Closing the Book on 2017-18

[Bingo]

Originally Posted by nik-

But I have thin skin?

Individual stats being accurate can co-exist with those stats being of low value to results. That's basically what this comes down to.

Gospel joke vs questioning someone's integrity. I think you know the difference. Maybe you don't.

85% correlated. If you want to believe it's a coin toss go for it.

How's leaving this thread working out for you so far?

[rubecube]

Originally Posted by Bingo

Come on, you know the difference.

A team can carry the play but get let down by not converting on their chances or having their goaltender out played by the opponent's goaltender.

Your definition of "carrying the play" seems to be based entirely on the amount of shots they took. I don't think anyone who watched the Flames last year thought they were consistently in control of anything. And it is cherry-picking because you're only using the stats that support your argument and ignoring the stats that demonstrate that they were a bad team.

[the2bears]

Wow, stats really trigger some people.

Stats themselves aren't wrong, unless they're actually measured incorrectly. In my day job, if the software system I'm responsible for isn't performing well, I want to look at all the numbers. From there, I try to build a story that explains things. If that turns out wrong, I look at the stats again, collecting more if appropriate, and try to explain again.

But if we're measuring correctly, the stats themselves aren't "wrong".

I see this thread's revival as Bingo saying, "Here is a more complete set of stats from last year. Some interesting numbers there." Instead of a "Hmmm... they still don't match my eye-test, that makes me wonder why?" the response was a bit over the top by some. Maybe they were hurt by an Actuary in the past, I don't know.

[DJones]

Can't argue the stats, the only thing that bothers me with stats guys is the assumption that luck is responsible for a lot of the variances. In the NBA, no one assumes people are going to shoot league average. Instead they look at the shooting percent and use that as a skill indicator.

PDO in general seems stupid to me. Two unrelated metrics added together like it means something and compared to other teams.

We were a decent 5 on 5 team last year that sucked at shooting and decision making. If anything I think we were lucky to have scored so much the first half of the year.

[rubecube]

Originally Posted by the2bears

Wow, stats really trigger some people.

Stats themselves aren't wrong, unless they're actually measured incorrectly. In my day job, if the software system I'm responsible for isn't performing well, I want to look at all the numbers. From there, I try to build a story that explains things. If that turns out wrong, I look at the stats again, collecting more if appropriate, and try to explain again.

But if we're measuring correctly, the stats themselves aren't "wrong".

I see this thread's revival as Bingo saying, "Here is a more complete set of stats from last year. Some interesting numbers there." Instead of a "Hmmm... they still don't match my eye-test, that makes me wonder why?" the response was a bit over the top by some. Maybe they were hurt by an Actuary in the past, I don't know.

It's not the stats that trigger people. It's that he's using them to build a demonstrably false narrative about the team (unlucky vs. bad). If you can't stop teams from scoring on you more than you score on them, to the tune of a -30 GD, you're not a good team. Your time of possession, SOG, etc., are completely irrelevant.

[Bingo]

Originally Posted by rubecube

Your definition of "carrying the play" seems to be based entirely on the amount of shots they took. I don't think anyone who watched the Flames last year thought they were consistently in control of anything. And it is cherry-picking because you're only using the stats that support your argument and ignoring the stats that demonstrate that they were a bad team.

The whole topic is about the underlying stats from last season to which I presented them all.

That's not cherry picking.

I agree their powerplay sucked
They sucked at home
Smith sucked at home big time
They led the league in shooting wide

If you want to discuss those things I'm happy to. But I started a topic looking at the shot metrics for last season and didn't avoid any of them.

[Bingo]

Originally Posted by the2bears

Wow, stats really trigger some people.

Stats themselves aren't wrong, unless they're actually measured incorrectly. In my day job, if the software system I'm responsible for isn't performing well, I want to look at all the numbers. From there, I try to build a story that explains things. If that turns out wrong, I look at the stats again, collecting more if appropriate, and try to explain again.

But if we're measuring correctly, the stats themselves aren't "wrong".

I see this thread's revival as Bingo saying, "Here is a more complete set of stats from last year. Some interesting numbers there." Instead of a "Hmmm... they still don't match my eye-test, that makes me wonder why?" the response was a bit over the top by some. Maybe they were hurt by an Actuary in the past, I don't know.

Very well said.

I don't think it's the stats that trigger all of them though, I think it's anything that doesn't fit the "Gulutzan was a terrible coach" script.

[Bingo]

Originally Posted by rubecube

It's not the stats that trigger people. It's that he's using them to build a demonstrably false narrative about the team (unlucky vs. bad). If you can't stop teams from scoring on you more than you score on them, to the tune of a -30 GD, you're not a good team. Your time of possession, SOG, etc., are completely irrelevant.

Do you mean "false" or do you mean the narrative that you choose to believe?

[Locke]

Originally Posted by Bingo

Do you mean "false" or do you mean the narrative that you choose to believe?

They were a good team that under-achieved due to bad luck.

[rubecube]

Originally Posted by Bingo

Do you mean "false" or do you mean the narrative that you choose to believe?

No, it's false. Apply that same logic to any other aspect of life and tell me you'd have the same response. If you had an employee that was good in 1 or 2 aspects of his job and crappy in two others, would you call him an "unlucky" employee or a bad one? If you had a car that, despite having a working engine and brakes, kept getting into accidents due to malfunctioning headlights, would you consider it a piece of crap or unlucky?

I can't think of a single aspect of life that is results-based where performance isn't judged on said results.

[Bingo]

Originally Posted by Locke

They were a good team that under-achieved due to bad luck.

I've talked about shooting wide and an inability to finish. I responded to Flash's notion in the affirmative that Treliving thought it was a finishing issue and that's why he changed the roster.

And they had a lot go wrong. Some luck, some injuries.

But to suggest I'm only suggesting luck isn't accurate.

[Bingo]

Originally Posted by rubecube

No, it's false. Apply that same logic to any other aspect of life and tell me you'd have the same response. If you had an employee that was good in 1 or 2 aspects of his job and crappy in two others, would you call him an "unlucky" employee or a bad one? If you had a car that, despite having a working engine and brakes, kept getting into accidents due to malfunctioning headlights, would you consider it a piece of crap or unlucky?

I can't think of a single aspect of life that is results-based where performance isn't judged on said results.

Never said it was all luck.

[Table 5]

edit, whoops looks like I missed the difficulty rating in the opening thread! So never mind.

[Bingo]

Originally Posted by Table 5

Curious, is there any data to show where the shots came from?

One could assume that the farther from the net you are, the higher your chances of missing the net would be...perhaps the reason they missed so many is that they were shooting from places that had little chance of making a mark... which could feed into the narrative of the Flames being a perimeter team.

For sure.

That's why the addition of high, medium and low danger shot attempts and shots (made it through) is new information.

When you have 65 shot attempts with no break down of where they came you may have a team that is a perimeter squad that never generates anything.

But when you have a team that is off the charts in high and medium chances, but not as high in low, they're actually shooting from dangerous areas more often than most NHL teams.

This graphic pretty much shows the breakdown. Lots of heat in the middle in front of the net.



Compared to what they gave up.



I'm honestly not making this stuff up guys.

[Locke]

Originally Posted by Bingo

What stat is wrong then? Would love to know.

The standings? 84 Points.

Oh. Sorry. That stat is actually correct.

[Lanny_McDonald]

Bingo, I always appreciate your efforts in trying to find new and creative ways in presenting information to inform your reader's understanding of the game. Great job as always.



I think what that this does kind of closes out the book on advanced stats though. Clearly all of those stats mean nothing. The Flames are way behind last year's totals and trends, and are clearly a much better team. If the stats you're presenting are supposed to prove your case that the Flames are better than they were compared to outcomes, and the stats prove it, then this years outcomes and the underlying stats are proving just the opposite. The stats prove the opposite of your thesis.



It is no secret that I have always been highly skeptical of these faux stats, based on their erroneous underlying relationships which are to indicate causality, and the sheer lack of consistency in data collection from building to building across the league. I think that the consistency in the stats, the poor outcomes they support, and the absolute failure to provide consistent results in predictability and outcomes, indicates they are useless measures.



You made a comment in one of the game threads that I found extremely interesting. When discussing hits, you stated you didn't believe the data was reliable in any shape or fashion, and as a result didn't think the measure provided any value. The validity of the stat was questioned because of the inconsistency in the data collection across the league. After witnessing the Nashville game, and the creative accounting used by their off ice officials to calculate SOG, I would say any statistic relying on SOG as a measure is questionable.


My question to you in this. If one collected stat is questionable, then aren't all stats collected questionable? Does this not cause the house of cards to collapse?

[The Fonz]

These stats are inaccurate because they only consider half of the equation. It's not enough to just get shots from high danger areas... shots alone from those spots are not dangerous.

It is shots from those areas combined with your opponent being out of position, which if you play a slow game (Gulutzan), is very rarely.

You'd be crazy not to see how much quicker the Flames are playing this season, and so when they get shots off from these high danger areas (or any area for that matter), there is a much higher likelihood of it going in the net as the other team is often scrambling as a result of our speed.

There's a massive flaw in Gulutzan's coaching for corsi and shots from high danger areas, because he's not taking the other team's positioning and awareness into account.

[Bingo]

Originally Posted by Locke

The standings? 84 Points.

Oh. Sorry. That stat is actually correct.

So you don't really have an answer then? Got it.

[Locke]

Originally Posted by Bingo

So you don't really have an answer then? Got it.

I'm sorry....is there a statistic that counts for more?

I was under the impression that the points in the standings determine eligibility for the playoffs. Is that incorrect?

[Bingo]

Originally Posted by New Era

Bingo, I always appreciate your efforts in trying to find new and creative ways in presenting information to inform your reader's understanding of the game. Great job as always.



I think what that this does kind of closes out the book on advanced stats though. Clearly all of those stats mean nothing. The Flames are way behind last year's totals and trends, and are clearly a much better team. If the stats you're presenting are supposed to prove your case that the Flames are better than they were compared to outcomes, and the stats prove it, then this years outcomes and the underlying stats are proving just the opposite. The stats prove the opposite of your thesis.



It is no secret that I have always been highly skeptical of these faux stats, based on their erroneous underlying relationships which are to indicate causality, and the sheer lack of consistency in data collection from building to building across the league. I think that the consistency in the stats, the poor outcomes they support, and the absolute failure to provide consistent results in predictability and outcomes, indicates they are useless measures.



You made a comment in one of the game threads that I found extremely interesting. When discussing hits, you stated you didn't believe the data was reliable in any shape or fashion, and as a result didn't think the measure provided any value. The validity of the stat was questioned because of the inconsistency in the data collection across the league. After witnessing the Nashville game, and the creative accounting used by their off ice officials to calculate SOG, I would say any statistic relying on SOG as a measure is questionable.


My question to you in this. If one collected stat is questionable, then aren't all stats collected questionable? Does this not cause the house of cards to collapse?

First off thanks, I appreciate it.

To answer your question hits, giveaways, shots on goal and takeaways have always been counted in buildings (at least it was). Underlying stuff has always been counted by independent people going over previous games to build data sets for each club.

So they are consistent in how they are counted across the league for the collector of the information. This has been proven out in that a lot of the data is similar through three different sets of eyes.

Add to that that a shot attempt is an easy thing to assess when watching. He either shots or he doesn't. Hits, blocks, giveaways and takeaways are more "is that a hit?" "No" so there is a much looser definition riding in it

And finally these stats aren't faux in any way. They are simply a count of events that aren't all that subjective.