Boston Bruins Advanced Stat Thread

[burstnbloom]

Just gonna roll this little grenade through the door and then get back to work...

There is no such thing as the eye test. Anecdotal observation is well and fine, but there's no coherent criteria to administer the eye test. Everyone's free to amass a series of gut feelings based on watching the game. No one would ever argue that you can't have an opinion formed this way--just that it's incomplete. Every field of study relies successfully on data driven analysis.

This is obviously true. The "eye test" as people on this board discuss it is called "confirmation bias" in any other forum. People have a "feeling" about something and they pick anecdotes to validate that building bias. There is no process of validation or independent control. The "eye test" narrative is a way for someone to look at irrefutable, quantified evidence of a truth (player x had these results) and say "no i dont believe that." It's bananas.

 

[ON3M4N]

There is a lot of misuse/misunderstanding in this post.

1) Not caring about what he did in Ottawa is incorrect when you're using statistical samples. The more variables a trend carries through, the more valid the data. Seeing consistent performance on multiple teams with multiple partners isolates performance. It just doesn;t suit your preconceived nothing which I think is that relative stats mean everything?

2) using ozone start% lacks context. I'll give you an example. Everyone talks about Forbort being buried in the dzone. He has a 40% ozone start. Wow. Super tough minutes! Ya he starts in the dzone 9.25 shifts per 60 minutes this year. Gryz starts in the d zone 8 shifts per 60. Given their TOI, thats 1 dzone start per 4 games more for Forbort. Forborts ozone start% is 40% and Gryz's is 57%. The reason is Gryz is capable in the ozone so his exta shifts are there where forbort's are not.

3) The relative stats are frankly useless. Especially when its a D core with widly divergent outcomes, THat said, all of your underlying numbers showing better than average and your GF numbers behing less than (within the margin of error really) it points to bad luck/flukiness

4) The Forbort is a pk horse myth needs to die, especially with this bad correlation does not equal causation argument when he was hurt. The penalty kill when Forbort was out gave up similar chances to the month before, but their PDO was in the .800's in November. The regression came and Forbort came back and everyone gave him credit. It's infuriating. Forbort has the worst numbers of an of the regular PK players on this team in the following categories:

CA/60 (shot attempts against)
FA/60 (unblocked shot attempts against)
xGA/60 (expected goals against)
SCA/60 (scoring chances against)
HDCA/60 (high danger chances against)
PDO

so what does that mean? It means he's giving up more shots than any other d, more unblocked shots, more expected goals, more scoring chances of all kind, more high danger chances against and he's been the luckiest by PDO. His on ice sv% is 92.06. Charlie McAvoy, for example, is .875. It has long been proven that their is no way for defenseman to influence on ice sv% and that its shifty and nebulous, but there isnt a correlation between good play and poor play by D. so - tl:dr - he's fine on the PK but everyone else is better, and given how pitiful he is at 5v5, its probably not worth rostering him. The bruins disagree, clearly, but I think thats based on a role ideal than anything quantifiable.

This post right here is exactly why people hate the analytics crowd. Could I sit here and keep this going? Ya pretty easily, but no matter what analytic I bring up you'll tell me why it's useless and what you're looking at is better. I like analytics myself and look at them frequently.... and even I don't want to deal with this. You notice how no one else wanted to answer your question? It's because they know your post above is what's going to be coming and be presented as an aboslute.

Your post basically comes off like your smarter than the Bruins front office, coaching staff and everyone in their analytics department. That's what drives non-analytics fans nuts.

With that said, I'm out on this thread.

 

[sarge88]

They are different but they are similar methodology, is my understanding. The use of expected results as a metric is prevalent in every major sport in the world and it’s dwidely accepted as the best methodology we have for measuring performance.

I don’t doubt you. It would be nice, as Dom says, if the teams released what they use…exactly, so that we can be better able to compare them.

 

[burstnbloom]

I don’t doubt you. It would be nice, as Dom says, if the teams released what they use…exactly, so that we can be better able to compare them.

The NHL is not exactly forthright with anything proprietary. We still rely on third party websites for contract informations.

 

[burstnbloom]

This post right here is exactly why people hate the analytics crowd. Could I sit here and keep this going? Ya pretty easily, but no matter what analytic I bring up you'll tell me why it's useless and what you're looking at is better. I like analytics myself and look at them frequently.... and even I don't want to deal with this. You notice how no one else wanted to answer your question? It's because they know your post above is what's going to be coming and be presented as an aboslute.

Your post basically comes off like your smarter than the Bruins front office, coaching staff and everyone in their analytics department. That's what drives non-analytics fans nuts.

With that said, I'm out on this thread.

this is what I mean when I say "bad faith." You have no interest in having a real discussion, you just wanted to say I was wrong and this is your projection lol.

Your characterization of my post is just way off. You're using bad stats- thats not because I dont like them, its because they incorrectly make the point you were trying to make. I told you why and you take your ball and go home.

I don't think I'm smarter than anyone but data is about objectivity. I don't think the bruins have made the right decisions when it comes to Forbort and Reilly. I'm allowed to have that opinion. Mine is based on the objective numbers that we have.

In general, I would say its a bit too sensitive to come to the exiled "advanced stats thread" to use bad stats and get upset when you're called on their lack of context. My lord.

 

[neverwatchthegames]

I don’t doubt you. It would be nice, as Dom says, if the teams released what they use…exactly, so that we can be better able to compare them.

My understanding (guaranteed to be partially wrong and definitely incomplete) is that, for the most part, teams use basically the same techniques as the public models. The main difference is that they have more robust models and larger data sets that come from subscriptions to analytics companies or collected and built by their inhouse teams. A quick google search will show you a huge selection of companies--some quite large--that collect hockey data, build models, and sell them to teams, media companies, investment groups, gambling outfits, and more. This stuff isn't meant for the fans per se, but some of us definitely enjoy this aspect of the game, and find it's insights enhance the viewing experience. That's why the public models exist at all. While the teams likely don't rely on the public models for decision making, they certainly pay attention to them.

 

[Dizzay]

Question for the advanced stats folks in here:

Is Bergy the odds on favorite to repeat as the Selke winner for this season? Who are the other two most likely nominees and how close are they to Bergy? Thanks!

 

[TCB]

Question for the advanced stats folks in here:

Is Bergy the odds on favorite to repeat as the Selke winner for this season? Who are the other two most likely nominees and how close are they to Bergy? Thanks!

Not an advanced stats guy, but I don't see how he doesn't win it. He's still the best two-way forward in the world and he's on a team who's on pace to set an all time wins and points record. He's still the best in the dot (2nd behind Toews in percentage) and don't look now, but he's slowly creeping up among the plus/minus leaders. Simply put he's the best !!!!

Nico Hischeir imo would be his closest comp and he's a great young Center and he'll win a few before his career is finished but not yet.

 

[JoeIsAStud]

Just gonna roll this little grenade through the door and then get back to work...

There is no such thing as the eye test. Anecdotal observation is well and fine, but there's no coherent criteria to administer the eye test. Everyone's free to amass a series of gut feelings based on watching the game. No one would ever argue that you can't have an opinion formed this way--just that it's incomplete. Every field of study relies successfully on data driven analysis.

So the Bruins should fire all their scouts?

 

[neverwatchthegames]

So the Bruins should fire all their scouts?

I'm struggling to find that implication in my statement.

 

[PB37]

So who counts all the happenings that advanced stats track? It seems like a really monotonous, tedious job to count all the hundreds of stick checks that happen during a game, for example. Or to try to declare a winner of a close board battle. Do they go back and check their work or just roll with what they saw live and pump it out into the stat-mosphere for us to read and take as gospel?

I haven't read this entire thread, so apologies if this has been asked and answered already.

 

[neverwatchthegames]

So who counts all the happenings that advanced stats track? It seems like a really monotonous, tedious job to count all the hundreds of stick checks that happen during a game, for example. Or to try to declare a winner of a close board battle. Do they go back and check their work or just roll with what they saw live and pump it out into the stat-mosphere for us to read and take as gospel?

I haven't read this entire thread, so apologies if this has been asked and answered already.

It's super labor intensive. The puck and jersey tracking chips would help with some of that, but there's still lots of context to add to every interaction both on and off the puck. Some of the public models scrape data off of the NHL's officially released data set. There are others that have a bunch of volunteers recording things like passes and zone exits/entrances. The NHL has a bunch of official trackers who watch each game and add to the data set. There are definitely issues with human-error, both in the NHL's data and independently collected data. People try to account for bias and error in building the models, but it's not perfect by any stretch. I don't think there will every be a completely pristine error-free model. That being said, the larger the sample sizes the less and less each individual error matters.

 

[Gee Wally]

So the Bruins should fire all their scouts?

of course not. All inputs available should go into the decision making process. Add even a third, the players personality makeup. Then a fourth, who is the agent.

I'm struggling to find that implication in my statement.

I think you covered it perfectly with :

”Everyone's free to amass a series of gut feelings based on watching the game. No one would ever argue that you can't have an opinion formed this way--just that it's incomplete”

 

[Dennis Bonvie]

Wasn't much defenceman in that era now was there?

And comparing Lysell to Wayne or Mario isn't the way to go. Comparing Lysell to his teammates is. Then you can figure out who is at least attempting to play the team's system.

Actually was talking about the "speed" generated by Gretzky and Mario when coming back defensively.

Is there a speed if there is no coming back?

 

[Dennis Bonvie]

Eye test guys definitely put ketchup on their hot dogs

I do not.

 

[neverwatchthegames]

of course not. All inputs available should go into the decision making process. Add even a third, the players personality makeup. Then a fourth, who is the agent.

This. The scouting department and analytics department should be in joined suites. The scouts and the nerds are so much more effective when they share insights. Any well run organization should know this.

 

[Dennis Bonvie]

Just gonna roll this little grenade through the door and then get back to work...

There is no such thing as the eye test. Anecdotal observation is well and fine, but there's no coherent criteria to administer the eye test. Everyone's free to amass a series of gut feelings based on watching the game. No one would ever argue that you can't have an opinion formed this way--just that it's incomplete. Every field of study relies successfully on data driven analysis.

I feel bad now.

Having to watch hockey for 60 years and all that time I really didn't know what was really going on.

Then again, for most of that time players, coaches and management didn't either.

 

[neverwatchthegames]

I feel bad now.

Having to watch hockey for 60 years and all that time I really didn't know what was really going on.

Then again, for most of that time players, coaches and management didn't either.

I'm really not trying to make that point. We watch the game and we think about the game. We digest it and enjoy it. I don't think there's anything wrong with that. Our brains aren't wired to see everything that appears before our eyes. Like light beyond the visual spectrum, there's so much to hockey that we can't comprehend from our sofas. I'd be disappointed if folks weren't out there developing new and improved understandings of the sport. I

 

[Dennis Bonvie]

I'm really not trying to make that point. We watch the game and we think about the game. We digest it and enjoy it. I don't think there's anything wrong with that. Our brains aren't wired to see everything that appears before our eyes. Like light beyond the visual spectrum, there's so much to hockey that we can't comprehend from our sofas. I'd be disappointed if folks weren't out there developing new and improved understandings of the sport. I

Or from analytics.

 

[MarchysNoseKnows]

Or from analytics.

To be fair, you use an analytic to inform your opinion all the time.

 

[neverwatchthegames]

Or from analytics.

Yes. That's the point.

 

[False Start]

so what does that mean? It means he's giving up more shots than any other d, more unblocked shots, more expected goals, more scoring chances of all kind, more high danger chances against and he's been the luckiest by PDO. His on ice sv% is 92.06. Charlie McAvoy, for example, is .875. It has long been proven that their is no way for defenseman to influence on ice sv% and that its shifty and nebulous, but there isnt a correlation between good play and poor play by D. so - tl:dr - he's fine on the PK but everyone else is better, and given how pitiful he is at 5v5, its probably not worth rostering him. The bruins disagree, clearly, but I think thats based on a role ideal than anything quantifiable.

I have this argument with my buddy who's pro Forbort and I basically call him out the same way you do.

My point is, okay even if we respect that he's great on the PK. Then what?

Do the Bruins, as a team, just accept that he makes every single partner on the defense miserable 5 on 5?

Do we all agree that when he's on the ice it's OKAY to be coincidentally out shot and out chanced on a consistent basis 5 on 5?

I'm not sure why people use his PK prowess as a way to justify his lack of anything everywhere else.

Great guy, I'm sure great in the locker room. He also makes good defensive and surprisingly good offensive plays sometimes. The stats just say on average he's a net negative when it comes to pushing the puck forward away from the goal. He's also negatively impacting his own D-partner in the same manner. That's 1/3rd of your defensive pairs that play 18-20 minutes a night in this system.

 

[MarchysNoseKnows]

I have this argument with my buddy who's pro Forbort and I basically call him out the same way you do.

My point is, okay even if we respect that he's great on the PK. Then what?

Do the Bruins, as a team, just accept that he makes every single partner on the defense miserable 5 on 5?

Do we all agree that when he's on the ice it's OKAY to be coincidentally out shot and out chanced on a consistent basis 5 on 5?

I'm not sure why people use his PK prowess as a way to justify his lack of anything everywhere else.

Great guy, I'm sure great in the locker room. He also makes good defensive and surprisingly good offensive plays sometimes. The stats just say on average he's a net negative when it comes to pushing the puck forward away from the goal. He's also negatively impacting his own D-partner in the same manner.

One of the great mysteries to me is how Forbort passes the "eye test". It's not just analytics - if you watch him, you see how much worse he is than anyone else, especially Grizz. But, as Fluto so eloquently put it yesterday, he is "stout".

 

[False Start]

One of the great mysteries to me is how Forbort passes the "eye test". It's not just analytics - if you watch him, you see how much worse he is than anyone else, especially Grizz. But, as Fluto so eloquently put it yesterday, he is "stout".

To be honest he's not as bad as anti-forbort people make it seem.

He's an NHL defenseman with a niche. On this team in particular we just project higher expectations (You'd think for 3 mil this would be the case).

Orlov since he's not played with Forbort, dominated.

Clifton's stats away from Forbort, fantastic.

No players stats with Forbort are better with than without haha. Every other defenseman on this team is carrying him when he's on the ice at even strength.

Just my opinion but he's the weakest link on a team with not very many weaknesses. And to me that's ok to point out.

 

[Dennis Bonvie]

To be fair, you use an analytic to inform your opinion all the time.

True. But it usually confirms what I see.

As the poster said, there's so much to hockey that we can't comprehend.

When Bobby Orr played, there was no access to plus/minus numbers for fans. But everyone knew he was heads and shoulders above the rest. Retro plus/minus confirmed that.