Did you crunch the numbers?

This doesn't have anything to do with +/- btw
- Giroux_Is_God

What doesn't have anything to do with +/-?

What you call false data is actually the off-puck stuff that Corsi/Fenwick captures better than any other stat.
- Tomahawk

I agree that Corsi/Fenwick is the best thing we have right now. Doesn't eliminate the issue of false data that makes it unreliable for rating individual players.

Absolutely there is. If you're talking about a team number, then the issue of false data isn't there. With individual players, there is a lot of it. As much as 30-40% of it is false data. Same issue as +/-.
- MJL

Corsi = Shot Attempts = Shots + Missed Shots + Blocks

Fenwick = Shots + Missed Shots

That's it. If a player is on the ice for any event, his Corsi reflects it. Again, it's a statistic that shows trends, which is why any serious stats person will caution against small sample sizes and uses other things like zone starts and entries and quality of both teammates & competition to get a fuller picture.

But a guy being on the ice for a shot for or against is pretty cut and dried.

Corsi = Shot Attempts = Shots + Missed Shots + Blocks

Fenwick = Shots + Missed Shots

That's it. If a player is on the ice for any event, his Corsi reflects it. Again, it's a statistic that shows trends, which is why any serious stats person will caution against small sample sizes and uses other things like zone starts and entries and quality of both teammates & competition to get a fuller picture.

But a guy being on the ice for a shot for or against is pretty cut and dried.
- Jsaquella

No it's not!

I never realized there was this much disdain for emrick. I don't love him but I don't get annoyed listening to him. I only hear him for national broadcasts though.

Who do people hate more emrick or buck?
- PhillySportsGuy

the key here is you only hear him for national broadcasts

Absolutely there is. If you're talking about a team number, then the issue of false data isn't there. With individual players, there is a lot of it. As much as 30-40% of it is false data. Same issue as +/-.

5 skaters are on the ice. They register 5 shot attempts for, and other team registers 3 shot attempts against them. All 5 players skate to the bench after the shift with a +2 in Corsi attempts for that shift. Anybody who watches the game knows that it's quite possible that a player can be on the ice for that shift, and have zero to do with any of it, or just part of it. One player can make the mistake that led to all 3 shot attempts against. But the other players are also punished for it. Same thing with +/-. A player can get a plus or a minus when it is not deserved.
- MJL

Based on your logic, no stat is EVER relevant in the real world because it is not performed in a vacuum.

Corsi = Shot Attempts = Shots + Missed Shots + Blocks

Fenwick = Shots + Missed Shots

That's it. If a player is on the ice for any event, his Corsi reflects it. Again, it's a statistic that shows trends, which is why any serious stats person will caution against small sample sizes and uses other things like zone starts and entries and competition to get a fuller picture.

But a guy being on the ice for a shot for or against is pretty cut and dried.
- Jsaquella

It's not cut and dry. Not even close. A player can be on the ice for a shift, where he played his position flawlessly, made zero mistakes, and he can come out of that shift on the negative end of Corsi. When you have a high level of false data involved in forming a players individual Corsi rating. It's impossible for it to be cut and dry. Now if we want to look at it on a team basis, then that issue is pretty much removed. Only issue then is whatever human element is involved. Which I don't think much can be done about.

You can have that opinion if you'd like. But if you know my stance on Corsi and some of the problems I have with it, you'd know that is not the case.
- MJL

Whatever

Based on your logic, no stat is EVER relevant in the real world because it is not performed in a vacuum.
- PhillySportsGuy

Obviously you don't get my logic.

Yeah, you're probably right
- Giroux_Is_God

good call

I agree that Corsi/Fenwick is the best thing we have right now. Doesn't eliminate the issue of false data that makes it unreliable for rating individual players.
- MJL

Takes all five guys doing the right thing to get the puck up the ice and in a position to be shot on net... on the other side of the puck, we're all familiar with the idea that defense requires all five guys to work in tandem.

Crediting everybody on the ice for each shot event makes perfect sense in 99% of cases. Only one I can think of that might qualify as false data is if the defense plays everything perfectly but the goalie lets in a long-range softy... otherwise, even a solo end-to-end rush requires the other 4 guys to draw off some defenders, hence doing their jobs off-puck.

Obviously you don't get my logic.
- MJL

My logic has a bigger Richard than your logic

http://www.nhl.com/ice/news.htm?id=727093
- Tomahawk

thanks. thanks a lot. i'll check back in 6 hours.

thanks. thanks a lot. i'll check back in 6 hours.
- hammarby31

/bow

My logic has a bigger Richard than your logic
- PhillySportsGuy

thanks. thanks a lot. i'll check back in 6 hours.
- hammarby31

See you tonight for the night shift bud. Always a pleasure.

Takes all five guys doing the right thing to get the puck up the ice and in a position to be shot on net... on the other side of the puck, we're all familiar with the idea that defense requires all five guys to work in tandem.

Crediting everybody on the ice for each shot event makes perfect sense in 99% of cases. Only one I can think of that might qualify as false data is if the defense plays everything perfectly but the goalie lets in a softy... otherwise, even a solo end-to-end rush requires the other 4 guys to draw off some defenders, hence doing their jobs off-puck.
- Tomahawk

There is no evidence that it makes perfect sense in 99% of cases. In fact there is plenty of evidence that it is far less then that. Such as maybe 60-70% of cases.

The author makes some really good points, in that if we try to break down each shift, and make a judgement on who is responsible for what, it would be difficult to develop a standard for determining that. Which is why he said he is uncomfortable with that. And he's right. If they used game film to try and remove the false data as much as possible, what is the criteria? How do we remove the human element in making that judgement, and standardize it across the League from team to team to make it a comparable stat. Probably would be very difficult and that further developed stat would have it's own issues. So like you said, Corsi/Fenwick is the best we have to this point. But that doesn't make it reliable for looking at individual players. Or remove the flaws of it.

How bout another one:

"The best way to make sure you're correctly calculating how many feet are in 3.2 miles is to make sure you know 5,280 feet are in a mile."

Babcock said "The best way to use advanced stats is to make sure the numbers are correct"

Well no (frank)in poop Sherlock.
- Giroux_Is_God

Listening to a podcast yesterday I learned the following:

The way most fans and media use advanced statistics has absolutely no relationship to the way teams employ those same numbers because fans and media don't have access to the materials teams have. Teams break down games into individual plays and cross reference them. If they notice a trend in the statistical information they can go and view every occurrence to see how and why the data trends that way. Most teams have stat categories leaps and bounds ahead of what fans and media have access to.

According to some, advanced stats will be the death of the stay at home defenceman. Most likely coming with the advent of the new sports view camera set up.

That's what I was taught yesterday. I'm glad the Leafs hired the boy wonder. Plenty of hockey coverage yesterday to ease my excessive drive time.

Edit: Oh and dump and chase hockey is going the way of the dinosaur thanks to advanced stats

Stats. Numbers. Percentages.

Listening to a podcast yesterday I learned the following:

The way most fans and media use advanced statistics has absolutely no relationship to the way teams employ those same numbers because fans and media don't have access to the materials teams have. Teams break down games into individual plays and cross reference them. If they notice a trend in the statistical information they can go and view every occurrence to see how and why the data trends that way. Most teams have stat categories leaps and bounds ahead of what fans and media have access to.

According to some, advanced stats will be the death of the stay at home defenceman. Most likely coming with the advent of the new sports view camera set up.

That's what I was taught yesterday. I'm glad the Leafs hired the boy wonder. Plenty of hockey coverage yesterday to ease my excessive drive time.
- mayorofangrytown

Never thought of that. Pretty cool.

I love getting the "inside" look at that. That's why I love 24/7 and the locker room footages and the draft stuff. I always think it's so cool just to see a fraction of how much REALLY goes on behind the scenes.

I could do without the "high school freshman using a thesaurus" vocabulary stylings, but I don't hate him. I'm weird, I'd generally rather watch sports without announcers.
- BulliesPhan87

I find that he's trying to catch up on the game frequently because he's been bloviating about some minor uninteresting fact about a 4th line player. When he's actually calling the game, he's really good.

I feel like I remember a few people in here pining for Gary Thorne recently. I hated that guy. I feel like he deliberately botched very easy to pronounce foreign names. Lazy.

The author makes some really good points, in that if we try to break down each shift, and make a judgement on who is responsible for what, it would be difficult to develop a standard for determining that. Which is why he said he is uncomfortable with that. And he's right. If they used game film to try and remove the false data as much as possible, what is the criteria? How do we remove the human element in making that judgement, and standardize it across the League from team to team to make it a comparable stat. Probably would be very difficult and that further developed stat would have it's own issues. So like you said, Corsi/Fenwick is the best we have to this point. But that doesn't make it reliable for looking at individual players. Or remove the flaws of it.
- MJL

Basically that entire false positive/negative narrative was constructed by Staples because he's hawking his own set of "advanced" stats and he's admitted in the past that he digs into Corsi as means to draw attention to his own stuff.

And, as you've mentioned, and as Dellow showed, Neilson numbers are much more problematic... like a sludgy pot of subjectivity.

The way most fans and media use advanced statistics has absolutely no relationship to the way teams employ those same numbers because fans and media don't have access to the materials teams have. Teams break down games into individual plays and cross reference them. If they notice a trend in the statistical information they can go and view every occurrence to see how and why the data trends that way. Most teams have stat categories leaps and bounds ahead of what fans and media have access to.


- mayorofangrytown

This is a great point, and one I often felt to be true. NHL teams aren't logging on to websites and looking at Corsi numbers. They are developing their own more accurate analytics with more accurate ways of gathering data. The principle of Corsi is very sound. It's the data collection that is the issue. No matter how good the software is. Garbage in, garbage out. Not saying that Corsi is garbage.

Basically that entire false positive/negative narrative was constructed by Staples because he's hawking his own set of "advanced" stats and he's admitted in the past that he digs into Corsi as means to draw attention to his own stuff.

And, as you've mentioned, and as Dellow showed, Neilsen numbers are much more problematic... like a sludgy pot of subjectivity.
- Tomahawk

ANALOGY (and/or simile) ALERT

This is a great point, and one I often felt to be true. NHL teams aren't logging on to websites and looking at Corsi numbers. They are developing their own more accurate analytics with more accurate ways of gathering data. The principle of Corsi is very sound. It's the data collection that is the issue. No matter how good the software is. Garbage in, garbage out. Not saying that Corsi is garbage.
- MJL

Can I say its garbage?