(01-11-2019, 03:41 PM)BostonCard Wrote: There are two major problems with this analysis, …
The first, of course, is acknowledged in the piece, which is that in large part, any play sequence that ends in a run is likely to have been more successful than one that ends in a pass ...
The second critique is more broad. Play-calling is not completely independent, even between sets of downs. That is, the idea of "establishing the run" is in part that running early benefits the offense later in the game. ….
I suppose you could group most of the fundamental issues into those groups. The problem is that those two sets of problems are not in the category of "this is all basically correct but you might end up with a little more variance than predicted", but rather more in the category of "and this is where my theory falls to the ground".
Or at least they might be. None of these kind of articles that I have seen really addresses the robustness of their conclusions.
The first issue is raised in the piece, but only to say that one can ignore findings that don't fit the premise of the article. That isn't very helpful.
The second critique is even broader than you state. Play calling and play success are far from random events but are highly correlated. An example on the play success side of the coin: I don't think much of anyone would claim that Russell Wilson is the best or most talented thrower among the QBs in the NFL, yet Seattle had the third best pass efficiency rating in the league. That's at least partly because they didn't throw as often as most (fewest attempts in the league). If Seattle passed as often as Pittsburgh or Indy their efficiency would certainly have declined.
This strong correlation means the standard statistical machinery doesn't apply; not that the machinery
almost applies.
If it did apply, that would mean every team should pass on first down every time. And as noted, it would mean every team should run on third down every time. The author acknowledges the second not so much because it is patently absurd, but because it doesn't match his narrative. I'd say the first is also patently absurd.
That's not to say that most teams wouldn't benefit from a careful analysis, including proper statistical analysis, of what they are doing. It does say that most teams would not benefit from this sort of shoddy statistical analysis.
---------------
BostonCard Wrote:... However, even though I can rightfully be called a "Shawpologist", even I wished we would pass more on first and second run.
Even this past year?
How much did we already pass on first down and run on second?
I do think that Stanford more than most teams tends to have particular plays tied to particular personnel groupings and formations, i.e. telegraphing the play. This is inherent in the approach, which is to have the best possible personnel for particular plays, i.e. to win the play by superior personnel.
Stanford thinks that 8 linemen will do the best blocking for some run plays, or 4/5 receivers give the best options on many pass plays. Stanford also feels that opposing defenses will either mess up trying to constantly swap personnel or simply not be able to match up with some groupings.
Those are the positives of what Stanford tries to do with all of the different formations and personnel groupings. There are also some obvious negatives to this approach, discussed over and over on this board.
I am not sure what is a better option for Stanford overall, but I do think Stanford needs to have a two minute (or one minute) package without personnel changes for end of game situations without adequate timeouts.
