Suppose we sort each running back’s carries in descending order by yards gained. How many carries would we need to take away from Murray in order to drop his YPC average to at or below league average? Same question for Charles. I’ll give you a moment to think about this one.
[Final Jeopardy Music]
[Are you ready?]
[Your time is now up. Post your answer in the comments!]
For Murray, the answer is just seven.1 The Cowboys star had runs of 51, 44, 40, 32, 28, 27, and 26 yards this year, for a total of 248 yards. Without those runs, Murray would have had 1,597 yards on 386 carries, which translates to a 4.14 YPC average. For Jamaal Charles, the answer is just five. His top five runs went for 63, 47, 39, 36, and 28 yards, a total of 213 yards. Without those carries, Charles would be at 826 yards on an even 200 carries, a 4.13 YPC average.
There were 19 running backs2 last season who recorded at least 100 carries and averaged more yards per rush than league average. Among those players, Justin Forsett is the answer to the question: Which player could lose the most number of top carries and still be above average in YPC? For the Ravens runner, the answer is 9. Forsett rushed 235 times for 1,267 yards last year3, and averaged 5.39 YPC. The far right column shows the minimum number of top carries each back would need to have erased in order to bring his YPC average below 4.16.
So, what is the point of this study? Well, for one thing, I just think this whole thing is kind of interesting. Consider Jeremy Hill, who had a great rookie year, rushing 222 times for 1,124 yards, producing an impressive 5.06 YPC average. But you only need to take away four of his 222 runs — which went for 85, 62, 60, and 30 yards — to bring his YPC average down to 4.07. Even if you just take off his top three runs, that brings Hill down to 4.19 YPC. That just seems kind of crazy to me.
This is a good reminder of how sensitive YPC is to outliers. Even within a season, YPC regresses by about 70% to the mean; from season to season (based off of larger sample sizes), the regression is about 65%. And we know that YPC is not very useful when trying to predict success rate. According to Danny Tuccitto, YPC is basically a bunkum stat.
The other interesting thing here is I like the way this stat automatically adjusts for number of carries. Seven running backs had higher YPC averages than Murray, but only one of those players could lose six carries and still be above average like Murray. So from a ranking standpoint, this is kind of neat. And by the way, I really slept on Lamar Miller having such a great year. I assumed his 5.09 YPC was inflated by his 97 yard touchdown run, but he still shines in this exercise. Miller also ranked 3rd in DVOA and had the best success rate of any running back with 100+ carries last year.
- An informal offline poll of two people produced identical answers of fifteen. [↩]
- For those curious, the number is a mind-boggling 21 for Russell Wilson, 12 for Cam Newton, and 9 for Colin Kaepernick. [↩]
- Note that because I used the play-by-play logs to conduct this exercise, there may be slight differences between the numbers in this table and the official numbers. [↩]