How box scores distort perceptions of running back success

Joe Nicholson-USA TODAY Sports

I present to you two running backs and their box scores.

RB #1: 10 carries, 54 yards rushing, 5.4 YPC (yards per carry)

RB #2: 10 carries, 49 yards rushing, 4.9 YPC

One might say that all other things being equal (lack of touchdowns etc) that the two players are about the same with RB #1 maybe being slightly better this would be even more pronounced if maintained over the course of a season.

The problem with this and other aggregate measures and the simple efficiency statistics like YPC is they ignore the effects of individual plays in favor of the (presumed) greater stability of the bulk. This is a perfectly reasonably approach if you lack detailed information (an individual play can vary from as much -99 yards to 100 yards though much less in reality) and it makes the math easier for the average fan but it often leads to bad players getting far too many touches and/or the game to get wildly out of hand before coaches/teams make adjustments.

So let's return to the above comparison shall we?

RB #1 gained his 54 yards on the following runs -1 yard, 8 yards, -2 yards, 5 yards, -3 yards, 4 yards, 0 yards, 40 yards, 3 yards & 0 yards.

RB #2 gained his 49 yards on the following runs 6 yards, 0 yards, 10 yards, 2 yards, 15 yards, 4 yards, 5 yards, 3 yards, 0 yards and 4 yards. RB #2 had no home run runs (heh!) and I doubt many would be interested in choosing RB #2 over RB #1 with this information. But let's put this through a measure of statistical rigor eh?

Assume for a moment that you had the same RB's play the exact same game 1000 times and you needed to get a 1st down with 3 consecutive runs from both players, which player gets you more 1st downs?

RB1 1st Downs

# of 1st Downs by RB1

RB #1 gets a 1st down 43.5% of the time, not great but not terrible either. Let's look at RB #2 then.

RB1 2nd Downs

# of 1st Downs by RB2

RB #2 gets a 1st down for you 74.6% of the time an incredible difference which makes running the game not only feasible, but *good* in the greater scheme of the offense despite the lack of homerun ability.

Football Outsiders and Sharp Football statistics both try to get at this with their DYAR/DVOA for RBs with the former and Success Rate with the latter. Football Outsiders compare a given run play to how all other RBs perform with a similar play/situation and Sharp Football Statistics assesses how many yards towards a 1st down that a given play gets to identify it as successful or not.

These measures both (from my own use in fantasy football) and their own internal benchmarking tell you a lot more about who is likely to consistently perform well game-to-game, year-to-year and if you're into betting, likely win you some money.

These measures aren't always right but they're a fair sight better than volume measures and an improvement on basic efficiency measures. Don't use YPC, and if you absolutely have to, calculate it after removing the longest run in that game.