Some insight into why Philbin was fired and Gase was hired.

[Spider-Dan]

Sorry, you'll have to do better than "straw man" after saying this:

I use the statistics that are most strongly correlated with winning, which gives them their validity.

No one needs advanced stats to read the W-L column.  The entire reason advanced stats exist is to tell you when W-L may be misleading you.

You said that you dislike subjective stats and prefer objective stats.  However, you then immediately turn around and say that subjective stats are great, but only if they correlate with wins... which misses the entire point of advanced stats in the first place.

[Rich]

The entire reason advanced stats exist is to tell you when W-L may be misleading you.

Well not just that. It is also to tell you why this pass rusher still had a great season despite only getting 4 sacks or why this wide receiver wasn't really that good despite 1200 yards receiving.

It adds color to the raw stats because many raw stats can be misleading. A perfect example is Tannehill's game against the Texans. He had a perfect rating, but mostly because his receivers turned short passes into huge plays.

[Spider-Dan]

Well, W-L and other obvious objective stats (as you cited: sacks, yards, etc.).

Nobody needs advanced stats to figure out how many TDs Tannehill threw, or what his W-L record was.  Advanced stats give you reason to think that the obvious stats like wins or INTs or yards may not be telling the whole story.

[Dolfanalyst]

You said that you dislike subjective stats and prefer objective stats.  However, you then immediately turn around and say that subjective stats are great, but only if they correlate with wins... which misses the entire point of advanced stats in the first place.

What if a set of subjective statistics and a set of objective statistics are both "advanced" (to use your terminology), but one of them correlates with winning more strongly than the other?

[Dolfanalyst]

It adds color to the raw stats because many raw stats can be misleading. A perfect example is Tannehill's game against the Texans. He had a perfect rating, but mostly because his receivers turned short passes into huge plays.

And we don't all need to watch that film to make that determination.  We need only one person to watch the film (assuming he's reliable) and tell us how far Tannehill's passes were thrown in the air, and how far his receivers ran with them.

[Spider-Dan]

What if a set of subjective statistics and a set of objective statistics are both "advanced" (to use your terminology), but one of them correlates with winning more strongly than the other?

Objective stats are objective and stand on their own merit.

Any stat (like, say, DVOA) that depends on an arbitrarily-designated weighting of the value of a play is necessarily subjective.  There's nothing wrong with subjective stats, but when you jump on a soapbox and declare that subjective stats (as a whole) are insufficiently rigorous, you cannot then turn around and promote other subjective stats that you need to build the foundation of your preferred conclusion.

[Rich]

And we don't all need to watch that film to make that determination.  We need only one person to watch the film (assuming he's reliable) and tell us how far Tannehill's passes were thrown in the air, and how far his receivers ran with them.

Several people can watch the film and still come to different conclusions.

One person can say that Tannehill had easy throws all day and therefore did very little.

Another person can say that the footwork and savvy to successfully run a screen play or the ability to throw a slant right where the receiver can catch and run without slowing down was a huge factor in those plays.

Like I said before, the beauty of it, no matter what stat or metric you throw at someone, is that the film still tells the ultimate story.

[Dolfanalyst]

Objective stats are objective and stand on their own merit.

Any stat (like, say, DVOA) that depends on an arbitrarily-designated weighting of the value of a play is necessarily subjective.  There's nothing wrong with subjective stats, but when you jump on a soapbox and declare that subjective stats (as a whole) are insufficiently rigorous, you cannot then turn around and promote other subjective stats that you need to build the foundation of your preferred conclusion.

You don't seem to be comprehending that I haven't done that, even though I've repeated it several times now.  I simply noted that PFF's subjective pass blocking grades -- when someone else brought them up -- don't correlate with measures of QB play.

And on top of that, how do you know what my perferred conclusion is?  Are you a mind-reader via the internet?

[Dolfanalyst]

Several people can watch the film and still come to different conclusions.

One person can say that Tannehill had easy throws all day and therefore did very little.

Another person can say that the footwork and savvy to successfully run a screen play or the ability to throw a slant right where the receiver can catch and run without slowing down was a huge factor in those plays.

Like I said before, the beauty of it, no matter what stat or metric you throw at someone, is that the film still tells the ultimate story.

Though according to you (above), it tells a story subject to various and wide-ranging interpretations.

If you line up 10 different bathroom scales in a row, step on each of them one at a time, and each tells you something different, what in fact is your weight as measured by those bathroom scales?

[Dolfanalyst]

Objective stats are objective and stand on their own merit.

Any stat (like, say, DVOA) that depends on an arbitrarily-designated weighting of the value of a play is necessarily subjective.  There's nothing wrong with subjective stats, but when you jump on a soapbox and declare that subjective stats (as a whole) are insufficiently rigorous, you cannot then turn around and promote other subjective stats that you need to build the foundation of your preferred conclusion.

Also, the weighting you mentioned (bolded above) isn't arbitrary.  Like I said, it's based on the probability of obtaining a first down based on the downs and yardages involved in the play.

Obviously a team wants to be more likely than not to obtain a first down.  A play that results in the team's being less likely than not to obtain a first down obviously isn't as valuable as one that results in its being more likely than not to obtain one.

That distinction in the value of two types of plays is hardly arbitrary.

[Spider-Dan]

Assigning -1 point (and not -2 or -0.5) to the value of a 3-yard loss (and not a 2-yard or 4-yard loss) is as arbitrary as any rationale for pass blocking grades.

[Rich]

Though according to you (above), it tells a story subject to various and wide-ranging interpretations.

If you line up 10 different bathroom scales in a row, step on each of them one at a time, and each tells you something different, what in fact is your weight as measured by those bathroom scales?

Again, another silly analogy. Can you find something better than a device that serves one purpose?

The point I am making in watching film is that this is what the advanced stats sites do. Watch film and evaluate. So at the end of the day, their results may not mesh with what someone else sees. There is no other way to do this.

Lastly, all of these sites are evaluating without knowing what the play call is.

[Dolfanalyst]

Assigning -1 point (and not -2 or -0.5) to the value of a 3-yard loss (and not a 2-yard or 4-yard loss) is as arbitrary as any rationale for pass blocking grades.

If the assignment of such points corresponds to the probability of obtaining a first down, then it's far less problematic (if at all) than a pass-blocking grade whose rationale is unknown and which may be inconsistently applied across players and teams, as well as across graders.  At least the assignment of points (i.e., DVOA) has 1) a sound theoretical rationale, and 2) is applied consistently, which in turn permits reliable comparisons among players and teams.

The major problem PFF will always have with its subjective grades is that it doesn't determine their reliability, and so their validity is unknown.  As soon as they indicate that several people produced similar grades over a large sample of players, teams, and plays (i.e., inter-rater reliability), they will have made a major stride.  But they would need to publish their methods and data in that area, so that people can evaluate their soundness.

But, as I've been arguing with Rich (above), they're unlikely to pull that off, since, as even he said, film lends itself to various and wide-ranging interpretations.  The PFF people are no less susceptible to that inter-rater unreliability than anyone else.  John Doe working for PFF may see something totally different than Joe Shmoe, also working for PFF.

[Rich]

PFF constantly reviews and tweaks its methodology and all of their analysis is "peer-reviewed" twice.

[Dolfanalyst]

Again, another silly analogy. Can you find something better than a device that serves one purpose?

The point I am making in watching film is that this is what the advanced stats sites do. Watch film and evaluate. So at the end of the day, their results may not mesh with what someone else sees. There is no other way to do this.

Lastly, all of these sites are evaluating without knowing what the play call is.

I don't think you realize how DVOA, for example, differs in that regard from PFF's subjective grades.

DVOA, at its core, is based on yardage.  There is no arguing that a player gained X number of yards on a play.  He gained that number of yards in the eyes of anyone who isn't delusional.  So the core of the evaluation system is indeed objective.  It consists of a unit of measurement (the yard) that everyone can agree on.

On the other hand, PFF's subjective grades are based on, what?  Do we know?  When Lamar Miller for example receives a grade of -1 on a play, that was based on what?  And what happens when the other guy at PFF thinks he got a +1 instead?  Which guy is right?

When we're using yards instead, everyone is right.  The guy gained X yards, in everyone's eyes.