For roughly the same reason that a pair of dice might come up snakes eyes. - PT21
That's my point. It's just a probability, it doesn't guarantee or affect anything. It's a great tool for noticing trends in an ideal world, but that's about it.
I thought it was hilarious, in a good way. Getting outshot 51-21 and still talking poop. Love that dude. Kind of like Marchand lite, but without the cheap shots, and more willing to take a hit.
That's my point. It's just a probability, it doesn't guarantee or affect anything. It's a great tool for noticing trends in an ideal world, but that's about it. - jmatchett383
In other words, it's fantasy. It's not even good on a predictive level and even worse as an evaluation tool. The problem is by it's nature, the sport of hockey is not an ideal sports world.
Why take it so personally? Like I said, we can have a discussion on the stat and I'll explain my point of view. Something being widely used doesn't mean that it's actually good. - MJL
I didn't take it personally personally, I never do, but if a guy posts an argument using a stat, and the other guy's response is that the stat is worth used toilet paper, then the original poster might well be justified taking it personally.
Something being widely used surely does not make it worthwhile, but the onus is on the critic to explain why he thinks it is worthless when lot of people think otherwise.
And regarding the discussion you seek, those are simple stats. It looks at a large database, and sees which factors correlate with goals scored, and then sees which player was involved in them and to what degree.
Could such an aggregate get things wrong in an individual game? Of course. Thats called variation.
Could it get things wrong most of the time? No. Because if it did, the correlation mentioned above would not have survived.
And because of that, it deserves a look. And that is why Flyers got lucky yesterday. Run up stats like they did, and a large fraction of time, you get the L.
In other words, it's fantasy. It's not even good on a predictive level and even worse as an evaluation tool. - MJL
It's better than just using shot attempt differentials, but it still exists in the vacuum of an ideal world, which does not exist. This is not baseball where you have a static 1-on-1 matchup, this has 10 other players interacting on a sheet of ice with a rubber disc floating around.
Again, it's one tool in a toolbox to evaluate players, and should be treated as such.
I didn't take it personally personally, I never do, but if a guy posts an argument using a stat, and the other guy's response is that the stat is worth used toilet paper, then the original poster might well be justified taking it personally.
Something being widely used surely does not make it worthwhile, but the onus is on the critic to explain why he thinks it is worthless when lot of people think otherwise.
And regarding the discussion you seek, those are simple stats. It looks at a large database, and sees which factors correlate with goals scored, and then sees which player was involved in them and to what degree.
Could such an aggregate get things wrong in an individual game? Of course. Thats called variation.
Could it get things wrong most of the time? No. Because if it did, the correlation mentioned above would not have survived.
And because of that, it deserves a look. And that is why Flyers got lucky yesterday. Run up stats like they did, and a large fraction of time, you get the L.
Heart or no heart. - PT21
I'm not sure you'll find an argument on this. However, aside from the first and last 10 minutes, watching any sample of 2 minutes or longer would tell you which team was dictating the majority of play. Granted, you can't quantify it by doing so, but it was pretty apparent which skaters were more effective at driving play
I didn't take it personally personally, I never do, but if a guy posts an argument using a stat, and the other guy's response is that the stat is worth used toilet paper, then the original poster might well be justified taking it personally.
Something being widely used surely does not make it worthwhile, but the onus is on the critic to explain why he thinks it is worthless when lot of people think otherwise.
And regarding the discussion you seek, those are simple stats. It looks at a large database, and sees which factors correlate with goals scored, and then sees which player was involved in them and to what degree.
Could such an aggregate get things wrong in an individual game? Of course. Thats called variation.
Could it get things wrong most of the time? No. Because if it did, the correlation mentioned above would not have survived.
And because of that, it deserves a look. And that is why Flyers got lucky yesterday. Run up stats like they did, and a large fraction of time, you get the L.
Heart or no heart. - PT21
This is the issue. You're looking at the science of the stat and ignoring the data. Ignoring the noise. It's not a small sample of players. It's a large sample. The number of variables that they cannot account for, makes the stats mediocre and really unreliable as a strong tool for evaluation. Most of them are just okay and have usefulness for helping to show what happened in a game but they can't tell you why. That's why you're wrong for using a stat such as expected goals to show that this player was better than that player. The minute you start using the stats in that manner, you're going in the wrong direction. They cannot account for the variables and the false data accumulated. For the overwhelming number of players, that false data can tilt the scales in either direction.
As an example, what happens, what is different when looking at the teams overall stats, versus an individual player?
It's better than just using shot attempt differentials, but it still exists in the vacuum of an ideal world, which does not exist. This is not baseball where you have a static 1-on-1 matchup, this has 10 other players interacting on a sheet of ice with a rubber disc floating around.
Again, it's one tool in a toolbox to evaluate players, and should be treated as such. - jmatchett383
We're on the same page for the most part but I don't think any of them should be used to evaluate a player. It's not the science that is the issue, it's the data.
That's my point. It's just a probability, it doesn't guarantee or affect anything. It's a great tool for noticing trends in an ideal world, but that's about it. - jmatchett383
You are mistaken about the predictive value of probabilities.
Lets say you flip a die one hundred times. You notice that 2/3 of the times, the face shows a # 3 or more.
Does that mean that the next roll will definitely produce a value between 3-6? No.
Does that mean the past pattern you discovered has no value in predicting the future? Also nope.
The value lies in the following: think not only of a yes or no or correct/incorrect answer, but degrees of likelihood.
How likely is it that the next roll will result in a value between 3-6? The answer is twice as likely as the alternative.
This is the issue. You're looking at the science of the stat and ignoring the data. Ignoring the noise. It's not a small sample of players. It's a large sample. The number of variables that they cannot account for, makes the stats mediocre and really unreliable as a strong tool for evaluation. - MJL
The largeness of the sample cuts down on the noise. You have it in reverse.
And the value of any stat is correlation. If what you said was true, and it was indeed a mediocre stat for the purpose it was being used, it would weed itself out.
Location: Be nice from now on, NJ Joined: 03.17.2006
Sep 4 @ 4:46 PM ET
I didn't take it personally personally, I never do, but if a guy posts an argument using a stat, and the other guy's response is that the stat is worth used toilet paper, then the original poster might well be justified taking it personally.
Something being widely used surely does not make it worthwhile, but the onus is on the critic to explain why he thinks it is worthless when lot of people think otherwise.
And regarding the discussion you seek, those are simple stats. It looks at a large database, and sees which factors correlate with goals scored, and then sees which player was involved in them and to what degree.
Could such an aggregate get things wrong in an individual game? Of course. Thats called variation.
Could it get things wrong most of the time? No. Because if it did, the correlation mentioned above would not have survived.
And because of that, it deserves a look. And that is why Flyers got lucky yesterday. Run up stats like they did, and a large fraction of time, you get the L.
Heart or no heart. - PT21
From the athletic>>>
The Islanders certainly had the edge by advanced metrics in this game. That said, the shots-on-goal counter was maybe a bit misleading in terms of the degree of New York’s dominance. By even-strength expected goals, the Flyers merely trailed 2.87 to 2.57, good for a not-horrible 47.24 percent xG For rate. In reality, the Flyers largely hung with the Isles when the two sides were evenly matched in manpower, with their stellar second overtime making up a large portion of the gap.
But the Isles still had a sizable edge in all-situations xG (4.52 to 3.19), which speaks to a serious problem the Flyers desperately need to resolve if they have any intentions of winning their third consecutive game and punching their ticket to the Edmonton bubble for the conference finals …
I've been less critical of Braun but that shift that led to second goal, holy sh!t. He pinched but couldn't get the puck deep which led to the turnover, he tripped the player who ended up sliding into Hart, then Martin's shot deflected off of him and through the 5 hole. Tough freaking 20 seconds, bud.
I thought it was hilarious, in a good way. Getting outshot 51-21 and still talking poop. Love that dude. Kind of like Marchand lite, but without the cheap shots, and more willing to take a hit. - jmatchett383
Yeah Philly was getting outshot but the scoreboard read 4-4. That had to be frustrating to the Isles so Konecny's mockery could serve to further agitate and get them off their game. Perfect time to be an a$$hole.
Fair enough. More stats to amend the original stat leads to a more nuanced pic.
Flyers dominated in game other than OT2.
But some of that domination comes from pp effectiveness disparity.
And once that is taken out and OT2 is added in, domination stats improve considerably.
See - stats add value and more stats continue to add value and highlight things.
Anyone know what’s more boring than a discussion about the validity of stats in hockey? No, me neither. - Dkos
I will make it spectacularly thrilling for you. You will be enthralled.
You are a dude in an foreign country. You are standing at a bus stop. You happen to know that in that country, all families are man-woman-2 kids. Each kid is a boy or a girl with equal probability.
A beautiful woman comes by. You have no compunctions with chatting up a married woman, so you do your thing. And this woman asks you to guess the gender (same as sex here) of her 2 kids.
You find this question stupid, but you want to play along (no pun intended). But before you go make your guess, a little girl comes by and hugs the woman, and the woman says:" this is my little girl. Now you know the gender of one of my kids. Whats your guess about the gender of the other?"
What should you say?
EDIT: Forgot to add critical piece: no twins in this country.
I will make it spectacularly thrilling for you. You will be enthralled.
You are a dude in an foreign country. You are standing at a bus stop. You happen to know that in that country, all families are man-woman-2 kids. Each kid is a boy or a girl with equal probability.
A beautiful woman comes by. You have no compunctions with chatting up a married woman, so you do your thing. And this woman asks you to guess the gender (same as sex here) of her 2 kids.
You find this question stupid, but you want to play along (no pun intended). But before you go make your guess, a little girl comes by and hugs the woman, and the woman says:" this is my little girl. Now you know the gender of one of my kids. Whats your guess about the gender of the other?"
What should you say?
EDIT: Forgot to add critical piece: no twins in this country. - PT21
You would guess a boy. And you still have a 50-50 shot at being wrong to several levels of precision...assuming there is a fairly large population in this country.
