1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Merge: Report: Joe Philbin wanted to replace Ryan Tannehill with Derek Carr

Discussion in 'Miami Dolphins Forum' started by muskrat21, Jan 10, 2016.

  1. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Again, I'm arguing against your claim that it's possible that if OL variance is minimal, "QBs are affected significantly by offensive lines is incorrect". Think about it. It doesn't matter whether there is ANY variance in OL play to determine if that statement is true or false. You just need to show QB's are significantly affected by OL whether either has ANY variance or not.

    And QB's are significantly affected by the OL: 60 vs. 95 ratings in pressure vs. no-pressure situations.

    OK, that resolves this issue. You originally stated it's possible "the variation in offensive line play is actually caused by the variation in QB play" which implied "all" and that's all I was arguing against.
     
  2. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    So, just so I'm clear, not only do all QBs face the same amount of pressure, and all olines play roughly the same, it's just the QBs that handle it differently?

    This is quite the hypothesis you've come up with.
     
    Fin D likes this.
  3. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    I think where we're getting lost here is in the fact that what I'm focusing on overall is the extent to which QB play -- in general -- is explained by offensive line play. For the finding that QB ratings decrease under pressure to support the contention that it is, there would have to be an interaction effect in that case (i.e., the offensive line's being a moderating variable between pressure and QB rating), and we don't know there is one.

    OK cool. ;)
     
  4. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    OK compare two situations: 1) You have no OL. 2) You have an average NFL OL.

    Which do you think will lead to more pressure on the QB? You ask whether the OL is a "moderating variable". How can you say it's NOT???

    To be technical, say the independent variable is defensive play attempting to pressure the QB and the dependent variable is QB play. The OL is by all accounts perfectly described as a moderating variable (changes the probability of QB pressure).
     
  5. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    No, I didn't miss the point. My post addresses exactly that.
     
  6. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    More precisely, it's that there is a weak correlation between pressure surrendered on quarterbacks and measures of quarterback play, in part because there is so much less variation in pressure than there is in quarterback play. The quarterback play, which varies a great deal, can't possibly be explained by the offensive line play, which doesn't.

    Consider the fact that in 2015, there was a mere -0.10 correlation between percentage of pressured pass dropbacks and QBR, and a mere -0.17 correlation between percentage of pressured pass dropbacks and quarterbacks' DVOA, on a team-by-team basis.

    Those figures mean that in 2015, 99% of the variation in QBR was associated with things other than the percentage of pressured pass dropbacks, and 97% of the variation in DVOA was associated with things other than the percentage of pressured pass dropbacks.

    To put it more simply, when you look at how QBs vary throughout the league in terms of things that are fairly strongly correlated with winning (such as QBR and DVOA), you have to go well beyond pressure to account for how they vary. Pressure explains almost none of that variation among QBs. That means other things are explaining that variation much more strongly.
     
  7. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    If there were no offensive line, sure, it would very likely be a moderating variable, but that isn't what we see in the league. What we see in the league, rather, is that every team has an offensive line, and they don't vary a great deal across teams in how much pressure they surrender on QBs, certainly nowhere near as much as the QBs vary in their own performances.

    It's a moderating variable in theory, based on the configuration you've outlined, but it isn't a moderating variable in reality if there isn't enough variation within it to make the effect of pressure vary significantly from QB to QB. See the correlations I posted above for evidence of that.
     
  8. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    I think I found out where we're differing here. Your dependent variable is pressure, whereas mine is QB play in general.
     
  9. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    No our differences are simple. I'm arguing against your claim that the OL doesn't significantly influence QB play if there's no OL variance. You keep focusing on variance. For some reason you're not seeing the obvious: variance is irrelevant in this entire debate.

    OL influences QB play tremendously and that can be determined without knowing a single thing about variance or correlations. Just look at two different pressure situations that are due to the OL and see the effect on QB play.

    Not sure I can explain it better man..
     
    resnor and Fin D like this.
  10. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    No, no it didn't. There were gabs where Tannehill and the offense played well in second halves, but they still lost, because the defense let up scores in the second half. People blamed Tannehill because the offense had the defense in the field Suu much in the first half they were tired.

    Yet the Seahawks didn't suffer the same fatigue, apparently.
     
  11. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    Then this is where we're having our problem. I'm looking at the effect of pressure on QBs league-wide (i.e., the post above, with the correlations in it), and you seem content to stop at the sort of analysis above, which delves nowhere into whether there are differences in offensive line play across teams and how they correlate with quarterback play.

    Surely there are plays in which pressure on the QB can be attributed to the offensive line, like you stated above. I'm not denying that. However, the task now becomes determining whether there is 1) significant variation in those plays across teams, which would suggest that some offensive lines are significantly better than others, and 2) whether that variation, if it exists, correlates with quarterback play. You aren't doing either of those things by stating that plays like the ones in your quote simply exist in the league.
     
  12. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    You're asking for a statistical analysis to prove water is wet.

    I know you're in "ignore Fin D because he challenges me to the point my stance no longer makes sense" mode right now, but if you believe we need statistics to prove it before we can claim it, then you'd HAVE to believe that any 5 random people off the street would perform the same as 5 NFL lineman.

    If you don't believe 5 random people can do it, then you'd have to allow that we're talking about humans and they are all different. Some are better than others at a given task. That's one place you're screwing up.

    Another place you're screwing up is you are flat out misusing stats. Stats are a tool and the way you're using them is similar to hammering a nail with a screwdriver. Stats are used to indicate what is likely to happen not what will happen. They are predictive not guarantees. Even if a stat says 80% (which even the best QB correlations to winning are not that high) of the time when X happens Y will happen that still means 20% of the time it doesn't. Again predicative not guarantees.

    The third problem you have, is that you are not objective. You have a notion and set out to prove it, so you manipulate what data you use or even disqualify simple truths like "not all olines are the same", to keep your already held belief "statistically accurate".

    You have, over the years, consistently struggled with these concepts.
     
  13. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    I missed this earlier. This here gets at my point above as well, because if it happened in reality, then there would be significant variation in offensive lines in the league (with this one comprised of high-schoolers varying sharply downward, presumably), and the correlation between offensive line and QB play would become stronger, presumably because the quarterback on this team would be unable to function and unable to compile any respectable statistics of his own.

    But this isn't what we see in the league in reality. What we see in the league in reality, rather, is that there are a bunch of offensive lines that are much more interchangeable in nature, with no single one being that much better or worse than any other in terms of how they affect QB play (per the correlations I posted above).

    So there is no equivalent of the "high school" line in the NFL. That's part of the problem of placing one's team under such a high-powered microscope. You get to thinking your team functions in some unique way, though when you pan back and look at the league as a whole, you find your team isn't all that different. Then it becomes much harder to attribute, for example, Ryan Tannehill's functioning to his offensive line, when his offensive line isn't much different from the average one in the league, and the QBs who play better, on average, don't have significantly better offensive lines of their own.
     
  14. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    I'll converse with you to the extent that you remain polite and respectful.
     
  15. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    That's clearly not true as you ignored the rest of the post you quoted. Essentially, proving the part you did quote.
     
  16. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    I didn't ignore it. I read it. I figured it was addressed in post #415.
     
  17. roy_miami

    roy_miami Well-Known Member

    1,385
    560
    113
    Oct 11, 2013
    From that same article Manning had a passer rating of 122 vs no pressure to 90 with pressure, do these numbers also prove there is great variance in o-line play across the league?
     
    cbrad likes this.
  18. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    This gets exactly at my point that there would have to be significant variation among offensive lines for it to moderate the relationship between pressure, which happens to all teams, and QB play. What you'd have to show is that the pressure that happens to all teams results in poorer QB play for some (but not all) teams, via the impact of those teams' poorer offensive lines. Here's what that looks like graphically:

    http://www.palgrave-journals.com/jibs/journal/v45/n9/carousel/jibs201450f1.jpg
     
  19. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    I don't see how. There's more than just that one point in there, and even that one was expounded upon to the point it rendered 415 inaccurate.
     
  20. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    We'll have to agree to disagree, then.
     
  21. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    As I said, you proved my "joke" to be accurate.
     
  22. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    And we'll have to agree to disagree about that, as well.
     
  23. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    No we don't. You can keep ignoring my other points and thereby proving my accusation and I'll keep pointing it out.
     
  24. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    Is it okay with you if I simply don't agree with your other points, and state as much? You're certainly still free to believe those points -- I'm saying nothing about what you should or shouldn't believe, yourself.
     
  25. Shane Falco

    Shane Falco Banned

    916
    468
    0
    Nov 22, 2015
    Any guesses who this new dude is?
     
  26. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    lol, you can do whatever you want. But just know, this is why you get the grief you get from everyone.\

    Your argument doesn't change. After Thill's first year, you tried the argument that all receivers were basically the same. I eventually got you to admit the error of your ways, which you promptly forgot two days later. Here we are a few years passed and now your stance is that the olines are basically the same.
     
    Shane Falco likes this.
  27. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    That's merely your opinion. I strongly doubt that the vast majority of other people take issue with being asked politely and respectfully to agree to disagree about something, and in turn wish to give "grief" to the other person involved on the basis of it.
     
  28. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    You're right they don't. What they don't like, however, is your obvious passive aggressiveness and your refusal to answer points that show you to be wrong. This is a forum for discussion not a place for you to make declarations and to be unchallenged.

    Look, you already made a post wondering/complaining about how you're treated. I explained why. Accept it or don't.

    I get a lot of **** on this board too, but its not a mystery to me as to why.

    Again, you used to argue that all WRs were basically the same, now you've changed that to all Olines are basically the same. That's absurd and you do it because , for some reason, you've never liked Thill and you're willing to ignore common sense and logic to try and prove that bias statistically.
     
    resnor likes this.
  29. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    I disagree with everything you've written above, my disagreement with it won't change, and I don't wish to discuss it any further. Please respect that. I respect that you are free to believe whatever you'd like, and I don't wish to have any influence on your thinking or beliefs.
     
  30. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Dude, you can't post something that isn't correct, then say, "agree to disagree" when people dispute it.

    Water is dry. Gravity doesn't exist. The sky is green. Don't argue with me. These are what I believe. Let's just agree to disagree.

    Shane, yeah, I think most know who this is.
     
    Fin D likes this.
  31. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    If you believe it isn't correct, I believe it is, and our positions are unlikely to change, then a polite and respectful "agree to disagree" is certainly in order.

    I'm going to take a break from this thread until something substantive about the Miami Dolphins is raised for discussion. The topic of how people conduct themselves on a message board, which is what this has now evolved to, isn't interesting to me. Others are certainly free to continue to discuss it, however.
     
    Steve-Mo likes this.
  32. Sceeto

    Sceeto Well-Known Member

    13,501
    6,245
    113
    Oct 13, 2008
    New York
    C'mon guys, back to the Statistics 101 debate.

    :cry:
     
  33. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Yeah the question of how much of the variance in QB play is a result of variance in OL play is a completely separate issue from the points you made in post #324 that I attacked in post #338. The first point I attacked in #338 was precisely this implicit claim that you need to know something about variance in OL to determine if the OL has a huge influence on QB play. You need to know nothing about variance in OL to determine that.

    What you're asking is how much of the variance in OL play accounts for the variance in QB play. I never attacked anything on that issue, except as I'll point out again the input statistics can't be taken seriously.

    Maybe I should just summarize what I think about different stats used in football analysis. There are 3 types of stats I think that are OK to use:

    1) Simple descriptive stats like "yards", "Y/A", etc.. Importantly, NO subjective weights are included, so passer rating doesn't fall into this group.

    2) Stats that include subjective weights IF (and ONLY if) the formula is transparent. The main example here is passer rating but you could put AYA or so into it too. Very importantly, the justification for this class of stats has to be well understood. Because subjective weights are put on them, the ONLY justification I can accept for using them is correlation with win %. That is, you use them not because they are "natural" descriptors of what is going on, but because you can show high correlation with wins (predictive power).

    3) Win probability. This is the only type of actual statistical analysis (as opposed to just descriptive stats or attempts to predict things) that I have respect for. Show an argument that depends only on win probability and that's a good analysis assuming you have enough data to make the argument.

    DVOA in principle falls into class #2, but it really doesn't because it's not transparent. I know you're willing to overlook that, but I'm not. It's also why I dismiss ESPN's QBR. So to summarize, I was arguing specific points, not your overall points about how variance in one accounts for variance in another. I've already pointed out the reason I wouldn't take that kind of analysis too seriously: I don't respect the input stats, but as I said before that's not a knock on you per se, just the inputs to the analysis.
     
  34. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    That's what it turns into when you don't want to discuss what you post.

    And how can you say you believe it? You have been called on several things, which you then say are only hypotheticals. Like, you state something as a hypothetical, but when people disagree, you have your out, because it was "only a hypothetical."

    But clearly these are things you believe, not simply hypotheticals.
     
  35. roy_miami

    roy_miami Well-Known Member

    1,385
    560
    113
    Oct 11, 2013
    Yeah because your belief that Ryan Tannehill is an elite QB being made to look mediocre by the things around him is totally on par with saying the earth is round...
     
  36. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    OK that's clear now. Thanks for taking the time to break it all down like that.

    Just curious -- is there an offensive line statistic you believe has merit? If so I'd be happy to do the work of determining its correlation with YPA, since that seems to be a quarterback statistic you believe has merit. I'm always interested in doing the work, regardless of what it turns up.
     
    cbrad likes this.
  37. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    You ask a great question. The answer is not necessarily yes as your intuition might suggest, in fact for the toy argument I made the answer is no.

    It depends on how you measure OL play. In general, the best types of measurements are efficiency measurements (e.g. Y/A) because you exclude the effect of one player just having more opportunities than another, assuming both have enough opportunities so that sample size isn't a problem.

    Consider a toy measure of OL efficiency that just looks at "percent of defensive pressure nullified". So one OL might nullify 90% of the pressure, while another 85%, etc..

    The different numbers there give you "variance" in OL play by such a hypothetical measure. Now, suppose there is NO variance in OL play. So suppose across the league all OL nullify 90% of defensive pressure. Does that mean every play looks the same? No.

    In one case the defense may apply more pressure by design than another play. So nullifying 90% (with no variance) all the time still leads to situations where there is more pressure vs. less pressure. That's how you can have no variance in OL play yet get "pressure" vs. "no pressure" situations. Good question regardless.
     
  38. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Sadly no. The reason there shouldn't be (regardless of claims some people make) is because in order to have such a stat you need many different QB and OL combinations. If the composition of the OL never changes for a given QB, then no stat could ever tease apart their relative contributions. Obviously, OL personnel changes occur, but no one is going to convince me there is enough of that going on to get any decent OL stat.

    This is what I would do instead. Just use any stat where you can argue the OL must have played a very important role. Rushing yards (even efficiency), or obviously pressure/sack stats will work. We don't know precisely (or even approximately) how much is due to the OL, but you can argue there MUST be a good portion due to it. So if you show correlations between any of those stats and whatever you're interested in, you can at least make the argument that variance in OL play accounts for a non-trivial portion of the variance in something else.
     
  39. RevRick

    RevRick Long Haired Leaping Gnome Club Member

    7,191
    3,940
    113
    Nov 25, 2007
    Thomasville, GA
    Dude.... That is just NOT right. I have never seen a quarterback fend off a raging defensive tackle with his left hand and throw touchdown passes with his right - and that is the logical extension of the ridiculous statement quoted above.

    There have been a number of possibly good quarterbacks who are now sitting at home because their body was beaten down to a nub because their offensive line could not pass block at all - and the Dolphins are not very far from having this quarterback join that club.
     
    Sceeto, resnor and Fin D like this.
  40. Laurence

    Laurence Banned

    80
    19
    0
    Jan 20, 2016
    One would think, based on a superficial analysis (and I don't mean that to sound insulting to you). But what if the raging defensive tackle is in the backfield because his team is up by three scores in the fourth quarter, because the opposing quarterback has had trouble scoring points, the defensive tackle knows the other team has to pass the ball, he doesn't have to respect the run game, and so with that advantage, he can dominate his opposing offensive lineman much more easily than he would otherwise?

    Now, certainly that sort of situation doesn't explain every presence of a defensive tackle in an opposing backfield, but if it explains far more of them than does the variation in the quality of the offensive lines in the league, then we have a situation where the quarterback is indeed responsible for his own pressure.

    A deeper analysis is needed to investigate these possibilities.
     

Share This Page