Beyond Corsi - Let's Dig a Little Deeper

[Pyromancer]

Finally, a rant beyond my wildest expectations!

I totally agree. What a comprehensive analysis of advanced stats based on intangibles & reality.

 

We can sit here & complain about advanced stats but Flame111 dove right into the religion and then realized it is perhaps not what many would have you believe.

 

The hockey gods now appear to scoff at advanced stats.

Perhaps cccsberg & kehatch can make us believe again.

[kehatch]

I can say it ten thousand times but the detractors won't listen. I am NOT saying that they provide a secret recipe. Barely anyone is. It provides information. Knowing how many face off a player is taking in the D zone vs O zone is good info. Know how many times a shot is directed against each net while a player is on the ice is good info. Knowing how well a player performs with player x vs player y can be helpful.

If someone is telling you they can tell the entire quality of a team or player based on a single stat then they don't know what they are talking about. The fact that some moron tries to use them to suggest the Oilers are good and the Flames are bad doesn't make the data bad.

In a week or two they will be accessible on www.nhl.com. People will get used to them. Eventually stats like Corsi won't mean anything more then blocked shots, hits, or take aways. It's just one stat. Most of us that follow it know that. It's the detractors that are usually trying to make it out to be something more then it is.

[Flyerfan52]

I can say it ten thousand times but the detractors won't listen. I am NOT saying that they provide a secret recipe. Barely anyone is. It provides information. Knowing how many face off a player is taking in the D zone vs O zone is good info. Know how many times a shot is directed against each net while a player is on the ice is good info. Knowing how well a player performs with player x vs player y can be helpful.

If someone is telling you they can tell the entire quality of a team or player based on a single stat then they don't know what they are talking about. The fact that some moron tries to use them to suggest the Oilers are good and the Flames are bad doesn't make the data bad.

In a week or two they will be accessible on www.nhl.com. People will get used to them. Eventually stats like Corsi won't mean anything more then blocked shots, hits, or take aways. It's just one stat. Most of us that follow it know that. It's the detractors that are usually trying to make it out to be something more then it is.

What is the fancy stats take on the new & improved Dubnyk? Not a dig, mere curiousity.

 

(BTW, if it kills a cat I don't care. )

[kehatch]

What is the fancy stats take on the new & improved Dubnyk? Not a dig, mere curiousity.

(BTW, if it kills a cat I don't care. )

It doesn't take too much crunching to know that a guy with a career SV% of 0.911 can't sustain 0.940. It's not like Minnesota has been easy on its goalies this season.

But I don't think that means he is going to fall off the planet or that Minnesota needs him to be that good to stay competitive.

If you look at Minnesota's 6 game winning streak they had a PDO (save percentage plus shooting precentage) of 1.05 and a save percentage of 0.96. That isn't sustainable. But it didn't need to be. Minnesota out shot their opponents 30 to 26 over that span. They were only out shot once.

They can't outscore opponents 16 to 6 like they did over that span. But you don't need that kind of performance to win games.

The question isn't can Dubnyk keep playing like that. Of course he can't. The question is can Minnesota keep out shooting their opponents and can they keep scoring goals. If they can they will keep winning even when Dubnyk slows down.

[KACaribou]

The topic has meandered granted, but the point was the Analytics guy from TSN was calling the Flames a bad hockey team on TSN Radio 1260 in Edmonton. It wasn't Corsi at all. He just wanted to prove though stats that the Flames were every bit as bad at Toronto or Edmonton, but lucky. The TNS analytics guys is like a bad scientist who tries to fit the results of his experiment into his pre-conceived formula, rather than develop a formula after the results of the experiment.

[cccsberg]

Analytics are fine. Peoples conclusions are sometimes suspect.

That said, it's tough to deny that the Flames have had their share of luck this season. Forget about advanced statistics. How many times have we won in OT? How many times have we come back in the third? How many times have we scored with our new empty?

The answers to those questions isn't all luck. Good fitness. Never say die attitude. Etc. But luck is a partial factor.

Calgary has had their share of luck, and their share of misfortune.  I don't think its the reason the season has gone so well.

 

I believe the best answer for the third periods and overtime is the skating, and yes, possession.  Calgary has a small group of fast, great skaters that can control the play and use that effectively to create turnovers and create odd-man or otherwise great shot attempts that go in at a higher average than their normal.  This is especially so 4-4 when their defensemen are often a cut above the opposition.  When there are fewer skaters and more open ice these things become much more prominent and important.  

 

The other thing that's really key is the Flames work hard, and by that I mean they drive to the net well and are tenacious on the forecheck, as well as swatting the puck out of harm's way.  The last thing is they block shots, in fact they lead the NHL in blocking shots.  Because they do so well blocking shots their Corsi typically looks terrible but if the shots don't get through they won't go in.  The Corsi may look bad but really that's a poor measure of success.

 

I was reading an article on TSN yesterday about the Predictive Value of stats, specifically Corsi, Goal Differential and Winning %, for the whole NHL over a period of several years.  The results showed that their was a rough correlation for each, with the three having more value in the order I listed.  However the predictive value at the very best was 0.35-0.4, meaning that even the very best data only could predict 1/3 to 40% of the time.  Or another way to look at it is a coin flip would have more predictive value...   The other thing of note was that all three measures were within 0.05 of each other.  NOT what you're really looking for.

 

I was thinking the other day that I should look at the Flames this year and every category that they currently lead the NHL (or better, Top/Bottom 5) in stats as a way to focus in on the things that they are exceptional at to see the drivers to their success.  Perhaps I will over the next couple days.  But in the meantime, anyone help me out here?

[DirtyDeeds]

The topic has meandered granted, but the point was the Analytics guy from TSN was calling the Flames a bad hockey team on TSN Radio 1260 in Edmonton. It wasn't Corsi at all. He just wanted to prove though stats that the Flames were every bit as bad at Toronto or Edmonton, but lucky. The TNS analytics guys is like a bad scientist who tries to fit the results of his experiment into his pre-conceived formula, rather than develop a formula after the results of the experiment.

Some people around here do not understand that to gain acceptance in Edmonton they need only to bash the Flames.....

 

It does not matter what they say, or what the message is, as long as it bashes the Flames then a good portion of Oiler Fans will accept the message overall.

 

I found this to be totally true on the Oiler forums.. Soon as a few of the posters found out I was a Flames Fan then I was open  season to bash. It did not matter what my message was, I was wrong...

[KACaribou]

It is just frustrating listening to analytics fans rather than hockey fans. Two different animals altogether. Oh, and by the way the Oilers hired an analytics guy and have a lot of management that believed in analytics. They made many of their changes going into this season based on analytics. Analytically speaking, they really thought they had a good team going into the season according to the numbers they crunched.

[tmac70]

Some people around here do not understand that to gain acceptance in Edmonton they need only to bash the Flames.....

 

It does not matter what they say, or what the message is, as long as it bashes the Flames then a good portion of Oiler Fans will accept the message overall.

 

I found this to be totally true on the Oiler forums.. Soon as a few of the posters found out I was a Flames Fan then I was open  season to bash. It did not matter what my message was, I was wrong...

Cause they don't understand hockey logic on the Oiler board. Oiler fans live in a different reality.

[Flame111]

[Flame111]

[Flamespatriot]

I always found the stat to be over rated and inaccurate just like Plus minus

[darth_henning]

The +/- I find does have its relevance. But more-so for defencemen than for the forwards.

 

For the forwards it really depends on line matchups, and what role they're being used in. For the defence it really does show how good a role they're playing in defending the net from being scored on.

 

Also, its a simple stat to understand and read after a 1 sentence explanation.

 

What bugs me about Corsi is that I have yet to actualy get (after several paragraph attempts) the slightest understanding of what they are and what they're supposed to tel you. Or is it called Fenwick? Or....

[DirtyDeeds]

In its day +/- was also an advanced stat. The NHL had a +/- award too but took that out after the 2008 season. 

 

So just because the NHL has decided to track some newer advanced stats does not mean much.........

[kehatch]

The thing that I find frustrating is that fans go out of their way to bash it without any understanding of what they are. If your not interested that's fine. If you don't agree with what someone is using them to say that's fine. But every time someone posts anything to do with advanced stats the entire discussion is derailed.

It's the stereotypical arguments too. The Oiler's hired an analytics person so advanced stats must be bad. Someone mentioned CORSI in an attempt to say the Oilers aren't bad so advanced stats must be bad. The data is telling me something I don't want to hear so advanced stats must be bad.

Stats guys can't even debate back because the entire argument is based around the NA NA NA NA BOO BOO I CAN'T HEAR YOU defence.

I had originally made an effort to clear this single thread so stats people had somewhere to discuss the stats rather then hear about why people are passionately ignorant about them. But apparently that's what this post is for again. Oh well.

[Carty]

The thing that I find frustrating is that fans go out of their way to bash it without any understanding of what they are. If your not interested that's fine. If you don't agree with what someone is using them to say that's fine. But every time someone posts anything to do with advanced stats the entire discussion is derailed.

 

You are assuming that those of us that do not agree with how some stats are used, or even dispute some stats validity due to exclusions in the way in which they are accumulated know nothing about statistics...   Sounds like Flame111 has delved into advanced stats and chosen to dispute their validity, and he is not the only one...

 

Stats guys can't even debate back because the entire argument is based around the NA NA NA NA BOO BOO I CAN'T HEAR YOU defence.  

 

Using an insult as a reason would seem to be a little bit ironic...

 

I had originally made an effort to clear this single thread so stats people had somewhere to discuss the stats rather then hear about why people are passionately ignorant about them.

 

If mods started deleting every post they did not agree with, what would the purpose of the board be ?   to become someones blog for the promotion of their personal opinion as opposed to a board for discussion?    and to call those who do not agree "passionately ignorant" would seem to be a little bit harsh for an assumption that just because they do not agree with someone on something that they are not informed on the subject...

 

Speaking for myself I have had a difficult time finding examples of advanced stats that can be shown to be consistently accurate, but when examples of advanced stats can be shown to be accurate and relevant I for one will be more than willing to give them serious consideration...

 

[kehatch]

1. There are some detractors that have actually bothered to learn about them.

2. I didn't insult anyone. But look above. It's full of silly pictures bashing the stats without context.

3. Not every stats thread needs to be a debate on the merits of advanced stats.

[Flyerfan52]

It's when the advanced stats say that bottom feeders are really much better teams then the 1s actually making the playoffs that they seem rather irrelevant.

We see games. We read standings. Yet advanced stats say this shouldn't be happening.

It depends if you believe in charts or reality.

***********************************************************

With advanced stats how many playoff contenders have the #s? According to them the Jets & Flames don't. 29th place Oilers do.

 

Go back 5 years. How accurate were their #s on who competed & who pretended. Polls allow for a margin of 3-5%. Were they right 95-97% of the time?

Remember the Oilers & Isles were both supposed to be in the playoffs last year. Same this year. The Isles will be but I predicted that even before they added Boychuk & Leddy. The Oilers brain lack are trying to figure how they can lose every game this season while pretending not to tank.

[kehatch]

It's when the advanced stats say that bottom feeders are really much better teams then the 1s actually making the playoffs that they seem rather irrelevant.

We see games. We read standings. Yet advanced stats say this shouldn't be happening.

It depends if you believe in charts or reality.

***********************************************************

With advanced stats how many playoff contenders have the #s? According to them the Jets & Flames don't. 29th place Oilers do.

Go back 5 years. How accurate were their #s on who competed & who pretended. Polls allow for a margin of 3-5%. Were they right 95-97% of the time?

Remember the Oilers & Isles were both supposed to be in the playoffs last year. Same this year. The Isles will be but I predicted that even before they added Boychuk & Leddy. The Oilers brain lack are trying to figure how they can lose every game this season while pretending not to tank.

You keep saying this FF. Stats don't SAY anything. It's like saying that the blocked shots say that team X is good but team x finished low in the I so blocked shots are irrelevant. Corsi just shows shot attempts. Pdo just shows the contrast of save percentage vs shot percentage. Zone starts just shows where a player is starting their shifts. Zone entries just shows how often a player or team carries or dumps the puck in. Etc.

It's just info. Like shots, or assists, or PP percentage, or face offs, or hits, or blocked shots.

Some people try and draw conclusions. Such as team x has a higher Corsi so they must be the better team. Or the PDO is high or low so the team must be lucky or unlucky. Sometimes they are wrong. It doesn't make the data bad.

I get that their is too much value out on Corsi in the media and even by some stat sights. Heck, quality of competition and similar metrics use Corsi as the underlying metric. Some groups use Corsi like on base percentage was used on moneyball. Corsi = success isn't proven and may be a flawed assumption.

But none of that makes the data irrelevant. Before all of this debate there were a number of examples on how advanced stats can be used. In fact, the post itself is beyond Corsi and acknowledges some of its limitations.

But despite that, someone posts an article about some dufus using Corsi to draw an unsupported conclusion so we are having the same debate.

I say it AGAIN. Stats provide useful information. Some of us consider it interesting information. None of us are saying the Oilers are good or the Flames are bad. Just because there are stupid people using good information in bad ways isn't a reason to derail every stats based discussion.

[Kulstad]

You keep saying this FF. Stats don't SAY anything. It's like saying that the blocked shots say that team X is good but team x finished low in the I so blocked shots are irrelevant. Corsi just shows shot attempts. Pdo just shows the contrast of save percentage vs shot percentage. Zone starts just shows where a player is starting their shifts. Zone entries just shows how often a player or team carries or dumps the puck in. Etc.

It's just info. Like shots, or assists, or PP percentage, or face offs, or hits, or blocked shots.

Some people try and draw conclusions. Such as team x has a higher Corsi so they must be the better team. Or the PDO is high or low so the team must be lucky or unlucky. Sometimes they are wrong. It doesn't make the data bad.

I get that their is too much value out on Corsi in the media and even by some stat sights. Heck, quality of competition and similar metrics use Corsi as the underlying metric. Some groups use Corsi like on base percentage was used on moneyball. Corsi = success isn't proven and may be a flawed assumption.

But none of that makes the data irrelevant. Before all of this debate there were a number of examples on how advanced stats can be used. In fact, the post itself is beyond Corsi and acknowledges some of its limitations.

But despite that, someone posts an article about some dufus using Corsi to draw an unsupported conclusion so we are having the same debate.

I say it AGAIN. Stats provide useful information. Some of us consider it interesting information. None of us are saying the Oilers are good or the Flames are bad. Just because there are stupid people using good information in bad ways isn't a reason to derail every stats based discussion.

 

I have to agree with Kehatch here.  Information is just that: information.  Take it or leave it as you will, and ridiculing the information serves no purpose.  It's almost like making fun of a brick just because it's a brick (regardless of the purpose it does or does not serve for you).

 

I've said it before, that I personally don't care for this kind of information, and I'll fully admit partly because I have not seen a practical application for it yet.  The fact that some find it useful and some find it interesting is great.  There are any people out there that find a particular method of basket-weaving useful and/or interesting.  I don't, but that doesn't lessen it's attraction to those who do.

 

I make fun of the people who use it in conversations with me, but the teasing (at least on my behalf) is always intended in a light-hearted, non-derogatory manner (although I know it can be taken as flippant and dismissive).

 

Having said this, I can absolutely understand Kehatch's frustration when the supposed "know it all" talking heads start spouting these new stats like they're gospel, but use them in a completely ignorant fashion (ie: the Edmonton doofus who thinks the Flames are "just lucky").

[KACaribou]

The thing that I find frustrating is that fans go out of their way to bash it without any understanding of what they are. If your not interested that's fine. If you don't agree with what someone is using them to say that's fine. But every time someone posts anything to do with advanced stats the entire discussion is derailed.

It's the stereotypical arguments too. The Oiler's hired an analytics person so advanced stats must be bad. Someone mentioned CORSI in an attempt to say the Oilers aren't bad so advanced stats must be bad. The data is telling me something I don't want to hear so advanced stats must be bad.

Stats guys can't even debate back because the entire argument is based around the NA NA NA NA BOO BOO I CAN'T HEAR YOU defence.

I had originally made an effort to clear this single thread so stats people had somewhere to discuss the stats rather then hear about why people are passionately ignorant about them. But apparently that's what this post is for again. Oh well.

The point was missed here. A TSN analytics guy said the Flames are a bad team, just lucky. You can throw all the stats you want, and I can throw them right back. But I watch every single game. I have seen bad Flames teams. This is a good Flames team, just getting better. Sometimes analytics people have to understand that the formula is bad if the results don't fit into it. In this case a different formula may prove to be better. I am not saying there can't be an anyalytics formula that works, but so for it doesn't. If 2 + 2 is 4 most of the time but not all the time, it would be a flawed formula. That's all. Keep trying to come up with something. Let me know when the formula works all the time. Then I will be more than interested. I will be sold on it.

[flamesfan85]

Let me know when the formula works all the time. Then I will be more than interested. I will be sold on it.

 

Really? Beyond fundamentally understood and accepted natural and mathematical laws I do not think there are any formulas that work "all of the time" (and even then most models break down under certain conditions). As Kehatch and others have tried to explain to the hard nosed detractors here - advanced stats are just a stat. That is it. They are not a gospel nor should they ever be treated as such.

 

If you want other real world corollaries to illustrate this idea... consider economic forecasts. They use models to attempt to illustrate and explain uncertain events/actions (agency of people within the economy) and the product of those events/actions. Have they ever worked all of the time? No. Will they ever work all of the time? No. Are they useful metrics for informing decisions and strategy? Absolutely.

 

That is all advanced stats offer and all they will offer. Some uninformed pundit decrying them as the holy book and stating that because of them the Flames success is due to luck is just a failure in your value assignment. That mans opinion doesn't hold value - he doesn't understand the interpretation of the statistics he is championing. Your holding him up as some kind of authoritative reason to detract from the value of advanced stats only serves to highlight your lack of understanding of them.

[DirtyDeeds]

Really? Beyond fundamentally understood and accepted natural and mathematical laws I do not think there are any formulas that work "all of the time" (and even then most models break down under certain conditions). As Kehatch and others have tried to explain to the hard nosed detractors here - advanced stats are just a stat. That is it. They are not a gospel nor should they ever be treated as such.

 

If you want other real world corollaries to illustrate this idea... consider economic forecasts. They use models to attempt to illustrate and explain uncertain events/actions (agency of people within the economy) and the product of those events/actions. Have they ever worked all of the time? No. Will they ever work all of the time? No. Are they useful metrics for informing decisions and strategy? Absolutely.

 

That is all advanced stats offer and all they will offer. Some uninformed pundit decrying them as the holy book and stating that because of them the Flames success is due to luck is just a failure in your value assignment. That mans opinion doesn't hold value - he doesn't understand the interpretation of the statistics he is championing. Your holding him up as some kind of authoritative reason to detract from the value of advanced stats only serves to highlight your lack of understanding of them.

When a poster here uses this guy as an example of poor use of advanced stats and calls him out for his "Flames are just lucky" comments. You say you agree but in same breath it is this posters fault and he has the lack of understanding??

 

The guy who made the claims should be called on his lack of understanding not the guy who heard him and reasoned he was incorrect...  

 

if he did not understand the advanced stats well enough, he would not have recognized that the guy was incorrectly using them and saying the Flames were lucky......

[DirtyDeeds]

----------------------------------------------------

There is little doubt in my mind that advanced stats are a whole lot of hype right now. The press is giving them all sorts of attention because a few hockey teams are finally adding some education about it to their systems.

The press has been negligent about advanced stats too and are attempting to compensate now.

All this hype and press is just that. Advanced stats won't change the face of the game, and they won't give a team the edge to be that much better than the next team. It is just a bit more data to consider is all.

[flamesfan85]

Apparently I misunderstood his post. I took it as he saw the TSN guy as incorrect, and then says that until they can use stats that are "always right" he will continue to not acknowledge them. Feels like you are inferring an understanding based on the post that I am not. Apologies if I was incorrect.