Question to advanced stats proponents

[SnowblindNYR]

I'm not sure if I should post this here or in the "by the numbers" thread but advanced stats get used here so often and this is a general thread. I have literally yet to see an advanced stats proponent that used the stats simply as a guide. It seems like you guys seem to use it as black and white and seem to think that common stats and legitimate watching of the game is not important. I seriously don't want this to be a "bashing thread" and want to know if you guys think that there is no room for subjective interpretation and you can just look at Corsi and Fenwick to determine objectively which players are the best, which teams are playing the best, which teams are getting lucky etc. I seriously would be interesting in knowing that, especially as stats and numbers enthusiast.

Click to read more...

 

[Bear of Bad News]

First, despite your exhortations, you seem to be setting this up as a bashing thread. If that's genuinely not your intent, I apologize.

Second, "us guys" actually have a multitude of opinions and thoughts. Would you like it if all Rangers fans were characterized with the same opinion? Believe it or not, we actually disagree with one another quite a bit.

Third, of course there's subjectiveness, and there's no one-size-fits-all statistics to measure hockey.

I recommend reading the threads on this subforum, although I'm not sure where you're getting your conclusions about "us guys".

 

[Mathletic]

I'm not sure if I should post this here or in the "by the numbers" thread but advanced stats get used here so often and this is a general thread. I have literally yet to see an advanced stats proponent that used the stats simply as a guide. It seems like you guys seem to use it as black and white and seem to think that common stats and legitimate watching of the game is not important. I seriously don't want this to be a "bashing thread" and want to know if you guys think that there is no room for subjective interpretation and you can just look at Corsi and Fenwick to determine objectively which players are the best, which teams are playing the best, which teams are getting lucky etc. I seriously would be interesting in knowing that, especially as stats and numbers enthusiast.

This would get ugly quick on any other board That said, pretty sure you'll get good responses here. People who take part of this board are for the most part pretty open minded to begin with, otherwise they wouldn't come to think of hockey in a different way. From my experience, it's not even close as to which "group" of people is more open minded to the ideas of the other group.

Besides, I really don't think anyone here thinks that Corsi is the be-all and end-all in analyzing hockey. I doubt anyone here thinks that Justin Williams is the best player in the NHL.

 

[SnowblindNYR]

First, despite your exhortations, you seem to be setting this up as a bashing thread. If that's genuinely not your intent, I apologize.

Second, "us guys" actually have a multitude of opinions and thoughts. Would you like it if all Rangers fans were characterized with the same opinion? Believe it or not, we actually disagree with one another quite a bit.

Third, of course there's subjectiveness, and there's no one-size-fits-all statistics to measure hockey.

I recommend reading the threads on this subforum, although I'm not sure where you're getting your conclusions about "us guys".

First of all, full disclosure, it gets really difficult sometimes to post on the main board when someone is an advanced stats proponent because it's often used as dogma. Pretty much every discussion on the main board that involves such proponents features a pretty inflexible opinion like it's a subjective fact. It ranges from respectful but unwavering to full out feisty. I paint said proponents with a broad brush because my experience is that there are a lot of commonalities between the proponents. We have a very respected and respectful poster on our board that has an interesting opinion that a lot of people disagree with (he's actually convinced me to hope that AV gives his idea a try). But it's really hard to argue with him because every argument is a statistic and is treated as dogma, like there's no subjectivity. I guess my once in a while annoyance may have been betrayed in the thread. That said, I legitimately wanted to know, because advanced stats people that I've come across are so unwavering with that the stats tell the whole story, I really was curious, if it's considered pretty much objective. For example, if you are playing roulette and play 1 number and win 5 spins in a row, you've very lucky. Those types of statistics are objective truths. Now is it objective truth that a team with a low Corsi winning a lot is very lucky? So maybe these stats were highly objective. So while my annoyance may have slipped through in the tone (I apologize), I legitimately want to know.

 

[SnowblindNYR]

This would get ugly quick on any other board That said, pretty sure you'll get good responses here. People who take part of this board are for the most part pretty open minded to begin with, otherwise they wouldn't come to think of hockey in a different way. From my experience, it's not even close as to which "group" of people is more open minded to the ideas of the other group.

Besides, I really don't think anyone here thinks that Corsi is the be-all and end-all in analyzing hockey. I doubt anyone here thinks that Justin Williams is the best player in the NHL.

I really don't want to call anyone out so I hope I keep this as obscure as possible, but I recently read the opinion that Anze Kopitar is the best player in the NHL because of "stats that matter". That's the kind of unwavering opinion I'm talking about.

 

[Bear of Bad News]

I really don't want to call anyone out so I hope I keep this as obscure as possible, but I recently read the opinion that Anze Kopitar is the best player in the NHL because of "stats that matter". That's the kind of unwavering opinion I'm talking about.

Supposing that this is true (and I have no reason to doubt you), that's one person's opinion. Some people develop an opinion, and then use whatever supporting evidence they can find to make their opinion sound credible. If "advanced stats" are the key to making that happen, then so be it. If "I've seen every Kings game since 1983" is the key, then so be it. Supporting the opinion is the goal.

Others use analytics to explore concepts, without any preconceived notions. You'll find more of that sort of individual in this forum, although we do all have biases (certainly). Most of us like exploring for exploring's sake.

 

[SnowblindNYR]

Supposing that this is true (and I have no reason to doubt you), that's one person's opinion. Some people develop an opinion, and then use whatever supporting evidence they can find to make their opinion sound credible. If "advanced stats" are the key to making that happen, then so be it. If "I've seen every Kings game since 1983" is the key, then so be it. Supporting the opinion is the goal.

Others use analytics to explore concepts, without any preconceived notions. You'll find more of that sort of individual in this forum, although we do all have biases (certainly). Most of us like exploring for exploring's sake.

This board seems a lot more open minded. I remember during the 11-12 season somebody on the main board would say how much the Rangers suck because of advanced stats and they'll fall down to earth. Well they finished with 109 points and a second half 100 point pace and made the ECF. Now I really think that that team overachieved. My point is that it seems not everyone thinks like that. It seems like the difference between saying "if you flip a coin 100 times, it's almost improbable that you don't have at least 1 Head" rather than "if you flip a coin 100 times, you're definitely getting a Head at least once".

 

[TheDevilMadeMe]

This board seems a lot more open minded. I remember during the 11-12 season somebody on the main board would say how much the Rangers suck because of advanced stats and they'll fall down to earth. Well they finished with 109 points and a second half 100 point pace and made the ECF. Now I really think that that team overachieved. My point is that it seems not everyone thinks like that. It seems like the difference between saying "if you flip a coin 100 times, it's almost improbable that you don't have at least 1 Head" rather than "if you flip a coin 100 times, you're definitely getting a Head at least once".

There are a refreshing number of posters on this board who are highly critical of the overuse and misuse of Corsi and Fenwick that had spread over the internet like a plague a couple of years ago.

 

[Mosby]

To piggyback on this thread, where would you direct a Corsi/advanced stat newbie like myself? What are the best books or websites dedicated to the matter?

 

[kdfsjljklgjfg]

I'm not sure if I should post this here or in the "by the numbers" thread but advanced stats get used here so often and this is a general thread. I have literally yet to see an advanced stats proponent that used the stats simply as a guide. It seems like you guys seem to use it as black and white and seem to think that common stats and legitimate watching of the game is not important. I seriously don't want this to be a "bashing thread" and want to know if you guys think that there is no room for subjective interpretation and you can just look at Corsi and Fenwick to determine objectively which players are the best, which teams are playing the best, which teams are getting lucky etc. I seriously would be interesting in knowing that, especially as stats and numbers enthusiast.

The idea isn't that "Player A has a higher Y stat than player B, so he's better"

The idea is that if you love something, it makes sense to want to understand it further. To be able to quantify what you see. "Player A has a higher Y stat" sounds better than "I saw player A do it really well", especially when it's difficult to have a fair chance to see everyone in the league play.

Even the most rabid fan who goes to as many games as humanly possible isn't going to have seen enough to give the eye test on every possible situation for every player in the league, and even then, you have to try to sort opinion from fact because everyone has their own idea of how things should be done. To one person, cherry-picking is abandoning the defense and giving up chances. To another person, cherry-picking could be taking a defensive risk to gain a greater chance to score for the greater good.

When it comes down to it, advanced stats are immune from bias and opinion. You're more likely to win an argument by pointing out a player's Fenwick than saying "I saw him do it and it looked like he did a good job".

There's 2 sayings to keep in mind:
The stats don't lie
The numbers don't tell the whole story

So long as you don't take them as absolute truths, and acknowledge that they only paint a piece of the picture, advanced stats are the best way to quantify the game without using opinion or allowing bias to come into play.

 

[sharks9]

To piggyback on this thread, where would you direct a Corsi/advanced stat newbie like myself? What are the best books or websites dedicated to the matter?

Here are some links to introductions about various advanced stats

http://nhlnumbers.com/2013/10/10/a-graphic-guide-to-advanced-hockey-stats

http://www.grantland.com/blog/the-t...the-fakers-guide-to-advanced-stats-in-the-nhl

http://www.habseyesontheprize.com/fancy-stats

 

[Alan Jackson]

I think stats like CORSI and FENWICK can be helpful when considered with some context, but aren't they really team-generated numbers?

How comfortable are we using possession stats to judge individuals when they can be largely dependant on who you play with or the situations you're put in?

 

[SnowblindNYR]

Damn reading about how big of a role luck plays in winning games makes me wonder why I even celebrate wins.

 

[SnowblindNYR]

Here are some links to introductions about various advanced stats

http://nhlnumbers.com/2013/10/10/a-graphic-guide-to-advanced-hockey-stats

http://www.grantland.com/blog/the-t...the-fakers-guide-to-advanced-stats-in-the-nhl

http://www.habseyesontheprize.com/fancy-stats

Damn, thanks. All I needed is another numbers obsession. I've always like numbers and hockey but for some reason never got around to Corsi. I even enjoyed Moneyball and I HATE baseball.

 

[Michael Gary Scott]

Damn, thanks. All I needed is another numbers obsession. I've always like numbers and hockey but for some reason never got around to Corsi. I even enjoyed Moneyball and I HATE baseball.

Extraskater.com is my favorite for finding statistics of these nature. Handy sorting tools and all kinds of depth.

 

[wgknestrick]

Here are my goto links:


Only site that has multi-year capabilities (That I've found) for sortable pt/min, SH%, G%, etc in all situations
http://stats.hockeyanalysis.com/ratings.php

Here is their blog page with some good (albeit TOR focused) articles
http://hockeyanalysis.com/

Great site for visual shot locations of players
http://www.sportingcharts.com/nhl/icetrack/

This is their normal stat page - good year to year graphs
http://www.sportingcharts.com/stats/nhl/player/

Old reliable - stat links at bottom
http://www.behindthenet.ca/

Carolina blog page -MUST READ- when your team plays the Canes. Gives detailed scoring chance data. I hate that no one does this for every NHL game, and that my team isn't the Canes. Then I realize my team is better than the Canes .
http://www.shutdownline.com/

Hockey Prospectus - used to be my favorite site, but has fallen a bit since ESPN plucked a couple of their mainstay writers for the ESPN insider. Home of GVT
http://www.hockeyprospectus.com/puck/wp/ - new pay site I guess -
http://www.hockeyprospectus.com/
http://www.hockeyprospectus.com/sortable/

 

[hella rights]

I think stats like CORSI and FENWICK can be helpful when considered with some context, but aren't they really team-generated numbers?

How comfortable are we using possession stats to judge individuals when they can be largely dependant on who you play with or the situations you're put in?

This is a great point, assigning values to individuals based on group outcomes can definitely be problematic.

 

[3 Minute Minor]

There are a refreshing number of posters on this board who are highly critical of the overuse and misuse of Corsi and Fenwick that had spread over the internet like a plague a couple of years ago.

Blame Moneyball

People were so focused on the numbers they forgot that baseball numbers make sense because it's pretty much an individual sport whereas hockey depends on so many variables that the math is impossible.

 

[CarpeNoctem]

Blame Moneyball

People were so focused on the numbers they forgot that baseball numbers make sense because it's pretty much an individual sport whereas hockey depends on so many variables that the math is impossible.

I'm right there with you. Corsi and all that crap is very easy to misuse in a team sport, where the top player has maybe 25 min TOI on average.

 

[Mathletic]

Blame Moneyball

People were so focused on the numbers they forgot that baseball numbers make sense because it's pretty much an individual sport whereas hockey depends on so many variables that the math is impossible.

I respect your opinion but hockey is such an easy sport to forecast performance compared to baseball, it's not really close. Besides, basketball is also a team sport, very much like hockey, and I consider the statistical analysis that has been done there has surpassed that in baseball by quite a fair margin. It's a given that if you approach a task telling yourself, well it's just not going to work. It's likely you won't find anything there. But anwyays, we'll always have that kind of debate until at least a team gives it a shot for good.

 

[Bear of Bad News]

I respect your opinion but hockey is such an easy sport to forecast performance compared to baseball, it's not really close.

Tell me more about this. Do you mean team performance or individual performance? In either case, what measures of performance are being forecasted? What studies are there to show that it's much easier to do than baseball?

 

[Mathletic]

Tell me more about this. Do you mean team performance or individual performance? In either case, what measures of performance are being forecasted? What studies are there to show that it's much easier to do than baseball?

From a team performance and matchups standpoint we can look at vegas odds for example. Odds given for baseball games are much tighter than in hockey. I'd even go as far as saying that despite tighter odds, on average, there's even a bit of a bias in favor of the favorite teams in baseball.

For example, tonight the Habs were favorite at -308 vs. the Sabres. It's the 5th time already this year we've seen a team favored at at least -300. The other matchups were Buffalo vs. LA, Calgary vs. Colorado, Carolina vs. St. Louis and Carolina vs. Boston. All favorites won their game.

In baseball, there were only 3 such matchups rated at -300 or higher for the favorite team all season long. The 3 being: Houston in Tampa and David Price. Houston in Detroit and Doug Fister and the Cubs in LA with Kershaw pitching. In case you didn't notice, 2 of those favorites actually lost their games against the weaker opponents. I think most would agree that we'd think of those matchups as absolutely lop-sided. Yet we wouldn't think of Colorado, for example, as having the same shot at beating Calgary at home than say Kershaw beating the Cubs at home. Although this is a small sample, you'll find the same pattern over many years of data.

There tends to be a lot more randomness in baseball than there is in hockey. Despite isolating for a lot of stuff, baseball is actually very hard to predict. On the other hand, even if hockey may seem to be much harder to grasp, it tends to be quite predictable. At least from a team and matchup perspective. Basketball is even easier to predict given the disparity between teams (simply from a money line perspective, in other words, who wins the game).

I can't point out any study or objective data to prove my point on an individual level. All I have are my own models, which I won't share.

 

[Bear of Bad News]

So to translate what you're saying - it's easier to predict the outcome of a single (given) baseball than a single (given) hockey game?

Give the nature of starting pitching, I don't find that surprising.

 

[habsfanatics*]

So to translate what you're saying - it's easier to predict the outcome of a single (given) baseball than a single (given) hockey game?

Give the nature of starting pitching, I don't find that surprising.

I must be extremely tired, because I interpreted what he was saying to be the exact opposite.

 

[Mathletic]

So to translate what you're saying - it's easier to predict the outcome of a single (given) baseball than a single (given) hockey game?

Give the nature of starting pitching, I don't find that surprising.

no, it's easier to predict the outcome of a hockey game than a baseball game

I should have added more information. In sports betting, -300 means that the team has to have a 75% chance of winning the game in order to break even on that bet. At -350, you need a 78% chance of winning and so on. This translates to, bookies (bettors in general), find it easier to predict the outcome of hockey games than baseball games.