clock menu more-arrow no yes

Filed under:

Perhaps "YPC" isn't a fair way to compare two backs

New, 23 comments

We've been talking a lot about how Quizz probably won't play on Saturday, but there still have been a lot of comparisons drawn between Oregon running backs and Quizz this week, notably over the difference in their "Yard Per Carry" averages. (ex: "Oregon has 12 players who get more yards per carry than Quizz, he'd be 13th on our depth chart") Well, here's the reason why we shouldn't completely rely on such an inconsistent statistic.

As I was preparing this post, I decided to look at the stats of Jeremiah Johnson and Jacquizz Rodgers. For the sake of simplicity, I decided to look specifically at each players' best game. For Jeremiah, that was Oregon's game a few weeks ago against Stanford.  For Quizz, it was the game against USC.

Let's first start by examining Jeremiah's performance. He ran the ball 15 times in the game for 135 yards, and came out with a 9.0 yard per carry average. However, two rushes-- a 41 yard touchdown in the first quarter, and a 47 yard rush from Oregon's 12 yard line in the third quarter- made up for over 65% of his yards in the entire game. I understand that those two rushes count too, but as the graph indicates below, Jeremiah only rushed for over his average of 9 yards three times in the entire game. Take out the two big gains, and Johnson had 13 carries for 41 yards, a 3.1 yard per carry average. All against the Pac-10's 7th-best rushing defense.

Jacquizz Rodgers had 187 yards on 37 carries against the Trojans. The first major difference is the amount of carries Jacquizz received compared to Jeremiah's (Oregon's other predominant back, LeGarrette Blount, had 10 rushes against the Cardinal). Quizz isn't the type of back who picks up yards in chunks of 40-- he pounds ahead for 3-8 yards per carry.

Now, the graph: (each hash mark on the axis represents 2 yards)

7699271a1e08e091b499b77a160b4618_medium

I pulled this "box and whisker" technique out of my sixth grade math arsenal-- It may not be the best way to show the information, but it works. The median of the data (the middle number) becomes the center line you see in the middle of the box. That creates two sets of data, and again, you take the median of each half of data, creating the left and right sides of each box. The "whiskers" extend from the medians of each half to the smallest and largest numbers. 

As you can see on Jeremiah's graph, the two long runs show up as "outliers" when you input the data into a graphing calculator.

Now, take a look at the red lines. Those are the averages. You can see that Jeremiah's average is outside his "box", meaning that at least 75% of his rushes were under his average. In this case, the actual number is 80%. 

In the case of Quizz, his average lies near the middle of his rushes. 

I realize that you can manipulate statistics to say whatever you want them to say, but hopefully this shows that Jeremiah and Quizz have similar stats. They've both been putting up good numbers, and just because Jeremiah has broken more long rushes than Quizz has doesn't mean that he consistently picks up more yards than Jacquizz every time he touches the ball. In other words, just because Johnson picked up 65% of his total yards in his best statistical game on two plays and his average is four yards higher as a result doesn't mean he's the better back.

And did I mention that Quizz's best game was against the Pac-10's best defense?

I realize, it's like comparing apples to oranges, but maybe you should give Quizz some more respect.