He gives you what he gives you and would be the safe choice.
I would rather give the opportunity to someone else.
Last edited:
Post-Game Talk: PS 2 - Niagara Falls.................... - Buffalo 4 BRUINS 1
- Thread starter Fenway
- Start date
He gives you what he gives you and would be the safe choice.
I would rather give the opportunity to someone else.
goldnblack
Registered User
- Jun 24, 2020
- 3,395
- 6,433
Was it televised? Even with my 21,000 iptv channels I couldn't find it on. Weird.
Yeah. The cheer was actually for a Buffalo prospect who's from New York, Grendal Goochmacher.
Hence,
"GOOOCH!" "GOOOCH!"
Yes. Both NESN and MSG.
Regarding Jack, the one thing that is really noticeable to me are the amount of times he's talking and something happens, then he seems to pause for a few seconds mid sentence and isn't sure what to say next. 5 10 years ago he had a much better flow.
You can tell he's speaking slower. He may of gotten speech therapy because its not as bad like last year. It sucks, but it was much better.
Things that make you go hmmmmmmmmm!
Things that make you go hmmmmmmmmm!
That stat is as reliable as +/-
I'm not a huge advance stats guy, but no Stat is very reliable if the sample size is one game.
UncleRico
Registered User
- May 8, 2017
- 8,020
- 10,109
Agreed, I would rather the growing pains of playing someone young and that the monotony of someone like heinen or brown.
MarchysNoseKnows
Big Hat No Cattle
- Feb 14, 2018
- 8,446
- 16,536
What it says is when those two were on the ice, the Bruins had the highest aggregate scoring opportunities. I don’t know why that’s controversial.
Donnie Shulzhoffer
Rocket Surgery
MarchysNoseKnows
Big Hat No Cattle
- Feb 14, 2018
- 8,446
- 16,536
You’re making my point though. Reacting to a preseason game’s advanced stats and using them to discard the notion that a stat used by virtually every analytics person, organization, or team is valid is what’s crazy.
- Mar 2, 2015
- 829
- 1,902
always curious about the source of the data as well, did the person classify high danger chances the same as another? is there a standardized universally accepted approach? If I go to a different site will I get completely different interpretation for the same game? seems like a lot of variability and bias can be introduced by whomever first manipulated/classified the data. I went to Natural Stat Trick and got different info interpretation on the same game. I have a background and one has to be very careful. Useful absolutely, can they misinterpret ..yes.
It's dependent of who your linemates are . Given last nights lineup there were plenty of questionable linemates. Given how it doesn't at all lineup with the eye test it likely comes down to luck of the draw.
I could go on but I've seen your work and I'm not up for a ride on the merry-go-round.
Particularly with a lineup like last night.
RussellmaniaKW
Registered User
- Sep 15, 2004
- 19,699
- 21,808
Things that make you go hmmmmmmmmm!
"this stat" (expected goals for percentage) basically just says "When Alex Chiasson was on the ice at 5v5 67.23% of the scoring chances were for his team".
It's not a "score" or a "grade" of a player. It's just "here's what happened when this guy was on the ice". Drawing conclusions like "X was the top player" are not what this stat says or is meant for. You always need context and other information to draw any significant conclusions about how an individual player performed. I cannot stress enough: NOBODY ON THIS BOARD IS CLAIMING THAT XGF% ALONE CAN TELL YOU IF A PLAYER PLAYED WELL.
As for whether it's "reliable", It's no less reliable than any other stat because it comes from generally reliable underlying data. It's all based on the most common basic stats that have been counted for many years now (shots, shot attempts, shot locations). As an indicator of individual player effectiveness it absolutely is more useful than +/- since it's based on shot data that's much higher volume than goal data (more data = more reliable). And small sample size of data is only one of the multiple flaws with +/- as an individual stat. Expected goals by itself doesn't tell you enough to evaluate an individual player's performance but it gets you a whole hell of a lot closer than +/-.
And against quality opponents...need to be way more physical, dont mix a lot often he use his stick instead of take the body...and the management has made it clear this team need to be more physical...
RussellmaniaKW
Registered User
- Sep 15, 2004
- 19,699
- 21,808
you're not saying anything that someone who uses these stats would disagree with, man
The issue I have with advanced stats is that there are too many variables for them to be reliablely used to compare players.
They are not totally irrelevant but at the same time they are to often presented as a be all end all counter argument , which I have no time for.
- Mar 2, 2015
- 829
- 1,902
Great write up/summary. I have yet to look at the actual algorithm they use but I assume each team has a proprietary different version that utilizes slightly different information to help normalize the data for their own team members (i.e. what is considered a high danger opportunity for player X vs player Y based on what they know about the player). Without me digging around, if 3 linemates are on the ice at the exact same time with each other for the game do they get the same xGF% as they would +/- stats?
MarchysNoseKnows
Big Hat No Cattle
- Feb 14, 2018
- 8,446
- 16,536
They would, yes. And absolutely there are different models, though the variation is pretty small.
When player tracking is released to the public next year, we’ll have a huge leap in what’s available to analyze. Right now only the teams have access to it.
Donnie Shulzhoffer
Rocket Surgery
- Mar 2, 2015
- 829
- 1,902
Yup, sure seems like it would improve upon the current model which would group all 3 players on the line (given on the ice at the exact same time during the game) the same stat ..similar to +/- stats. With the provision of available data that will come available (payer tracking), one could really fine tune it to the minutia and tailor the stat for each player ..get some really good insight ...likely something the B's office is currently doing compared to what we quote on this site from the available sources. AI will make it child's play, coaches will likely have realtime AI on the laptops on the bench that will give them insight on who to pair with who and against whom. Brave new world.
MarchysNoseKnows
Big Hat No Cattle
- Feb 14, 2018
- 8,446
- 16,536
Well there are other stats that isolate more but given how poorly a basic one like xGF is received it’s not really worth posting. Having five players get credit for a high danger chance for or a shot against because they were on the ice is much better than five players getting dinged for whatever Ryan Mast was doing last night, for example. Especially because it’s not Corsi, where any shot is rated the same - shot location and danger is taken into account. It also takes goalies out of the equation, unlike plus minus.
I’m not worried about the AI thing anytime soon. I do think player position when shooting, puck and skating velocity, etc, opens up a whole new world.
- Mar 2, 2015
- 829
- 1,902
Oh I'm not arguing +/- is better than xGF, just looking forward to better more tailored modeling that take into account other variables as you mentioned (which B's head office already does no doubt). Can you imagine a world where players have to input daily dairies of what they eat, do, watch, hang with etc to input as variables ....be valuable in identifying why a streaky player goes on streak and off a streak. Mike Babcock was ahead of his time with his phone debacle! LOL
AI is the only real approach to catch things we can't when it comes to the complexity of getting insight on statistics in real time, it'll be on the bench and practices sooner than later.
every stat has too many variables for them to be reliabley used to compare players. It's why people get paid to do these things such as scouts and analytics teams. It's about mitigating these variables in multiple ways, not finding one stat to act as the end all be all, which is why advanced stats are still a useful piece of the puzzle.
