In 2015, Thomas Rawls averaged 5.65 yards per carry, best in the NFL. The next year, Rawls averaged just 3.20 YPC, the second-worst rate in the league. That’s an extreme case (actually, the most extreme case), but it does represent the general idea that yards per carry is simply not very sticky from year to year.

In the graph below, I have shown all running backs who had at least 100 carries in back to back years since 1970. The X-Axis shows the YPC each player had in Year N, and the Y-Axis shows the YPC that player had the following year, Year N+1. The R^2 shows the correlation between those two numbers. The best-fit formula to predict Year N+1 Yards per Carry from Year N Yards per Carry is 3.05 + 0.28 * Year N YPC. The R^2, of course, is the square of that 0.28 coefficient. Take a look at this graph:

What if we raise the minimums to 150 carries both years?

Or 200 carries:

Even at a minimum of 250+ carries — and think how good a running back must be to get 250+ carries in consecutive years — the R^2 remains very low:

Yards per carry is very sensitive to outliers, and outliers are very inconsistent from year to year. As a result, I don’t think these results are surprising, but it’s always good to actually see the data.