Good question.

Yards per reception for RBs was nowhere near statistically significant, IIRC, so I chose to just use receiving yards.

As for yards/rush over rushing yards, whether I used both yards per rush and rush attempts or rushing yards, the correlation coefficient was roughly the same. But it seemed preferable to split up rushing yards in this manner; even if it doesn’t make a big difference when you look at hundreds of running backs, it felt like the right thing to do.

For what it’s worth, the formula spits out these projections:

300 carries for 100 yards = 988

200 carries for 100 yards = 915

I think that makes sense: I would imagine that extreme YPC averages on both sides are likely to regress to the mean, whereas number of carries is a very telling statistic.