2019/2020 Advanced Stats / Analytics Tracker

[Fezzy126]

Do you have last years? I was looking for it but couldn't find it. Made my own in a spreadsheet but not as fun to look at.

Team Charts Over Time: Expected Goal Share - 2018-19

 

[jc17]

A couple interesting things this morning:

1) From an Athletic Article:

He’s been a top tier defender to start, but for whatever reason isn’t being used as one.
That reason: the Sabres are winning a lot and leading a lot of games lately and Dahlin isn’t being trusted in those minutes. When tied or losing, Dahlin’s minutes shoot up. Considering he had some of the team’s strongest numbers while defending a lead last year (and the less talented Rasmus did not), so it’s hard to support that usage.

Makes it sound like you could argue Ralph is saving 26 for moments he feels like the team needs him, rather than benching him.

2) In line with the shot tracking, I think it's interesting how different sites have varying methods of tracking and measuring xG. Natural Stattrick has the sabres at 53% xGF, while evolving hockey is at 56%.

 

[Fezzy126]

A couple interesting things this morning:

1) From an Athletic Article:

Makes it sound like you could argue Ralph is saving 26 for moments he feels like the team needs him, rather than benching him.

2) In line with the shot tracking, I think it's interesting how different sites have varying methods of tracking and measuring xG. Natural Stattrick has the sabres at 53% xGF, while evolving hockey is at 56%.

I forget the specifics, but they each handle empty net situations differently, which could cause the numbers to vary slightly as well.

 

[joshjull]

Good article. WOWY is indeed a poor stat. Or at least poorly used by many.

 

[Fezzy126]

Good article. WOWY is indeed a poor stat. Or at least poorly used by many.

Common theme: People using stats without understanding them (Filipovic gets it). Most smart people can see when various stats are misused, and at the very least have an intelligent discussion about the data.

But not RAPM, no one really understands the algorithm behind the graphs, which is why I find them exceedingly dangerous. More and more, I find people on twitter using them as proof for 'player x' being good, or 'player y' being bad. The charts lack some serious context, omit extremely important data points, are wildly inconsistent, but yet create the illusion of a catch-all, NHL video game-style rating system for fans. bleh...

 

[jc17]

Common theme: People using stats without understanding them (Filipovic gets it). Most smart people can see when various stats are misused, and at the very least have an intelligent discussion about the data.

But not RAPM, no one really understands the algorithm behind the graphs, which is why I find them exceedingly dangerous. More and more, I find people on twitter using them as proof for 'player x' being good, or 'player y' being bad. The charts lack some serious context, omit extremely important data points, are wildly inconsistent, but yet create the illusion of a catch-all, NHL video game-style rating system for fans. bleh...

I think this is my issue with a lot of current stuff.

Also I think too many people want attention on their Twitter, article, website, etc and are more concerned with introducing a new stat or a new visualization, regardless of its true meaning.

I do like evolving hockey, watched and read about their model for rapm and war and its clear they are very bright, but I agree that most people have trouble interpreting

 

[valet]

A couple interesting things this morning:

1) From an Athletic Article:

Makes it sound like you could argue Ralph is saving 26 for moments he feels like the team needs him, rather than benching him.

2) In line with the shot tracking, I think it's interesting how different sites have varying methods of tracking and measuring xG. Natural Stattrick has the sabres at 53% xGF, while evolving hockey is at 56%.

the discrepancy probably has something to do with the NHL shot distance tracking fiasco that has the analytics community in an uproar. Last I heard they were fixing it, but some folks could still be using the old data

 

[Fezzy126]

the discrepancy probably has something to do with the NHL shot distance tracking fiasco that has the analytics community in an uproar. Last I heard they were fixing it, but some folks could still be using the old data

They both use the same data source - the NHL. The discrepancies have to do with the way they all calculate expected goals, which all involve different formulas and statistical techniques to calibrate the results. Based on those calculations, yeah the location data could have affected some of the models more than others (I don't know the details well enough to say for sure), but they are definitely using the same data inputs.

For what it's worth I think the player tracking systems that the NHL is discussing will completely overhaul all of these models. All the current expected goals models have the same caveat, and that is "based on publicly available data". Having shot and event data is better than not, but there is an entire world of data not yet available that is going to make analytics infinitely more useful. (At least to some of us nerds)

 

[Buffalo Norsemen]

Nice Chad!

 

[jc17]

So we've fallen below 50% in the big 4: CF%, xGF%, SC%, HDSC%.

Buuut, we're still in the top 10 in most of those when the game's within a goal. So time will tell if its score effects coming into play, or a coincidence.

 

[Aladyyn]

So we've fallen below 50% in the big 4: CF%, xGF%, SC%, HDSC%.

Buuut, we're still in the top 10 in most of those when the game's within a goal. So time will tell if its score effects coming into play, or a coincidence.

We're above 50% in all of them with score+venue adjustment. We're doing fine.

(tbh this early in the season Corsi is the only stat I would take seriously)

 

[jc17]

We're above 50% in all of them with score+venue adjustment. We're doing fine.

(tbh this early in the season Corsi is the only stat I would take seriously)

Right I just wonder if this early, with still relatively small sample sizes at each score effect, if they're playing worse because of score effects or if its coincidental.

 

[jc17]

 

[yahhockey]

Hopefully there are fancy stats to back this up but it seems they jump out to early leads then defend, defend, defend and their analytics, especially third periods, pay the price. If they kept their foot on the gas it could be a different picture. For example in the Stars game they had the 2-0 lead and although they scored two goals in the third they only had two or three shots and the second goal was six minutes into the period.

 

[K8fool]

So we've fallen below 50% in the big 4: CF%, xGF%, SC%, HDSC%.

Buuut, we're still in the top 10 in most of those when the game's within a goal. So time will tell if its score effects coming into play, or a coincidence.

Does the score affect the players in an nhl game? .. Hmmmm. Does the trailing team possess the puck more.? .. Hmmm ... Should i watch hockey or stare at stats on the phone. Am i a parrot that talks ...

Does the seeming feeling of security ,however logically dismissed as folly
, affect the actions of an indivudual or a group...etc ?

What is life?

These are the questions that keep me from abandoning the idea of seeking out a lab near Afghanistan or Vietnam w a lifetime of diabetic supplies..

Ah the unified field theory for hockey without the key variables of context..

 

[Push Dr Tracksuit]

Does the score affect the players in an nhl game? .. Hmmmm. Does the trailing team possess the puck more.? .. Hmmm ... Should i watch hockey or stare at stats on the phone. Am i a parrot that talks ...

Does the seeming feeling of security ,however logically dismissed as folly
, affect the actions of an indivudual or a group...etc ?

What is life?

These are the questions that keep me from abandoning the idea of seeking out a lab near Afghanistan or Vietnam w a lifetime of diabetic supplies..

Ah the unified field theory for hockey without the key variables of context..

theres not liking the way analytics is end all be all around here and then theres this

 

[Buffaloed]

For what it's worth I think the player tracking systems that the NHL is discussing will completely overhaul all of these models. All the current expected goals models have the same caveat, and that is "based on publicly available data". Having shot and event data is better than not, but there is an entire world of data not yet available that is going to make analytics infinitely more useful. (At least to some of us nerds)

I worry that the league isn't taking a serious look at potential negative consequences of player tracking systems. Whenever they make changes, it's often an oops we meant well affair. Then they spend years tweaking. They can never put the worms back in the can.

Will this data be used in contract negotiations? Will it change the way players play the game? Could it be used to develop an algorithm that predicts how many useful years a player has left? Will it make outcomes predictable? What are the implications for gambling? How secure will it be? Won't this have a dehumanizing effect on the game? We'll be talking about needing players with a specific data set before you know it. With enough data models could be developed to create virtual players that are indistinguishable from human players. The league needs to look to outsiders to find out how this data can be exploited in ways they'd never think of.

 

[jc17]

Does the score affect the players in an nhl game? .. Hmmmm. Does the trailing team possess the puck more.? .. Hmmm ... Should i watch hockey or stare at stats on the phone. Am i a parrot that talks ...

Does the seeming feeling of security ,however logically dismissed as folly
, affect the actions of an indivudual or a group...etc ?

What is life?

These are the questions that keep me from abandoning the idea of seeking out a lab near Afghanistan or Vietnam w a lifetime of diabetic supplies..

Ah the unified field theory for hockey without the key variables of context..

If you want to argue something ive said come back when you're sober

 

[vcv]

I think this matches what I’ve seen. They are slowing down a bit and struggling after strong starts. How much of that is score effects remains to be seen. Let’s see them not score early for once...

 

[K8fool]

If you want to argue something ive said come back when you're sober

That's the problem .. I am brutally sober and the stats do not take into account change in strategy of the coach based on the fatigue of players on the second night of a back to back etc. In fact all of the stats in the L.A. game are invalid or at the very least misleading after the the first .. They lack the context of the moment.. Words like caved in as reflected by stats have no meaning to me in a 3 0 victory when the team showed grit and hustle to reduce real scoring opportunities to aa much as possible.. I done agree w the shell but they pulled it off and won 3 0.. They got killed in two periods by the stats which Do Not Reflect the excellent play by the team on the road.. Your comnent on sobriety is a passive aggressive attempt to dismiss through some misplaced self righteous nous about my choice to abandon a society that isnt honest with itself and is still trying the quantify the unquantifiable. Thats cool.. Please simply defend a stat that doesnt fail and mislead..as the situatiom evolves.. .

 

[K8fool]

View attachment 266343

I think this matches what I’ve seen. They are slowing down a bit and struggling after strong starts. How much of that is score effects remains to be seen. Let’s see them not score early for once...

Lets not

 

[K8fool]

I worry that the league isn't taking a serious look at potential negative consequences of player tracking systems. Whenever they make changes, it's often an oops we meant well affair. Then they spend years tweaking. They can never put the worms back in the can.

Will this data be used in contract negotiations? Will it change the way players play the game? Could it be used to develop an algorithm that predicts how many useful years a player has left? Will it make outcomes predictable? What are the implications for gambling? How secure will it be? Won't this have a dehumanizing effect on the game? We'll be talking about needing players with a specific data set before you know it. With enough data models could be developed to create virtual players that are indistinguishable from human players. The league needs to look to outsiders to find out how this data can be exploited in ways they'd never think of.

Even the physical affects of increased pinging..

 

[jc17]

the stats do not take into account change in strategy of the coach based on the fatigue of players on the second night of a back to back etc.

First of all, the stats I'm talking about dont refer to the Kings game alone, I was talking about the season. Second, sitting in your own zone giving up scoring chances isnt a strategy, its a symptom. Yeah I also expect the team to struggle 2nd night of back to back on the road, but again I wasnt talking about just this game.

Words like caved in as reflected by stats have no meaning to me in a 3 0 victory when the team showed grit and hustle to reduce real scoring opportunities to aa much as possible..

Words like grit and hustle have no meaning when the team actually didnt reduce real scoring opportunities.

This picture has very little to do with stats. Giving up shots very close to the net is not a good thing. Its not debatable. But go ahead saying they reduced real scoring chances. The only thing may have been odd man rushes.

Your comnent on sobriety is a passive aggressive attempt to dismiss through some misplaced self righteous nous about my choice to abandon a society that isnt honest with itself and is still trying the quantify the unquantifiable. Thats cool..
Please simply defend a stat that doesnt fail and mislead..as the situatiom evolves.. .

Does the score affect the players in an nhl game? .. Hmmmm. Does the trailing team possess the puck more.? .. Hmmm ... Should i watch hockey or stare at stats on the phone. Am i a parrot that talks ...

Who's passive aggressive?

 

[Dubi Doo]

I wouldn't say them getting caved in the last two games was a result of a 2 goal lead. They just got heavily out played. Hopefully it's due to fatigue, but it is cause for some concern. Let's see how they respond tonight with a bit of rest, though they're still on a 3-in-4 and 4-in-6 night games played, so the fatigue may still be there.

 

[K8fool]

First of all, the stats I'm talking about dont refer to the Kings game alone, I was talking about the season. Second, sitting in your own zone giving up scoring chances isnt a strategy, its a symptom. Yeah I also expect the team to struggle 2nd night of back to back on the road, but again I wasnt talking about just this game.



Words like grit and hustle have no meaning when the team actually didnt reduce real scoring opportunities.
View attachment 266361
This picture has very little to do with stats. Giving up shots very close to the net is not a good thing. Its not debatable. But go ahead saying they reduced real scoring chances. The only thing may have been odd man rushes.

Fine point by point.. You need a larger sample size to say a player or players are playing poorlt thru these highly subjective and vague stats..

Your shot charts appear to back up your point but you would have to look at every single shot to see if it was released cleanly .. If it had a screen ... Whether the shot placement was in the pads because player was under duress etc.. You know the effort and determination and physicality you dismiss..

Also the main point that should verify they caved for ten minutes after they went up 3 0 is two kovalchuck opportunities that should have been goals that dont even show in those stats other than vaguely in corsi..

Find better stats that dont demonize players who are told by coach to play a role or imply one player got caved in etc..

And yes i threw self righteous and passive aggressive dismissal right back as it seemed unfortunately appropriate considering how prejudiced your comment was about a joke you took seriously instead of the context i was exasperated about..

We may have to move on with this until you have stats that reflect a cleaner version than what i see.. All stats have some value but currently they should not make your decisions just make you look harder at tape to diagnose what is occuring in flawed stats.. Much like lame panacea doctor tests that do not root out cause..

I still dont like sobotka and i do believe the team is generally playing well despite the last two cocky alamo mode when up 2 0 that resulted in one bad loss to nasty birds .

Ok .. Have at it




Who's passive aggressive?

I put my reply inside expanded apparently to spare the board although it was accidental