HFBoards

HFBoards (http://hfboards.hockeysfuture.com/index.php)
-   By The Numbers (http://hfboards.hockeysfuture.com/forumdisplay.php?f=241)
-   -   Sportsnet.ca 'Shot Quality Project' (EDIT: Now with results!!) (http://hfboards.hockeysfuture.com/showthread.php?t=1521655)

LeafOfBread 10-22-2013 02:52 PM

Sportsnet.ca 'Shot Quality Project' (EDIT: Now with results!!)
 
http://www.sportsnet.ca/hockey/nhl/i...ality-project/

Pretty cool that this stuff is beginning to get more exposure. What do you guys think?

He says he'll be posting the results next week and that you 'won't believe what you see'.

Part 2:

http://www.sportsnet.ca/hockey/intro...oject-part-ii/

LeafOfBread 10-22-2013 10:18 PM

No replies? I thought this would've generated a lot of interest on the boards since people are becoming more and more acquainted with corsi and fenwick analysis

The Latvian 10-22-2013 10:20 PM

I'm sure the results will get more response.

Suntouchable13 10-22-2013 10:21 PM

I hate these made up stats. I will never give them the time of day.

Elever 10-22-2013 10:21 PM

The main board is sorta like the ****** of hfboards. I don't think people care much unless it's funny or unique or controversial so maybe this should get moved to the advanced stats section of the forum.

Anyways, I like this, good visual and has some meaning unlike a good chunk of advanced stats.

LeafOfBread 10-22-2013 10:23 PM

Quote:

Originally Posted by Suntouchable13 (Post 73078411)
I hate these made up stats. I will never give them the time of day.

You do realize the project in the link posted is aiming at quantifying the portion of stats that current advanced stats are missing right?

Arrch 10-22-2013 10:28 PM

I'll wait until the unbelievable results come out before commenting.

biturbo19 10-22-2013 10:36 PM

Seems like a lot of work to ascertain that ultimately, yes...better quality shots are more likely to end up in the net. :laugh:

loobarlow 10-22-2013 11:00 PM

Quote:

Originally Posted by Suntouchable13 (Post 73078411)
I hate these made up stats. I will never give them the time of day.

Yeah, science and math, who cares. They're for sissies!

LeafOfBread 10-23-2013 03:03 PM

Quote:

Originally Posted by loobarlow (Post 73081977)
Yeah, science and math, who cares. They're for sissies!

Lol, that same guy essentially told me "nobody cares don't post about it" when I was bringing up Fenwick and Corsi numbers in the Anaheim vs Toronto GDT before the game to emphasize that Anaheim was a good possession team.

billybudd 10-23-2013 04:13 PM

Maybe this'll be the smoking gun that puts the final nail in the coffin of the "shot quality is just noise" crowd, though I'd argue (and have argued) that scatterplots already did that.

Anyways, looking forward to it.

Master_Of_Districts 10-23-2013 04:41 PM

Quote:

Originally Posted by billybudd (Post 73109493)
Maybe this'll be the smoking gun that puts the final nail in the coffin of the "shot quality is just noise" crowd, though I'd argue (and have argued) that scatterplots already did that.

Anyways, looking forward to it.

Hey Billy - I got a question for you: What do you think the correlation is between a team's scoring chance differential and its shot/fenwick/corsi differential at even strength?

I mean, if shot quality is as important as you say it is, shot/fenwick/corsi differential should be correspondingly less important, and the relationship between shot/fenwick/corsi differential and scoring chance differential must not be very strong, right?

Also, how sustainable do you think team to team differences in shot quality are from one season to the next? I mean, if shot quality isn't just noise, as you're asserting, and is a reflection of a team's underlying ability, the differences should be moderately to highly sustainable, right?

Wesleyy 10-23-2013 05:03 PM

Quote:

Originally Posted by Master_Of_Districts (Post 73110753)
Hey Billy - I got a question for you: What do you think the correlation is between a team's scoring chance differential and its shot/fenwick/corsi differential at even strength?

I mean, if shot quality is as important as you say it is, shot/fenwick/corsi differential should be correspondingly less important, and the relationship between shot/fenwick/corsi differential and scoring chance differential must not be very strong, right?

shot differential is obviously less of a predictor than when the quality of the shot is also taken into account, assuming all data is accurate. and yes, the correlation should be low-moderate between shot differential and scoring chance differential.

Quote:

Originally Posted by Master_Of_Districts (Post 73110753)
Also, how sustainable do you think team to team differences in shot quality are from one season to the next? I mean, if shot quality isn't just noise, as you're asserting, and is a reflection of a team's underlying ability, the differences should be moderately to highly sustainable, right?

Pretty sustainable - Scroll to the bottom of this article. http://hockeymetrics.net/introducing...ls-percentage/

VinnyC 10-23-2013 05:13 PM

I really like that a major outlet picked up on this. Might be the first step towards shedding the brain-dead, numbers-without-context hockey analysis that permeates sports media.

Master_Of_Districts 10-23-2013 05:29 PM

Quote:

Originally Posted by Wesleyy (Post 73111813)
shot differential is obviously less of a predictor than when the quality of the shot is also taken into account, assuming all data is accurate. and yes, the correlation should be low-moderate between shot differential and scoring chance differential.

If you're asserting that the correlation between EV shot/corsi/fenwick differential and EV scoring chance differential at the team level is "low-moderate," you're wrong.

Plain and simple.

Quote:

Pretty sustainable - Scroll to the bottom of this article. http://hockeymetrics.net/introducing...ls-percentage/
The metric discussed in the link is irrelevant - expected goals are calculated on the basis of both shots taken and the quality of those shots. My assertion pertained only to shot quality.

And more importantly, I specifically said team to team differences in shot quality. The link discusses individual players, not teams. I never made any claim regarding differences in shot quality between players.

Noob616 10-23-2013 06:19 PM

Quote:

Originally Posted by VinnyC (Post 73112271)
I really like that a major outlet picked up on this. Might be the first step towards shedding the brain-dead, numbers-without-context hockey analysis that permeates sports media.

Really? I'd say precisely the opposite is what permeates the sports media.

billybudd 10-23-2013 07:36 PM

Quote:

Originally Posted by Master_Of_Districts (Post 73110753)
Hey Billy - I got a question for you: What do you think the correlation is between a team's scoring chance differential and its shot/fenwick/corsi differential at even strength?

I mean, if shot quality is as important as you say it is, shot/fenwick/corsi differential should be correspondingly less important, and the relationship between shot/fenwick/corsi differential and scoring chance differential must not be very strong, right?

I'm sorry, did I somewhere specify how important I believe shot quality to be? I believe I did not.


Quote:

Also, how sustainable do you think team to team differences in shot quality are from one season to the next? I mean, if shot quality isn't just noise, as you're asserting, and is a reflection of a team's underlying ability, the differences should be moderately to highly sustainable, right?
Is there an agreed upon measurement for shot quality that I'm to be making estimates based on (apart from that of the "there's no such thing because we have a model that doesn't use it" people. Interestingly the same thing was said of the idea of a heliocentric (edit: geocentric :facepalm: ) view of the solar system)?

The answer is, of course, no. If there was, this article would have nothing to tease.

Scatterplots and heat maps will tell you porous defensive clubs like Buffalo give up a higher percentage of shots from closer than a less porous one like Pittsburgh, but that's proof of existence, not a model. Which is again, the entire reason sportnet is teasing this article.

Now let me ask you a question in reference to the bold:

Is it your belief that year to year there are teams with

a) players that do not age
b) rosters that do not change
c) coaches that do not alter strategies

and

d) that the opponents of these teams have players that do not age, rosters that do not change, and coaches who make no strategic adjustments in comparison to previous season?

Seems like a rather odd request that I use a model that isn't public to evaluate its year to year effect on a static system that isn't static.

Wesleyy 10-23-2013 11:01 PM

Quote:

Originally Posted by Master_Of_Districts (Post 73112937)
If you're asserting that the correlation between EV shot/corsi/fenwick differential and EV scoring chance differential at the team level is "low-moderate," you're wrong.

Plain and simple.

:facepalm: This is honestly why I don't bother commenting on these threads. Your response does not contribute to the conversation at all besides asserting your believe based on the fact that you can't be wrong.

Quote:

Originally Posted by Master_Of_Districts (Post 73112937)
The metric discussed in the link is irrelevant - expected goals are calculated on the basis of both shots taken and the quality of those shots. My assertion pertained only to shot quality.

That's a good point; I misread your comment.

Master_Of_Districts 10-24-2013 04:03 PM

Quote:

Originally Posted by Wesleyy (Post 73130249)
:facepalm: This is honestly why I don't bother commenting on these threads. Your response does not contribute to the conversation at all besides asserting your believe based on the fact that you can't be wrong.

I apologize for that - the incivility from my end was inappropriate.

However, my belief regarding the strength of correlation between outchancing and outshooting at the team level does, in fact, have a factual basis.

For example, quite a few bloggers counted scoring chances for a variety of teams over the 2010-11 NHL season. I happened to have collected data from all games for which scoring chances were counted. It turns out that there was scoring chances data for 386 out of the 1230 games played that year, which isn't bad.

For that 386 game sample, I took the trouble of calculating each team's:

1. Scoring Chance Ratio
2. Fenwick Ratio
3. Corsi Ratio

The correlation between Scoring Chance Ratio and Fenwick Ratio was substantial, at 0.82. The correlation between Scoring Chance Ratio and Corsi Ratio was high as well, although somewhat less so, at 0.69.

As both scoring chance ratio and fenwick/corsi ratio have a reliability lower than 1 over the sample in question, it bears mentioning that the correlations would be even higher if the correlations were disattenuated to account for this factor.

One problem with the data is that the number of games included for each team was not uniform. For example, there were only nine Chicago games in the sample, but 82 games from Montreal and Edmonton. This has the potential to skew things.

Fortunately, the correlations don't change much if we weight each team's data by the number of games played (or more precisely, aggregate fenwick for and against).

For example, the correlation between Fenwick Ratio and Scoring Chance Ratio drops only slightly to 0.81.

I intend to the same thing for the 2011-12 season once I obtain the necessary data.

Master_Of_Districts 10-24-2013 05:11 PM

Quote:

Originally Posted by billybudd (Post 73117513)
I'm sorry, did I somewhere specify how important I believe shot quality to be? I believe I did not.




Is there an agreed upon measurement for shot quality that I'm to be making estimates based on (apart from that of the "there's no such thing because we have a model that doesn't use it" people. Interestingly the same thing was said of the idea of a heliocentric view of the solar system)?

The answer is, of course, no. If there was, this article would have nothing to tease.

Scatterplots and heat maps will tell you porous defensive clubs like Buffalo give up a higher percentage of shots from closer than a less porous one like Pittsburgh, but that's proof of existence, not a model. Which is again, the entire reason sportnet is teasing this article.

Now let me ask you a question in reference to the bold:

Is it your belief that year to year there are teams with

a) players that do not age
b) rosters that do not change
c) coaches that do not alter strategies

and

d) that the opponents of these teams have players that do not age, rosters that do not change, and coaches who make no strategic adjustments in comparison to previous season?

Seems like a rather odd request that I use a model that isn't public to evaluate its year to year effect on a static system that isn't static.

Re-reading everything, you're correct in that you never made any claim regarding the importance of shot quality - your claim was merely that it exists.

I apologize.

Wesleyy 10-24-2013 08:45 PM

Quote:

Originally Posted by Master_Of_Districts (Post 73159153)
I apologize for that - the incivility from my end was inappropriate.

However, my belief regarding the strength of correlation between outchancing and outshooting at the team level does, in fact, have a factual basis.

For example, quite a few bloggers counted scoring chances for a variety of teams over the 2010-11 NHL season. I happened to have collected data from all games for which scoring chances were counted. It turns out that there was scoring chances data for 386 out of the 1230 games played that year, which isn't bad.

For that 386 game sample, I took the trouble of calculating each team's:

1. Scoring Chance Ratio
2. Fenwick Ratio
3. Corsi Ratio

The correlation between Scoring Chance Ratio and Fenwick Ratio was substantial, at 0.82. The correlation between Scoring Chance Ratio and Corsi Ratio was high as well, although somewhat less so, at 0.69.

As both scoring chance ratio and fenwick/corsi ratio have a reliability lower than 1 over the sample in question, it bears mentioning that the correlations would be even higher if the correlations were disattenuated to account for this factor.

One problem with the data is that the number of games included for each team was not uniform. For example, there were only nine Chicago games in the sample, but 82 games from Montreal and Edmonton. This has the potential to skew things.

Fortunately, the correlations don't change much if we weight each team's data by the number of games played (or more precisely, aggregate fenwick for and against).

For example, the correlation between Fenwick Ratio and Scoring Chance Ratio drops only slightly to 0.81.

I intend to the same thing for the 2011-12 season once I obtain the necessary data.

What about (scoring chance differential)/(shot diffential) correlation between games? I assumed the correlation to be moderate between shot differential and scoring chance because having more shots shouldn't necessarily mean that the shot quality is higher. I guess you could also say that the dominating team (in shot differential) would naturally mean that they are the better team and therefore should get more scoring chances. It depends on what question is asked, is shot differential a accurate predictor of chances, or does a team outshooting their opponent tend to create better scoring chances (neglecting the obvious fact that the team will have more attempts at turning a shot into a scoring chance).

billybudd 10-25-2013 01:59 AM

Quote:

Originally Posted by Master_Of_Districts (Post 73161841)
Re-reading everything, you're correct in that you never made any claim regarding the importance of shot quality - your claim was merely that it exists.

I apologize.

No worries

Canada4Gold 10-29-2013 02:47 PM

next article on this has been posted

http://www.sportsnet.ca/hockey/intro...oject-part-ii/

Hammer Time 10-29-2013 07:43 PM

Quote:

Originally Posted by Canada4Gold (Post 73451271)
next article on this has been posted

http://www.sportsnet.ca/hockey/intro...oject-part-ii/

Very interesting work. So far the results don't seem to be anything surprising - definitely cross-ice passes and deflections are how a lot of goals get scored, while pretty much any NHL goalie can consistently stop the unscreened shot. But once this study is complete, we can start figuring out how to quantify shot quality, and that is a big step forward for hockey analytics. :handclap:

DopeyFish 10-29-2013 11:35 PM

Quote:

Originally Posted by Hammer Time (Post 73466847)
Very interesting work. So far the results don't seem to be anything surprising - definitely cross-ice passes and deflections are how a lot of goals get scored, while pretty much any NHL goalie can consistently stop the unscreened shot. But once this study is complete, we can start figuring out how to quantify shot quality, and that is a big step forward for hockey analytics. :handclap:

i was starting work on my SONQ stat (shot on net quality)

essentially... I was originally going to look through game tape (A LOT) and split the ice into zones

see if goalie is in position when shot is taken, if its a one timer (with and without windup), tipped, player moving, wind up slap, snap, slap

you go through about 3-5~ years worth of data, you can get fairly decent figures in terms of shooting percentage

then you can add up all the shoot% to get a total offensive output (and as a result, total defensive) and it all would plug in nicely with CORSI/fenwick and would make nice adjustments to QoT and QoC

it's really the missing piece in NHL stats and could provide SOOOOO MUCH information


All times are GMT -5. The time now is 12:05 PM.

vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
HFBoards.com, A property of CraveOnline, a division of AtomicOnline LLC ©2009 CraveOnline Media, LLC. All Rights Reserved.