Last year, I looked at the unusual running back by committee in Arizona in 2013. Rashard Mendenhall was the team’s primary back, but he averaged 3.17 YPC that season, while Andre Ellington averaged 5.53 YPC. To measure how “unusual” the split was, I came up with the following methodology: calculate the difference between the YPC of the top two running backs (as measured by carries) on each team, and multiply that difference by the number of carries given to the running back with fewer carries. So for the 2013 Cardinals, the difference between Ellington and Mendenhall in terms of YPC was -2.36; we multiply that by 118 to get a value of -278. For 2014, the most extreme result along this line came in Minnesota.
Matt Asiata had 164 carries for the Vikings but gained just 570 yards, for a 3.48 YPC average. Meanwhile, Jerick McKinnon rushed only 113 times but picked up 538 yards, a 4.76 YPC average. So McKinnon averaged 1.28 more yards per carry than Asiata. Then, we multiply -1.28 by 113, which produces a value of -145, the most extreme of the 32 teams last year.
The reason for this two-step process is that when dealing with backup running backs, you sometimes get small sample sizes. For example, Latavius Murray averaged 5.17 YPC on his 82 carries, but the Raiders split doesn’t count quite as extreme as the Vikings split based on this method. Also, the Cowboys split would look pretty funky if you didn’t penalize RB2s that had only a handful of carries: [click to continue…]