[qwerty49]

[quote author=CTcard link=topic=16635.msg191261#msg191261 date=1491455691]

I think the stats below are pretty interesting, though of course they don't tell you everything.
It's pretty common to focus on carries by running backs in trying to get a feeling for the OL run blocking, for example Football Outsiders uses RB carries as the basis for their adjusted line yards stat, which Hulk apparently referred to above.

For the last several years, here are the RB avg per carry (for efficiency), the RB yds/game (for total contribution/degree to which running carried the load), the Football Outsiders Ajusted Line Yards for the three years available, and notes giving the QB, main RB, and season finale.

I would be interested in knowing the median yards per attempt each year as well, and/or the standard deviation of the yards per attempt. I think that this years  averages are inflated by 1) the last 6 opponents being "bad" run defenses, and 2) CMAC breaking a few long runs. The problem can be seen as follows. Suppose a back runs 4 times and makes 5 yards each time. The mean and median are 5 yards per carry and we have the ball and a first down. Now, instead, he makes 20 yards, and then 0 yards 3 times. Same average, but we punt. The mean is 5 but the median is 0. That was us last year. CMAC/BL could work magic sometimes, but not enough. Granted, this is an extreme, contrived example, but things like it seemed to happen in the games we lost, which caused us to have to give up the ball much more often than usual. It is also very fair to note that the OL had to contend with defenses that didn't fear the pass much (Washington State and Colorado anyway) and loaded up against the run. In those cases it seemed we would get a 10+ yard run and then negate it by getting stuffed a couple of times. Note also in my contrived example above, we probably wouldn't have run on third and 10. We would have attempted a pass, most probably. It would be incomplete or a sack, and we punt. The run average goes UP to 6.66 YPC, but we still punt. The average can be a very misleading statistic if the distribution is skewed by  large outliers, which I think ours was.
Now, if those outliers are large enough that they result in touchdowns, that is a different case, and the fans (at least) don't care. I think the coaching staff smiles in public but worries in private about that case. But last year in a few games there were too few "big plays" to compensate for the no/short gains. Stanford isn't trying to be a "big play" offense, at least on the ground. We want to control the ball. We rely on the OL to stay away from the "bad" plays on the ground. If that doesn't happen, we have a problem.
[/quote]

Spot on.  I'd love to see a histogram distribution of CMac's carries because I suspect it confirms your observations.  And actually it would be a pretty complex analysis, because at the next level one would have to consider down-and-distance, i.e. I doubt the distribution on 1st down is the same as on 3rd and 2.  And yes, strength of defense (e.g. running vs Rice is a whole lot different than vs U-Dub) would complicate that theoretical even further.  Easy for an armchair statistician to say, LOL.  But hey, it's April!  ;)

[qwerty49]

[quote author=CTcard link=topic=16635.msg191261#msg191261 date=1491455691]

I think the stats below are pretty interesting, though of course they don't tell you everything.
It's pretty common to focus on carries by running backs in trying to get a feeling for the OL run blocking, for example Football Outsiders uses RB carries as the basis for their adjusted line yards stat, which Hulk apparently referred to above.

For the last several years, here are the RB avg per carry (for efficiency), the RB yds/game (for total contribution/degree to which running carried the load), the Football Outsiders Ajusted Line Yards for the three years available, and notes giving the QB, main RB, and season finale.

I would be interested in knowing the median yards per attempt each year as well, and/or the standard deviation of the yards per attempt. I think that this years  averages are inflated by 1) the last 6 opponents being "bad" run defenses, and 2) CMAC breaking a few long runs. The problem can be seen as follows. Suppose a back runs 4 times and makes 5 yards each time. The mean and median are 5 yards per carry and we have the ball and a first down. Now, instead, he makes 20 yards, and then 0 yards 3 times. Same average, but we punt. The mean is 5 but the median is 0. That was us last year. CMAC/BL could work magic sometimes, but not enough. Granted, this is an extreme, contrived example, but things like it seemed to happen in the games we lost, which caused us to have to give up the ball much more often than usual. It is also very fair to note that the OL had to contend with defenses that didn't fear the pass much (Washington State and Colorado anyway) and loaded up against the run. In those cases it seemed we would get a 10+ yard run and then negate it by getting stuffed a couple of times. Note also in my contrived example above, we probably wouldn't have run on third and 10. We would have attempted a pass, most probably. It would be incomplete or a sack, and we punt. The run average goes UP to 6.66 YPC, but we still punt. The average can be a very misleading statistic if the distribution is skewed by  large outliers, which I think ours was.
Now, if those outliers are large enough that they result in touchdowns, that is a different case, and the fans (at least) don't care. I think the coaching staff smiles in public but worries in private about that case. But last year in a few games there were too few "big plays" to compensate for the no/short gains. Stanford isn't trying to be a "big play" offense, at least on the ground. We want to control the ball. We rely on the OL to stay away from the "bad" plays on the ground. If that doesn't happen, we have a problem.
[/quote]

Spot on.  I'd love to see a histogram distribution of CMac's carries because I suspect it confirms your observations.  And actually it would be a pretty complex analysis, because at the next level one would have to consider down-and-distance, i.e. I doubt the distribution on 1st down is the same as on 3rd and 2.  And yes, strength of defense (e.g. running vs Rice is a whole lot different than vs U-Dub) would complicate that theoretical even further.  Easy for an armchair statistician to say, LOL.  But hey, it's April!  ;)