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

Take Away His 'X' Best Carries and He's Average

Discussion in 'Miami Dolphins Forum' started by DevilFin13, Apr 3, 2015.

  1. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    Not sure if you're being purposely obtuse or I'm not clear enough.

    What I'm getting at is, I want to see how this correlation number is calculated. What are the specific data points used? If on the days Lamar Miller runs for 0.5 ypc on carries 11 and 12 and Philbin shuts him down, but on the days he's tearing it up on the ground and Philbin gives him more carries, that doesn't mean he gets better when he carries the ball more. That means he gets more carries the better he runs.

    Seeing as he averaged 5.4 ypc on runs 1-10, yet 2.4ypc on runs 11-20 (before the 97 yarder), I'd trying to figure out how a formula would come to the conclusion that he gets better the more he runs.

    As those are two very conflicting pictures. That he can run for a 2.4 ypc on runs 11-20 (he never ran more than 18 times) yet you're saying his ypc is better when he runs more. That tells me coaches are limiting his chances when he's not running well hence your conclusion from the correlation is wrong.

    If his ypc got better on a per carry basis the more carries he got, yes I'd agree with that conclusion. But it's clearly not. If you are counting his over game ypc on games he got more carries then I'm calling negatory on that theory.

    He wasn't like this last year so it is either an anomaly (hence his answering he doesn't know when specifically asked about the issue) or it's the added weight. We'll see because he's even bigger this year.
     
  2. finsfandan

    finsfandan Well-Known Member

    2,547
    600
    113
    Dec 14, 2014
    I don't remember Miller being able to run between the tackles very well.

    Him bulking up this offseason tells me he acknowledges that he needs to be a better inside runner.

    He'll have to beef up if we don't improve our guards.
     
  3. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    He has a pretty narrow frame if I remember correctly (judging by looks, not any measurements). Putting weight on narrower guys isn't always a good idea.
     
  4. Piston Honda

    Piston Honda Well-Known Member

    7,853
    8,088
    113
    Sep 23, 2014
    No, the arguement is that he's most effective when getting 12-15 carries and that 12-15 should be the aim point going forward, thus the need for a second quality back to handle another 12-15 per game. I haven't seen anyone say he gets worse the more carries he gets, that's grossly over simplifying the issue. WADR I think you're better than that and perhaps just looking for debate if that's what you're arguing against.

    You want a better measure of RB performance? Watch the tape. It's not hard to see whether a back is running hard, seeking and or driving thru contact for additional yards, or taking a more passive approach. Looking at numbers on paper isn't gonna do it for ya.

    Imo the staff did a good job of managing his work load, especially w Moreno only playing one full game. Unfortunately for the team as a whole their inability to lean on the run game for a full 60 minutes resulted in several losses. That's what I believe will be remedied by a 2nd back. Unless you're claiming that Miller can handle it all on his own then I'm not sure what the debate is here.
     
  5. finsfandan

    finsfandan Well-Known Member

    2,547
    600
    113
    Dec 14, 2014
    Yeah, I prefer keeping Miller as the speedy back and drafting a power back but Miller should be more versatile. Defenses shouldn't only worry about outside runs.
     
  6. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    You seem to have lost sight of the original statement that I responded to here, which was that he "was less effective the more carries he got" (post #49, to which you gave a "fist bump").

    It sounds like you, too, disagree with that statement at this point? If so, then you and I are agreeing about that point.
     
  7. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    You need to really focus in on post #63 here. Again, if Miller's performance is such that later carries result in better YPC than earlier carries, the argument that he "gets worse the more carries he gets" hardly holds.

    Your distinction of "carries 1 through 10, versus carries 11 through 20" isn't fully informative, because again, it's entirely conceivable that carries 15 through 19 are better than carries 1 through 10, but aren't sufficient to alter the average of carries 11 through 20, perhaps because carries 11 through 14 are so bad.

    And if carries 15 through 19 are better, on average, than carries 1 through 10, then it can hardly be argued that he "gets worse the more carries he gets."
     
  8. Fin4Ever

    Fin4Ever Season Ticket Holder Club Member

    9,297
    2,738
    113
    Aug 26, 2014
    Vero Beach, FL
    I know that I don't forget that. All I did back then was to wish for a competent running game and a stronger defense. Unfortunately it was never more apparent than the Super Bowl and if we would have fired the DC AND drafted to our specific needs, I feel that Marino would have been in more title games and brought the Trophy home.Anyone remember the AFC Championship game in 85? The year we smoked the Bears. We would have had a rematch that I think we win because we matched up well against them but in the championship against NE Stratford had the huge fumble that cost us the game...grrrr..memories
     
  9. Fin4Ever

    Fin4Ever Season Ticket Holder Club Member

    9,297
    2,738
    113
    Aug 26, 2014
    Vero Beach, FL
    I am with you..that is why I would like to pair Todd Gurley with Lamar Miller...We would have the best tandem in the league and I am a fan of the RB's that UGA sends to the NFL. These guys always seem ready to hit the field running from Day1. We need a few more playmakers and Gurley #1 is a great start. We can trade down to 18 and pick our 3rd and 7th back up, draft Gurley, then in round 2 we could package one of our fifth rounders to move up and draft a WR.
     
  10. Larryfinfan

    Larryfinfan 17-0...Priceless Club Member

    I think the issue with our run game is that there were times when our OL and Miller couldn't produce when it had to. In my opinion, the OL was much more to blame than Miller or scheme or anything else, but the bottom line, unlike in Dallas' case (Murray and Dallas produced when it had to for the most part), when we HAD to have ground yardage, we were not very good despite what the overall numbers look like...
     
  11. Larryfinfan

    Larryfinfan 17-0...Priceless Club Member

    I don't have a real problem with Gurley, despite the injury but there are other holes to be filled than RB1/1A... I'd rather see a WR, given that Stills is 'unproven' yet and so is the bulk of our remaining WRs... How about a TE to back up our newly acquired injury case, er I mean Cameron ?? How about a LB or two ? CB ?? S ?? I see those positions as more in need than RB....
     
  12. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    I'm not sure there are many lines in the league that produce at an exceptionally good level when their team has to have ground yardage, i.e., when opposing defenses know they're going to run.

    I suppose one can strive to have the very best offensive line in the league, and dominate opponents on the ground no matter what the situation, but that isn't a very realistic expectation.

    A better and more realistic expectation in my opinion is to strive to have a quarterback and receiving corps that can execute the short passing game to near-perfection when the ground game is needed (i.e., New England in the Super Bowl), and thereby keep opposing defenses off-balance.
     
    Fin4Ever likes this.
  13. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    So let me get this straight. Since the numbers show, he gets worse as he gets more carries, I'm still wrong, because it is entirely conceivable, because we don't know, if he gets really good, at carries 15-19, but 11-14 are so unbelievably bad, his average through 90% of the season last year, was still only HALF the ypc of the first 10 carries.

    So you don't even know if that's the case, but because it's possible, I'm wrong and you're right.

    Even if I let you cut out runs 11 and 12, in post 63 you wrote his 13-19 carry average was 3.58 or something, almost a full 2 ypc below his first 10 carries.

    So you're saying, 3.58 ypc is better than 5.4 ypc. He only had what, 3-4 games with more than 15 carries? He's only had 12 carries the whole season, that qualifies as 16-19, but you are arguing, with certainty, he gets better the more carries he gets.

    The problem with that is, before he gets to 16-19, he has to go through 11-15, and last year, that was crap for most of the season.

    So that is your argument? Just want to be sure that's your argument, and that you're sticking with it.
     
  14. miamiron

    miamiron There's always next year

    2,354
    1,402
    113
    Jan 4, 2008
    Miller had 21 negative yard runs, 15 zero yard runs,18 one yard runs, 26 two yard runs, and 29 three yard runs
     
  15. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    What he's saying is, if Miller gets worse as he carries, then carries 18-20 should be the worst carries of them all. If they aren't, then it wouldn't be accurate to say that Miller gets worse the more he carries. It's a hypothetical.
     
    Tannephins likes this.
  16. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    What we know is that carries 11 and 12 in six games in 2014 were particularly bad (0.67 YPC on average), and carries beyond number 12 in other games were significantly better.

    Again, if later carries are better than earlier ones, assuming an adequate sample size, one can hardly argue that he "gets worse (or less effective) as he carries the ball more."

    The problem you're having here is that you're taking a quote from a news story and not digging into the data enough to determine whether its conclusion is warranted.
     
  17. Fin4Ever

    Fin4Ever Season Ticket Holder Club Member

    9,297
    2,738
    113
    Aug 26, 2014
    Vero Beach, FL
    I don't know if it was the offensive line or Miller that wears down, but somebody does..Also when Williams came in he did not do as well..Well under 4 yards a clip..so tell me? I know that if we Draft Gurley then those problems are solved with the Miller/Gurley tandem..Gurley will mean several extra wins as he will be able to run out the clock and he is really good catching the ball out of the backfield and pass blocking...who gives us more value at 14? :up:
     
  18. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Let's use completely hypothetical numbers:

    Carries 1-10: 5.1ypc
    Carries 11-15: 1.4ypc
    Carries 16-20: 6.2ypc

    VS

    Carries 1-10: 5.1ypc
    Carries 11-20: 3.8ypc

    Would those numbers suggest that the runner was running worse the longer he went? Yes, this hypothetical runner would still have to get through carries 11-15 to get to 16-20, but it would not be accurate to say that the more carries he gets, the worse he is. I realize this is completely hypothetical, and these numbers really in no way are Miller's. Just trying to show the point he is making. The way you list the carries/ypc could affect the overall outlook of the numbers.
     
  19. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    I think some are forgetting that if we have the lead in the 3rd/4th quarter, we're going to be running more and the defense knows it, so the YPC averages should go down and it could have nothing to do with Miller "wearing down".
     
    resnor likes this.
  20. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    I actually was going to make a post about this, but was too lazy.
     
    Fin D likes this.
  21. djphinfan

    djphinfan Season Ticket Holder Club Member

    111,648
    67,540
    113
    Dec 20, 2007
    or how bout that lazors scheme can only expose the defense for so long..idk, obviously its bizarre that miller at such a high yac with the line we trotted out last year
     
  22. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    I agree that's common wisdom for a typical NFL team, but a quick calculation doesn't seem to bear that out for the Dolphins. If you take the total point differentials at the end of the game, OR if you take total point differentials at the beginning of the 4th quarter, and plot total number of carries in the game by Miller against it, the best fitting line is basically flat (if anything it has a slightly negative slope, meaning by a tiny margin, the more we lead the less total carries he has in the game).

    Now, that's not a direct refutation of your claim because I didn't do the calculations for carries ONLY in the 4th quarter (that stat isn't as easy to get from the boxscore), but it does suggest maybe this idea that the Dolphins run more the more they lead is wrong.
     
  23. Fin D

    Fin D Sigh

    72,252
    43,684
    113
    Nov 27, 2007
    WADR, I don't understand how an incomplete study without all the numbers can...suggest anything.
     
  24. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    In your hypothesis, you stated several things that should happen IF we have the lead in the 3rd/4th quarter. All of the things you suggested occur after you have the lead in the 3rd/4th quarter. What that means is that in your hypothesis, you are not suggesting there are any other differences in the game, say before the 3rd/4th quarter begins.

    Now, I'm not saying you personally don't think there aren't more differences, but in the hypothesis you stated you are assuming all else is equal (that is, no matter what conditions led to the point differential in the 3rd/4th quarter, the claims you make should hold).

    OK, that means that the hypothesized differences in the 3rd/4th quarter should show up in the overall # of carries for the game. So what I did is actually a valid (though weak) test of your hypothesis and is (weak) evidence against it.
     
  25. resnor

    resnor Derp Sherpa

    16,327
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Well, I think the hypothesis SHOULD be that when teams knew we wanted to run the ball, whether we had the lead or not, we certainly struggled. I think that is why Lazor used so many short passes. So, for instance, in the third quarter on 3rd and 1, we struggled to run for those gains.
     
  26. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    And I'm saying, he had 4 carries all year that were 18 and 19 of the game. So one wants to stick with the original hypothesis with 4 data points? Exactly.

    I'm playing a game of what. Y'all playing what ifs.

    What if on carries 25-26 he's really good, but 11-24 he's crap? He gets better the more he carries still?
     
  27. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    I think it's in general true that it's easier to defend something the more accurately you know what you should defend yourself against.
     
  28. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    No one is saying he gets better the more he carries the ball; they're simply saying he doesn't get worse the more he carries the ball.

    To show he gets worse the more he carries the ball, as a function of "wearing down," one would need to show that either 1) there is a downward slope in his YPC as a function of his number of carries in a game, or 2) he hits a sort of "brick wall," whereby he plays at a certain level that drops sharply to a much lower level at a certain point, and stays there.

    Neither of those patterns exists in his play.
     
  29. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Wait.. now I'm confused. In both posts #55 and #66 you explicitly say he gets better as he carries the ball more.
     
    Tannephins likes this.
  30. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    I'll recant that at this point and go with the more "middle ground" statement that he doesn't get worse.

    Looking more into the data, I think we find the same confound in the overall correlation (0.45 between number of carries and YPC) that we find when we divide the carries between 1 through 10 and 11 through 20.

    There is really no meaningful relationship between the number of times he carries the ball and his performance.
     
  31. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    I know jdang307 asked about this earlier, but is there some better visualization of the data and/or correlation we can look at? It would really help if we could see the data points themselves. It's a little hard to know what to make of the data when you can't see it.

    If that's not possible, let's be clear about several things. You are saying the correlation between number of carries and YPC is 0.45 whether we look at the entire dataset or whether we look ONLY at 1-10 or ONLY at 11-20? Just making sure what the claim is because based on what was stated earlier the absolute YPC goes down for 11-20.
     
  32. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    The correlation is 0.45 for only the entire dataset (from 2014), taken from here:

    http://www.pro-football-reference.com/players/M/MillLa01/gamelog/

    And if you remove the 97-yard run, that correlation is 0.24.
     
  33. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
  34. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
  35. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    OK, since as has been pointed out, you have to go through carries 1-10 before going through any more, then a positive overall correlation must be the result of (something close to) a weighted sum of the correlations for 1-10 vs. 11-20, with 1-10 having much more weight.

    There are all kinds of possibilities here of course, but there seems to be only one that doesn't create a "jump" or "disconnect" in the middle AND is consistent with a lower average for 11-20: a positive correlation for 1-10 and strong negative for 11-20.
     
  36. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    Yes, and if you take away his best carry, his only run for more than 40 yards, in the last game, after we're knocked out of the playoffs, the 3rd longest run in NFL history, it isn't as pretty.

    I don't advocate taking away carries except for specific reasons, and this is one of them. For analytical purposes.
     
  37. jdang307

    jdang307 Season Ticket Holder Club Member

    39,159
    21,798
    113
    Nov 29, 2007
    San Diego
    As we all know, correlation doesn't equal causation. Is Miller getting more carries when he is running the ball well, or does he run better when he gets more carries? The data doesn't suggest the latter.

    And I'm not just talking about Philbin deciding to give Miller the ball more. I'm talking about extended drives, more plays and more possessions when he runs the ball well, leading to more carries.
     
  38. Piston Honda

    Piston Honda Well-Known Member

    7,853
    8,088
    113
    Sep 23, 2014
    Thats a flawed line of thinking. For one, the sample size is too small to make a determination statistically, how many carries 18-20 does he have, a handful at most?

    The question is, for a guy averaging 5 YPC, why does Miller only get 13 carries per game? What possible rationale is there other than they're trying to keep him fresh/durability issues?
     
  39. Tannephins

    Tannephins Banned

    1,818
    572
    0
    Dec 23, 2014
    Here are the correlations between 1) the number of the rush within the game (rush #1, #2, #3, etc.), and 2) the yards gained on the rush, in each game in 2014:

    Patriots 0.12 (11 carries)
    Bills 0.31 (11 carries)
    Chiefs 0.05 (15 carries)
    Raiders -0.33 (12 carries)
    Packers 0.01 (14 carries)
    Bears -0.24 (18 carries)
    Jaguars -0.07 (14 carries)
    Chargers 0.01 (11 carries)
    Lions 0.47 (4 carries)
    Bills -0.21 (15 carries)
    Broncos -0.55 (12 carries)
    Jets -0.22 (13 carries)
    Ravens -0.43 (12 carries)
    Patriots -0.30 (16 carries)
    Vikings -0.12 (19 carries)
    Jets 0.09 (19 carries)

    Now, if we correlate the number of carries in those games with the correlations themselves, we should have a measure of whether Miller performs worse as he gets more carries.

    That correlation is -0.38.

    So, as Miller gets more carries, the correlation between the number of the rush within the game and the yards gained on that rush decreases.

    However, if the outlier Lions game is removed (4 rushes total), that correlation drops to -0.01, which is meaningless and indicates no relationship in this area.
     
  40. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Umm.. the strategy (in bold) is meaningless. Imagine a situation where some RB consistently gets worse (or better) as he carries it more. Suppose it's so consistent that the correlations are always the same. Pick a number, say -0.2.

    OK, now if number of carries varies per game (# of carries is x-axis), and you plot all the correlations on the y-axis, then you get exactly a straight horizontal line = 0 slope, so you'd then conclude that the RB does NOT get worse as he carries more? Obviously, that can't be true because by assumption the correlation was whatever you set it to be (e.g. -0.2). So this approach won't work.


    No, the simplest (valid) analysis would be to collapse all the data from all games and just look at it (so x-axis is carry #1, 2, 3, etc.. and y-axis is yards gained). Or if you want to compute stats, compute the correlations for 1-10 vs. 11-20 etc..
     

Share This Page