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

Offensive Stat Improvement without Ajayi

Discussion in 'Miami Dolphins Forum' started by Galant, Nov 24, 2017.

  1. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    What if defenses key on passes to the backs over the next 7 games, and reduce the effectiveness of it? After the season, you might get stats that don't show an overall increase of production, but it would not mean that these last two games weren't more productive due to his absence.
     
    cbrad likes this.
  2. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Hey! Finally a good question.

    That is precisely the reason why in post #40 I wrote that even if we find that pre-Ajayi vs. post-Ajayi stats are consistent with random variation, the story isn't over and we have to dig deeper. I said I'd cover that later but why wait right?

    What you should do depends on the precise nature of the hypotheses, but in most cases simple correlations should be sufficient. For example, you could compare the correlation between some QB efficiency stat (which one depends on your hypothesis) and total passing yards with the correlation between some efficiency stat for Williams and Drake and total passing yards. The square of the correlations estimate the amount of "variance explained" in total passing yards, thus giving you direct evidence of the degree to which each proposed mechanism was responsible for changes in the target stat.

    And that should make sense intuitively too. If shutting down Williams and Drake leads to loss of production, then that will show up very clearly in a correlation between your chosen measure of their production and the target stat.
     
  3. djphinfan

    djphinfan Season Ticket Holder Club Member

    111,652
    67,546
    113
    Dec 20, 2007
    Ok just so I know what your trying to say for sure...what other factors, do the numbers thus far revealed, with this statistical discrepancy, have been more important, or have affected this discrepancy, than Ajayi himself?

    Also...do you see a big difference in the runnings backs abilities today to get into routes, catch and get upfield, then you did when Williams and drake were not getting reps and Ajayi was the third down back..?
     
  4. djphinfan

    djphinfan Season Ticket Holder Club Member

    111,652
    67,546
    113
    Dec 20, 2007
    Also Cbrad, we have to do a study on the sack drop...because the same holds true for me in this case..common denominator is Ajayi...and because of the abilities of the new backs we are better on third down and avoiding sacks because of the outlet they present to a qb.
     
    resnor likes this.
  5. SocaCasualuk

    SocaCasualuk Active Member

    223
    88
    28
    Sep 21, 2014
    U.K.
    DJ nailing it again. Always listen to those who don’t shout but talk.
     
  6. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    Like I said.. how the QB has played. When you combine both Cutler and Moore together into a single QB, then the game-by-game correlation between their passer rating (which is an efficiency stat) and total passing yards so far is 0.65 which is pretty darn big! That means you have a viable hypothesis here in claiming much of the variance in total passing yards is due to changes in how our QB's have played, measured by PR (about 42% of the variance is explained so far this way).

    Of course, if a t-test shows post-Ajayi is unlikely to be due to random chance, then that will trump any correlation because it means the stat is beyond a threshold at which a statistician will actually reject a hypothesis.
    Absolutely. I definitely think this is a viable hypothesis. Problem is it's not the only hypothesis consistent with the data, and of course I'm only reporting on my bias when asked what my "eyes" see.

    One practical issue with this hypothesis that we may have to revisit later on is which available stats are best to use for correlations to a target summary stat like total passing yards (same goes for incorporating sacks into this), but we'll worry about that only after the t-test for pre- vs. post-Ajayi.
     
    Last edited: Nov 25, 2017
  7. Rock Sexton

    Rock Sexton Anti-Homer

    2,553
    1,793
    113
    Mar 14, 2015
    Bingo.

    Having him in there tipped the cap to defenses and then he compounded matters by not following the designed runs.
     
  8. hitman8

    hitman8 Well-Known Member

    3,040
    2,485
    113
    Nov 11, 2016
    That's because Defenses are concentrating on stopping the run when ajayi was playing.
     
  9. hitman8

    hitman8 Well-Known Member

    3,040
    2,485
    113
    Nov 11, 2016
    No, actually gase's inflexible and non creative playcalling is what made matters worse. If you have a guy who scares defenses into stacking the box and stopping the run, a good coordinator would take that opportunity to attack with the pass and make them pay. Instead gase kept trying to run ajayi into a brick wall then got mad when ajayi tried to find a way around that wall on his own.
     
  10. hitman8

    hitman8 Well-Known Member

    3,040
    2,485
    113
    Nov 11, 2016
    Yes, drake and williams are better route runners and receivers, but what I was arguing against is that they are better pass protectors, they are not. Ajayi was the best pass protector of the three in terms of staying in the pocket and picking up blitzes and rushers. Williams has been terrible at picking up blitzes going back to last season.
     
  11. Rock Sexton

    Rock Sexton Anti-Homer

    2,553
    1,793
    113
    Mar 14, 2015
    Except A: Cutler couldn't/wouldn't attack downfield and B: There were still creases into what Ajayi was supposed to run into - EVEN when they had 8 in the box.
     
  12. Rock Sexton

    Rock Sexton Anti-Homer

    2,553
    1,793
    113
    Mar 14, 2015
    Ya, that's it. You quickly forget it's still Cutler back there.
     
  13. resnor

    resnor Derp Sherpa

    16,329
    9,874
    113
    Nov 25, 2007
    New Hampshire
    Maybe there is a reason that he didn't do that? It's probably not as simple as Gase is an idiot and just runs the ball up the middle instead of passing.
     
    Last edited: Nov 26, 2017
    rackhound likes this.
  14. cbrad

    cbrad .

    10,659
    12,657
    113
    Dec 21, 2014
    OK.. so now we have our 4th game post-Ajayi. Let me provide some updated offensive stats.

    So before people say we were playing NE, let's note something: NE was LAST in the league prior to this game in total yards allowed per game at 401.9 and also LAST in total passing yards allowed per game at 282.

    Also.. note that NFL defines total passing yards as passing yards minus whatever was lost through sacks. So total passing yards today is not 215 but is instead 154 because we had 7 sacks for 61 yards. Similarly, total offensive yards today is 221 because of the sacks subtracted from passing yards.

    Given that, as I predicted, regression to the mean is already happening and in a big way. That 2-sample t-test already with one extra game no longer shows that the first 7 games pre-Ajayi vs. last 4 post-Ajayi are statistically speaking unlikely enough to come from separate distributions that we can reject the claim they come from the same distribution.

    The updated correlation between combined QB passer rating (combining Cutler and Moore into a single QB) this year and total passing yards is now slightly higher at 66.46. The correlation with total offensive yards is even higher at 0.7. Guys.. these are very high correlations, so don't ridicule the hypothesis the primary determinant of total passing yards and total offensive yards is QB efficiency measured by passing rating!

    Some other updated stats: Moore's passer rating under Gase (2016 + 2017 + playoff game last year) is 89.6. That number is directly comparable to Tannehill's 93.5.

    Oh.. and regarding the Williams injury today, Moore actually had better stats post-injury than pre-injury. Pre-injury Moore had 12 completions in 18 passing attempts for 87 yards, 0 TD's and 1 INT, while post-injury Moore had 11 completions in 16 attempts for 128 yards with 1 TD and 1 INT. Thus, pre-injury his rating was 54.63 while post-injury it was 87.5.

    Anyway.. that was just one game and we'll see how the rest of the season goes, but regression to the mean was definitely seen today and that correlated very well with how Moore played.
     
  15. finsfandan

    finsfandan Well-Known Member

    2,547
    600
    113
    Dec 14, 2014
    This. Why did we gift the Eagles Ajayi for a low fourth rounder again?
     
  16. hitman8

    hitman8 Well-Known Member

    3,040
    2,485
    113
    Nov 11, 2016
    Because tbaum and gase are idiots, we always get the short end of the stick when trading. I can't remember the last time we actually made a good trade. We almost always overpay for mediocre players, our drafting also sucks balls.

    This FO definitively has to go. I don't know why Ross has been sticking with tbaum the fat **** idiot for so long. Tbaum should have been fired two seasons ago.
     
    finsfandan likes this.

Share This Page