The Advanced Stats Thread Episode III: Return of the Eye Test

[Filthy Dangles]

Because every goal is a shot, but not every goal is a scoring chance. When Manny posted his xG model online, he noted that it wasn't a better predictor for goals than shot attempts. However, the xG model built by DTMAboutHeart, he's stated that it is better than shot attempts.



http://www.corsica.hockey/blog/2016/03/03/shot-quality-and-expected-goals-part-i/

Also, because shot attempts are important

That's what I was missing. Got it.

Do you know what percentage of goals that are scored come from shots graded as scoring chances? League wide and for the Rangers specifically?

Also, what's your thoughts on guys like Weber and Josi etc. who are regarded as elite players but have bad advanced stats?

 

[gorangers0525]

Over the years, Weber's advanced stats have always been good when he hasn't been with Josi. Like McDonagh, he's been tied to a posession anchor and there is not much he could do about it. Pietrangelo is in a similar situation with Bouwmeester.

Josi on the other hand, destroys the posession stats of everyone he plays with, no matter the situation. Most overrated player in the league IMO.

 

[silverfish]

That's what I was missing. Got it.

Do you know what percentage of goals that are scored come from shots graded as scoring chances? League wide and for the Rangers specifically?

Also, what's your thoughts on guys like Weber and Josi etc. who are regarded as elite players but have bad advanced stats?

I don't :/ I believe one could accomplish this feat by grabbing all the raw data that Manny puts on Corsica and then conducting their own research *nudge nudge*

Don't really have much of an opinion on either Weber or Josi outside of just not thinking they are elite players

Rolling differentials for each Rangers d-man. This is u-g-l-y

[collapse=img]

[/collapse]

 

[chosen]

Ian doesn't know what he's talking about.

I can't remember if it was you or another stats guy who laughed when I called Columbus a good team a while back. I was also told to check the standings in January because it would all fall apart.

Maybe, but it's pointing in the opposite direction, at the moment.

 

[silverfish]

I can't remember if it was you or another stats guy who laughed when I called Columbus a good team a while back. I was also told to check the standings in January because it would all fall apart.

Maybe, but it's pointing in the opposite direction, at the moment.

Don't think any "stats" guy would've laughed at that. CBJ was good last year, but Bobrovsky was hot-garbage. Now CBJ is still good, and so is Bob.

 

[chosen]

Don't think any "stats" guy would've laughed at that. CBJ was good last year, but Bobrovsky was hot-garbage. Now CBJ is still good, and so is Bob.

You would be wrong.

I was told by one of the regular stats guys who posts regularly that their possession numbers were terrible and they would crash down to earth.

 

[silverfish]

You would be wrong.

I was told by one of the regular stats guys who posts regularly that their possession numbers were terrible and they would crash down to earth.

Find the post?

Although, their #s last year are worse than I thought they were... I hope it was me

 

[chosen]

Find the post?

Although, their #s last year are worse than I thought they were... I hope it was me

http://hfboards.mandatory.com/showpost.php?p=124647755&postcount=163

 

[silverfish]

http://hfboards.mandatory.com/showpost.php?p=124647755&postcount=163

Ah, so that's from this season. Well, to be fair, at that point in the year, they had the 5th worst CF% in the league, and since then, have had the league's best CF%.

Anyone know what happened in Columbus around 11/20 this year?

 

[Irishguy42]

Ah, so that's from this season. Well, to be fair, at that point in the year, they had the 5th worst CF% in the league, and since then, have had the league's best CF%.

Anyone know what happened in Columbus around 11/20 this year?

Torts' heart grew three sizes.

 

[chosen]

Ah, so that's from this season. Well, to be fair, at that point in the year, they had the 5th worst CF% in the league, and since then, have had the league's best CF%.

Anyone know what happened in Columbus around 11/20 this year?

So, in other words, advanced stats are useless as a predictor?

Anyone can say A sucks because the stats say A sucks. Then if the stats get better, it's okay to say that A is now good?

Isn't that kind of a nonsensical way of performing analysis?

 

[Mac n Gs]

So, in other words, advanced stats are useless as a predictor?

Anyone can say A sucks because the stats say A sucks. Then if the stats get better, it's okay to say that A is now good?

Isn't that kind of a nonsensical way of performing analysis?

I think the whole argument can be boiled down to small sample sizes and lack of context, but that's just my opinion.

 

[silverfish]

So, in other words, advanced stats are useless as a predictor?

Anyone can say A sucks because the stats say A sucks. Then if the stats get better, it's okay to say that A is now good?

Isn't that kind of a nonsensical way of performing analysis?

MH said CBJ is bad because they had an awful SA% and a high PDO. This is a classic recipe for an incoming downturn. The Rangers are in the midst of one right now, for the second season in a row.

There are a few factors at play here:

1) Did CBJ change anything on a team level?
2) Are they playing certain players more or less?
3) Did Torts change the game plan?
4) What's the strength of schedule?
Etc... Etc... Etc...

You're basically using one sample size of ten games compared to the next sample size of ten games, and since the SA% of one-team fluctuated so much, you're calling "advanced stats" a "useless predictor"

I can just as easily point to the Rangers here.

Through 11/20, the Rangers SA% was 47.11%
Since 11/20, the Rangers SA% is 42.76%

It was bad before, and it's still bad, and getting worse, because the PDO isn't as high. But I wouldn't point to this example as "advanced stats" being a good predictor of a bad team getting worse, because it's not a large sample size, it's one-team out of 30 in the league, and plenty of other reasons.

This article may be of interest in terms of predicting future performance:

https://hockey-graphs.com/2015/10/01/expected-goals-are-a-better-predictor-of-future-scoring-than-corsi-goals/

 

[chosen]

MH said CBJ is bad because they had an awful SA% and a high PDO. This is a classic recipe for an incoming downturn. The Rangers are in the midst of one right now, for the second season in a row.

There are a few factors at play here:

1) Did CBJ change anything on a team level?
2) Are they playing certain players more or less?
3) Did Torts change the game plan?
4) What's the strength of schedule?
Etc... Etc... Etc...

You're basically using one sample size of ten games compared to the next sample size of ten games, and since the SA% of one-team fluctuated so much, you're calling "advanced stats" a "useless predictor"

I can just as easily point to the Rangers here.

Through 11/20, the Rangers SA% was 47.11%
Since 11/20, the Rangers SA% is 42.76%

It was bad before, and it's still bad, and getting worse, because the PDO isn't as high. But I wouldn't point to this example as "advanced stats" being a good predictor of a bad team getting worse, because it's not a large sample size, it's one-team out of 30 in the league, and plenty of other reasons.

This article may be of interest in terms of predicting future performance:

https://hockey-graphs.com/2015/10/01/expected-goals-are-a-better-predictor-of-future-scoring-than-corsi-goals/

You're missing the point that it was Machinehead, an advanced stats advocate, not me, who drew upon the small sample size in order to declare that Columbus will crash and burn.

They still might, but I would argue that no team in the league is better than them at this point in time.

 

[silverfish]

You're missing the point that it was Machinehead, an advanced stats advocate, not me, who drew upon the small sample size in order to declare that Columbus will crash and burn.

They still might, but I would argue that no team in the league is better than them at this point in time.

Right, and using the evidence at the time, his prediction was not that far-fetched. Would it be better if we all didn't speak in absolutes on HF all the time? Of course. If MH simply said: "ya know, I think CBJ isn't good as they seem because their SA% is low and their PDO is high, I think they'll see a downturn soon".

Unfortunately, no one on HF (myself included) really posts like that

 

[chosen]

Right, and using the evidence at the time, his prediction was not that far-fetched. Would it be better if we all didn't speak in absolutes on HF all the time? Of course. If MH simply said: "ya know, I think CBJ isn't good as they seem because their SA% is low and their PDO is high, I think they'll see a downturn soon".

Unfortunately, no one on HF (myself included) really posts like that

.

I guess it was the ridicule of my opinion that Columbus was good, as absurd, made me think his statement was stupid and obnoxious.

We all post our opinions here, but when you stand behind advanced stats and laugh at opinions that differ from these facts, you are sending out the notion that those that disagree with you just don't understand what they are watching.

I understand that advanced stats are a nice tool, but sometimes it seems like some of its proponents treat it like religion in a way that belittles all other religions and non-religious beliefs.

 

[silverfish]

.

I guess it was the ridicule of my opinion that Columbus was good, as absurd, made me think his statement was stupid and obnoxious.

We all post our opinions here, but when you stand behind advanced stats and laugh at opinions that differ from these facts, you are sending out the notion that those that disagree with you just don't understand what they are watching.

I understand that advanced stats are a nice tool, but sometimes it seems like some of its proponents treat it like religion in a way that belittles all other religions and non-religious beliefs.

I'm not out here trying to belittle anything, but it is frustrating for the proponents of the use of advanced stats in analysis because these are metrics that are proven to be beneficial to analysis.

It seems like belittling when someone provides objective analysis to counter someone's subjective point of view because it's impossible for that person to see it as a counter-argument instead of someone ****ting on what they believe.

If I say something like: Girardi is great. He's been a warrior for the Rangers and I think he's been great for this team. If someone counters that by saying: "Girardi bleed shot attempts against, and it's more than likely that his play will continue to decline."

The 2nd person in this conversation is right. The first person is wrong. The first person doesn't want to believe that they are wrong because it's their opinion.

It's an impossible line to toe, and it's frustrating for the person using objective analysis when the person using completely subjective analysis is firing back.

There are attacks on both sides of the aisle. They only seem personal if we tie a personal affect to them.

None of this makes sense, I bet. Just the ramblings of a man with Monday brain

 

[Machinehead]

Columbus is now up to 12th in corsi and their PDO is coming towards 100.

They're legitimately improving instead of continuing to ride hot sticks.

 

[Mac n Gs]

I really wish you could apply score states to the rolling averages part of corsica.

I'd be really interested to see what our tied/close numbers have been for these past two weeks because even though we have the 3rd lowest TOI for tied hockey, the stats are still pretty good (besides CF%) xGF60 and xGA60 are both near the top.

 

[chosen]

Columbus is now up to 12th in corsi and their PDO is coming towards 100.

They're legitimately improving instead of continuing to ride hot sticks.

In other words they were a bad team when they were steamrolling the best teams in the league because their advanced stats were poor, but now that their advanced stats are prettier, they are a good team?

 

[silverfish]

Josh Manson is so ****ing good.

 

[Machinehead]

I can't remember if it was you or another stats guy who laughed when I called Columbus a good team a while back. I was also told to check the standings in January because it would all fall apart.

Maybe, but it's pointing in the opposite direction, at the moment.

Call the Daily News, Machinehead might have been wrong about something!

Wouldn't be the first time. I ate hair for ****s sake.

That doesn't mean this particular tweeter isn't also wrong in this case. We've been over it a hundred times. He's wrong.

 

[Machinehead]

In other words they were a bad team when they were steamrolling the best teams in the league because their advanced stats were poor, but now that their advanced stats are prettier, they are a good team?

In a way, yes, you might say that.

They were playing poorly but got lucky, and now they're playing well.

It's a long season.

The Penguins sucked at advanced stats last year through Christmas. The second half of the season that were the best corsi team by a comfortable gap.

 

[chosen]

In a way, yes, you might say that.

They were playing poorly but got lucky, and now they're playing well.

It's a long season.

The Penguins sucked at advanced stats last year through Christmas. The second half of the season that were the best corsi team by a comfortable gap.

Your response is the reason that I believe advanced stats is in its infancy and has yet to evolve into a more meaningful thing.

 

[Machinehead]

Your response is the reason that I believe advanced stats is in its infancy and has yet to evolve into a more meaningful thing.

So two weeks of play by the Blue Jackets holds more weight than a decade of top 5 CF% teams winning the Cup?