Gravity,
With respect to DVOA and the like, are these statistics totally independent of the team's efficacy in other phases of the game? Suppose Tannehill faces twice as many 3rd down and 10+ yards to go instances versus the average QB or the top QB's? I am assuming DVOA rewards a completion that results in a 1st down more than completions which do not. If so, is Tannehill being penalized under the assumption that he is facing more adverse down and distance situations than other QB's? And I am assuming that if he is put in these worse situations by being sacked more and having an ineffective running game, that the DVOA does not reflect that he is not responsible for being in more of these situations, for the most part.
Based on the functional equivalence between DVOA and YPA statistics, I think the following post I did the other day is relevant to your question:
Your line of reasoning inspired me to do a t-test comparing the top 10 QBs in the table in the original post with the QBs ranked 41 through 50 in the table, based on their season-by-season YPAs throughout their careers, using only the seasons in which they started 10 or more games.
If it's true that supporting cast variables are more responsible than individual QB ability for YPA, then the within-group variation should be at least non-significantly different from the between-group variation.
This isn't the case, however. The mean of the QBs in the 1-10 group is 7.73, with a variance of 0.659, and the mean of the QBs in the 41-50 group is 6.98, with a variance of 0.277.
This leads to a t-statistic of 6.75, with a p-value of 0.0000000004.
In other words, there is a four in one trillion chance that the observed difference in the means between the two groups would occur by chance.
This supports the analogy I used earlier in the thread regarding weather and climate. Although there may be variation in the weather (i.e., variation from season to season in QBs' YPAs), the climate (i.e., QBs' individual ability) is driving the bus to a far greater extent.
In other words, while Peyton Manning's YPA may vary between 6.5 and 9.2, and while Mark Brunnell's YPA may vary between 6.3 and 7.8, despite the variation in both QBs' YPAs, Peyton Manning's average YPA is significantly higher than Mark Brunnell's. And this is the case for the larger comparison as well, between QBs 1-10 and QBs 41-50 in the table.
That about does it on that front, folks. :)
The "table" referred to in the above post is this one:
http://www.pro-football-reference.c...pos_is_db=Y&draft_pos_is_k=Y&draft_pos_is_p=Y
In other words, while it certainly makes sense and is prudent to consider situational variables in theory, I think these things we're talking about are being driven to a much, much greater extent by QBs' individual ability. Even the QBs' DVOA rankings themselves correlate with the consensus perceptions of individual ability.
Then of course there is also the additional layer of complexity to consider when addressing issues of individual ability versus situational variables, in that quarterbacks with better individual ability are likely going to improve the play of everyone around them.
To use an extreme example, Peyton Manning is
so good at bringing the entire playbook to the line of scrimmage, making audibles to the correct play, and getting rid of the ball quickly, that we probably really have no idea how good or bad the Denver offensive line truly is. They could be utter garbage, and his play alone is going to mitigate that considerably. Likewise for running games, whereby defenses have to respect the ability of a great QB and can't sell out against the run game, lest the QB torch them downfield.
Now, I
will concede this, however: given that the YPAs of even great QBs do vary from season to season, likely due to situational variables, we don't yet have enough of a sample size to determine whether Ryan Tannehill's relatively low YPAs in his first two seasons represents such variation.
However, once again, when we look through the game-to-game data and find that his performance in 2013 didn't improve when he was either 1) sacked less, or 2) the beneficiary of a better running game, and we consider that he wasn't pressured any more often than the average QB, it isn't a promising sign in my opinion, because there is no objective evidence available, then, to suggest that his overall YPA in 2013 was due more to those situational variables than to his own individual ability.
