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

Thoughts on the Dolphins/Titans Game

Discussion in 'Miami Dolphins Forum' started by KeyFin, Sep 9, 2018.

  1. fin13

    fin13 Season Ticket Holder Club Member

    1,695
    1,237
    113
    May 29, 2009
    Waterloo
    wtf
     
    Pauly, resnor and Redwine4all like this.
  2. Irishman

    Irishman Well-Known Member

    573
    532
    93
    Oct 16, 2017
    High Point, NC

    Prescott is a young QB on a horribly coached team. Doesn't this equate to Tannehills' first 4 or 5 years here?
     
    Tin Indian, Fin D, danmarino and 3 others like this.
  3. Electric Boogaloo

    Electric Boogaloo Inventor of the question mark

    68
    80
    18
    May 11, 2013
    Miami
    Good write up
    Thought Tannehill looked sharp when given time. Clearly duffed a few though.
    Running game, WR group and secondary all looked encouraging. Even thought our DT rotation held their own and was better than I expected.

    My biggest frustration is the offensive scheme lacks any intermediate passing game. Gase, Lazor...doesn’t matter, everything is 5-10 yd passes, even on 3rd and 10-12 we throw short of the sticks by 5 yds. Every big time Qb with success, save Brady, ...Rivers, Rodgers, Ben, Brees, Ryan, etc picks up chunk plays on 15-20 yd intermediate throws. I do see that Gase is stacking this offense with YAC athletes, but it takes a 12 play drive to get a score. Too much room for error, imo.

    Also concerning that last years draft picks are looking very average. I’ll give Raekwon a pass on a one game sample size, but he’s looked a step slow to the sidelines, IMO. And Harris and Tankersley are non factors. No bueno. Good teams build their core via solid drafts in the first 4 rounds. Outside of Howard, Minkah and Drake, name a solid to great starter in our last 4 drafts. Maybe Ajayi, who we shipped out.

    Shouldn’t complain about a win at home, though. Got to hold serve against an inferior opponent with out their three best offensive players and we did what was needed to get the W. Fins up
     
    danmarino and KeyFin like this.
  4. Redwine4all

    Redwine4all Well-Known Member

    1,216
    686
    113
    Feb 4, 2016
    Yes. By year 7, hes out of excuses.
     
  5. Redwine4all

    Redwine4all Well-Known Member

    1,216
    686
    113
    Feb 4, 2016
    Tannehill plays way worse than his QBR, was my point.
     
  6. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    That's literally not possible.
     
    danmarino, Fin D, Redwine4all and 2 others like this.
  7. Slk018

    Slk018 Member

    29
    31
    13
    Aug 21, 2018
    Im not sure why everyone is carried away in awe by that. It was a standard deep post and he made a good throw. That is a throw every single NFL QB has the ability to make. He put that one on the money (and it was tipped and Stills held on). Many are acting as if its the throw he made to stills against the Chargers 2 years ago. THAT was athrow were you say "How many other guys can make that throw"
     
  8. Slk018

    Slk018 Member

    29
    31
    13
    Aug 21, 2018
    You are probably right in that we lose that game with Osweiler or Fales. But, you cant be sure. If they don't throw 2 picks and we get 3 picks and s special teams score and good running game we may still win.
    But, Osweiler and Fales are horrible QBs.
    Im encouraged with the offensive line. One play that bothered me the most besides the 3rd and 8 screen for 6 was the Wilson first down on 3rd and long.
    The defense left a guy uncovered to the left he had a clean pocket and threw the short corsser, this is my main gripe with RT. Wilson made a great play and got the first down. I would like to see him attack more and yesterday there was a lot there, lots of plays left on the field.
    But, it was his first game in almost 2 years, he was accurate and moved well for the most part. I think it will open up going forward.
     
    danmarino and KeyFin like this.
  9. Slk018

    Slk018 Member

    29
    31
    13
    Aug 21, 2018
    Sorry for all the replies. I am learning this forum so not sure how to respond to multiple.
    I wouldnt mind the call on 3rd down. I would rather try to pound it in with Gore on 2nd and goal. Gesicki can't get manhandled by corners off the line. Lets chalk it up to wet field and he slipped.
     
    danmarino and Dol-Fan Dupree like this.
  10. Dol-Fan Dupree

    Dol-Fan Dupree Tank? Who is Tank? I am Guy Incognito.

    40,533
    33,035
    113
    Dec 11, 2007
    Fair enough
     
    danmarino likes this.
  11. Slk018

    Slk018 Member

    29
    31
    13
    Aug 21, 2018
    Dak Prescott is NOT a good QB. they were able to hide him with an elite Oline and running game. Tannehill is better than Dak.
     
    Tin Indian and danmarino like this.
  12. KeyFin

    KeyFin Well-Known Member

    10,488
    12,821
    113
    Nov 1, 2009
    I think it will too, especially since RT is more of a steaky QB that needs to build up a rhythm. He didn't get that this past Sunday due to the delays, so it was almost like we played 3 different 1st quarters. Under the circumstances I was pretty impressed with how he played. There's definitely some stuff to clean up though.
     
  13. Redwine4all

    Redwine4all Well-Known Member

    1,216
    686
    113
    Feb 4, 2016
    Yes, it really is. An INT like his second counts the same as the one in the end zone. The second one didnt matter. The first one was a disaster.
     
  14. Carmen Cygni

    Carmen Cygni Well-Known Member

    2,422
    5,732
    113
    Dec 30, 2017
    Skinny post. Stills ran outside-in on the numbers per leverage and stemmed inside the hash.
     
  15. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    The only way you can argue someone played below their rating, imo, is if the receivers are getting massive gains off of 5 yard dumpoffs.
     
  16. Redwine4all

    Redwine4all Well-Known Member

    1,216
    686
    113
    Feb 4, 2016
    no.
     
  17. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Yes. Passer rating is simply a number equivalent of how they played. With the exception of what I've already stated, how is someone playing worse than their rating? It's not possible. It's simply something people try to assert a negative narrative about Tannehill.

    It's like saying a team is worse or better than their record. No, you are what your record says you are.
     
    djphinfan likes this.
  18. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    I mean.. both of you are right but the context is different. Yes, every stat simply records what happened so you can't do better or worse than that stat. But it's also true that no stat perfectly captures how well a player played because it doesn't include everything that matters (e.g., QB runs in passer rating) and doesn't disinclude everything that doesn't matter (the examples both of you gave).

    The real hard part of course is "disincluding everything that doesn't matter". That issue brings us back to the "division of credit" approach that for example ESPN's QBR and Football Outsider's DVOA incorporate, and there's simply no statistical approach to solving that problem. And of course there are tons of examples that demonstrate the problem, including what you pointed out but also stuff like an INT occurring because the WR let a perfectly thrown ball bounce off his hands, or a TD occurring because a ball bounced off a defender into a WR's hands, etc... In each case you could argue the rating isn't capturing how well the player played.

    I think the best way to resolve this issue in practice (i.e., given the limitations and strengths of stats) is to not rely too much on stats when sample size is small, but to rely much more heavily on them when sample size is large. When sample size is large, the correlation to win% is a quantifiable measure of how good that stat is, and it's actually very difficult for any human observer to match that. Humans may have tons more information to make their decision on, but they don't use that information well and are biased. Stats use sparse information, but they use it near optimally.

    So I'm much more inclined to argue that stats like passer rating truly capture how well a QB played if the sample size is very large (especially over a career), even given all its limitations (so not considering the technicality you're pointing out).
     
    resnor likes this.
  19. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Right. But, if a QB is say a 90 rated QB over a 7 year span, would you find it correct to say he's not as good as his rating says he is?

    I wouldn't.
     
    danmarino and cbrad like this.
  20. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    You can apply something called the Central Limit Theorem in statistics to get some intuition on how the confidence in passer rating over N games changes as you increase sample size N. Basically to have K times more confidence you need N = K^2 sample size.

    For example, you if you want your confidence in a QB's passer rating to be K = 3 times as large (i.e., the standard deviation of those estimates is 3 times as small), then you need N = K^2 = 3^2 = 9 times the sample size. So the confidence in passer rating over say 100 games is twice as large as over 10 games (because 10^2 = 100).

    Or if you chose 7 years vs. 1 year the difference is 2.64 times greater confidence with 7 years. But to answer your question of course 7 years is large enough for almost every application (exception would be if you're comparing QB's so similar that you just happen to need larger N).

    I'd start that comparison only from about 150+ passing attempts however. Anything less than that and the variance is so large you really can't reject any hypothesis lol (almost always you'll be able to say that the difference between Aaron Rodgers and crappy QB X over 5 games could still theoretically be explained by random variation alone lol).

    EDIT: Just noticed a brain fart with one sentence up there.. that 10^2 lol. That should be square root of 10 = 3.16 times confidence with N=100 compared to N=10. To have confidence twice as large you'd need N = 10*(2^2) = 40. Sorry about that!
     
    Last edited: Sep 13, 2018
    danmarino and resnor like this.
  21. Pauly

    Pauly Season Ticket Holder

    3,696
    3,743
    113
    Nov 29, 2007
    From a statistical perspective points scored -v- points allowed is a better predictor of future performance than the W -v- L record.
     
    Redwine4all likes this.
  22. Redwine4all

    Redwine4all Well-Known Member

    1,216
    686
    113
    Feb 4, 2016
    QBR doesn't tell the whole story. Proof...Rodgers QBR last year.
     
    KeyFin likes this.
  23. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Yeah, I wasn't saying that rating was useful for predicting anything. I was just saying that you are what your rating says you are. Just like you aren't better or worse than your record. You are what your record is.
     
  24. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Are you referring to the BS ESPN rating, or traditional rating?
     
  25. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Wait that's not true, except when you use 50+ games back or more. Here's the same graph as in post #62 of that "2019" thread except that I added point differential for comparison (and made the colors more "Dolphin" like lol):
    [​IMG]

    As you can see the best predictive power is actually using win/loss record over last N=10 games. However, if you're forced to use N>50 games back, then point differential becomes the better predictor.

    (oh.. I mislabeled the y-axis.. it should be "probability correct" since you're trying to predict the outcome)
     
    Irishman likes this.
  26. Pauly

    Pauly Season Ticket Holder

    3,696
    3,743
    113
    Nov 29, 2007
    Thanks cbrad

    I didn’t realize that the points scored/allowed being better than W/L is a shibbeloth. It’s a proposition that I’ve seen asserted on many stat sites (noticeably football outsiders) that I assumed that someone had done the base research.
     
    cbrad and Carmen Cygni like this.
  27. Pauly

    Pauly Season Ticket Holder

    3,696
    3,743
    113
    Nov 29, 2007
    The proper name for the traditional rating is. “Passer Rating”. It’s one of the things that annoys me about ESPN’s QBR is that they took a common misnomer/nickname of Passer Rating which makes it confusing when you’re discussing the 2 different beasts.
     
  28. Carmen Cygni

    Carmen Cygni Well-Known Member

    2,422
    5,732
    113
    Dec 30, 2017
    Thanks for my WOTD. :knucks:
     
    Pauly, KeyFin and resnor like this.
  29. Redwine4all

    Redwine4all Well-Known Member

    1,216
    686
    113
    Feb 4, 2016
    According to your logic, all you have to do to know who the best quarterbacks are is go by quarterback rating. I think that is ridiculously inaccurate. Not all passes are the same not all interceptions are the same not all completions are the same.
     
  30. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    No... I'm simply disputing your assertion that somehow, someone can play worse than their rating. With the exception of a QB throwing 5 yard dump offs that go for big gains consistently, I can't think of another way that's accurate.

    You have yet to give one shred of evidence for why you believe this. You don't like Tannehill, so you get around his good ratings by claiming he plays worse than his rating.
     
  31. Irishman

    Irishman Well-Known Member

    573
    532
    93
    Oct 16, 2017
    High Point, NC
    Does this also indicate the team has gotten more predictable, or that the method is better just using more recent games, or both?
     
  32. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    I think those curves are a composite of at least 2 different curves, which we could call curves A and B.

    Curve A dominates from maybe 25+ games and represent greater predictive power when you use more recent games. Curve A is basically a line, and if we extrapolate that line it intersects the y-axis at about 61%.

    Curve B dominates from N=1 to N=10 and is basically the effect of small sample size, and until N gets to be a decent size the effect of sample size will dominate. Put those two curves together and I think it explains that graph.

    However, the reason I cut the graph off at N=100 is because it suddenly levels off from around that point and is almost unchanged out to N=200. So really there are 3 curves: A, B and C. Curve C is a baseline curve at about 55% (it just stays there regardless of N) and most likely represents the cumulative effect of any source of stability in a franchise, from the GM, the coach, etc... Average turnover in the NFL at those positions isn't high enough to cause the overall graph to go to 50% at N=200, or 12.5 years. I suspect if you went to N=800 (50 years) it would go to 50%, but you'd need 100 years of NFL history to look at past 50 years because of games 50 years ago, so I can't prove that.
     
    Last edited: Sep 13, 2018
    Irishman likes this.
  33. Irishman

    Irishman Well-Known Member

    573
    532
    93
    Oct 16, 2017
    High Point, NC
    Thanks for your reply. It makes sense.
     
    cbrad likes this.

Share This Page