the defensive zone and having an active defense drive transition up ice in 5-man units. As to the offensive zone, that’s where they get iffy as it’s a mix of cycling and freewheeling on the fly. It works when there’s talent in place to showcase skill, but I don’t think we have the firepower to do this for 4 lines.

With that in place, integrate prospects in waves so they can get softer usage and acclimate to the NHL game.

I honestly think improving the style of play is the main goal for this season. How many times have we seen AV’s system trashed by Brooks and Carp recently?

One thing I'm really interested in seeing is the YoY changes of anyone who played a full year under AV last year and plays a full year under Quinn next year. Outside of normal aging and progression, I'm thinking we see some interesting stuff.

Not sure if anyone is paying attention, but Neil Greenberg is expertly trolling hockey Twitter right now. At least for his sake, I hope he's trolling because if he's serious this is just sad now.