Previous post:

Next post:

Yards per Carry and Points per Drive

by Chase Stuart on September 10, 2013

in Rushing, Statistics

I’ve written a couple of times about “yards per carry” as a key statistic to grade running backs. The usual argument in favor of using YPC is that a running back who rushes for 1200 yards on 300 carries is less valuable, all else being equal, than one who rushes for 1200 yards on 250 carries. But when it comes to running backs and yards per carry, “all else” is is never equal. Two players come to mind whenever yards per carry is cited for running backs: Eddie George and Curtis Martin.

From 1996 to 2002, George led the league with 2,421 carries. Only Martin (2,236) was within 300 carries of George. In NFL history, Eric Dickerson is the only player to ever record more carries during a player’s first seven seasons. But some would have you believe that George wasn’t very good during those seven years, because he averaged just 3.71 yards per carry. During that stretch, the Titans went 68-44, giving them the fourth best record in the NFL and the second-best mark in the AFC during that span. But, the yards-per-carry proponents would argue, Jeff Fisher didn’t know what he was doing when he kept handing the ball off to George, play after play, game after game, year after year.

In 1998, the Jets went 12-4 and earned a first-round bye; New York went 12-1 in Vinny Testaverde’s thirteen starts, and finished in the top five of the league in points, yards, and first downs. That season, Curtis Martin received 369 carries despite missing one game due to injury. Martin rushed 25+ times in seven games and recorded at least 17 carries in every game that year… and averaged only 3.49 yards per carry. There are some who would have you believe that the Hall of Fame head coach didn’t quite know what he was doing that year, and the Jets would have been even better had the team called Martin’s number less frequently.

It’s not that I think yards per carry is meaningless, or that it’s better to produce a low YPC average. I just think that “yards per carry” is only one small piece of evidence to let us know how talented or valuable or productive a running back is. And in my opinion, the number of carries he receives is a much bigger piece of evidence of that player’s abilities. But today, I want to look at yards per carry in connection with another statistic: points per drive.

First, we should keep in mind that YPC has not been consistent throughout history, and was at a relatively low point during the primes of the careers of Martin and George. The graph below shows the league-wide yards per carry average — among only running backs — since 1970:

RB YPC

That’s an interesting graph1, but that is only the supporting material to get us to what I really want to examine. There are thirty-seven running backs since 1970 who have met the following criteria:

  • Finished at least 8% below league average in rushing yards per carry; and
  • Recorded at least 70% of their team’s rush attempts (among running backs)

This lets us see those players who were both workhorses and had poor YPC averages (taking 8% off of a 4.25 YPC average would include all players who averaged fewer than 3.91 yards per carry). My thinking is that if you are the workhorse for a productive offense, that’s very strong evidence that you are doing something right: great offenses don’t tend to waste three hundred plays on mediocre players. This is something Jason Lisk, Doug Drinen, Neil Paine and I used to discuss behind the scenes when the four of us contributed to the PFR Blog, and Jason was the one who suggested we compare running backs by both yards per carry and points per drive.

The table below lists these 37 running backs, but the table is sorted by the Offensive Points Per Estimated Drive produced by that team.2 LaDainian Tomlinson comes up first on the list, and here’s how to read his line. In 2008, a 29-year-old LT put up a 292-1110-11 stat line. That’s a 3.80 yards per carry average in a year when all running backs averaged 4.25 yards per carry. Tomlinson therefore produced just 89.5% of the league-average YPC, but he received 75.8 of all carries that went to Chargers running backs that season. And in 2008, San Diego ranked 2nd in Offensive Points Per Estimated Drive. George also appears several times at the top of this list, including some brutal yards per carry averages during years when the Titans still had a very good offense.

Running Back
Team
Yr
Age
Rsh
RshYd
RTD
YPC
LgAvgYPC
% YPC
%ofTmCar
OPPED Rk
LaDainian TomlinsonSDG2008292921110113.84.2589.5%75.8%2
Curtis MartinNYJ199825369128783.493.9787.8%79.9%4
Edgar BennettGNB199526316106733.383.9785.1%87.5%4
Eddie GeorgeTEN200330312103153.34.1879%72.4%6
Karim Abdul-JabbarMIA199723283892153.15478.7%70%6
Rudi JohnsonCIN2006273411309123.844.291.5%86.8%7
Eddie GeorgeTEN2002293431165123.44.1581.8%81.5%10
Ryan GrantGNB200826312120343.864.2590.8%83.4%12
Eddie GeorgeTEN20012831593952.984.0473.8%81.6%12
Jamal AndersonATL199725290100273.46486.3%75.1%15
Jamal LewisBAL200627314113293.614.285.9%78.3%16
Adrian MurrellNYJ199727300108673.62490.4%77.7%16
Edgerrin JamesARI200729324122273.774.1690.6%88.5%17
Edgerrin JamesIND20022427798923.574.1585.9%70.7%17
Lamar SmithMIA20013131396863.094.0476.6%81.9%17
Edgerrin JamesARI200628337115963.444.282%89.4%19
LaDainian TomlinsonSDG2001223391236103.654.0490.3%91.4%19
Ricky WilliamsMIA200326392137293.54.1883.7%89.5%20
Jerome BettisRAM199422319102533.213.7585.7%90.6%20
Cedric BensonCIN201028321111173.464.1882.8%81.5%21
Matt ForteCHI20092425892943.64.2984%78.9%21
Rodney HamptonNYG199425327107563.293.7587.7%72.3%22
James WilderTAM1985273651300103.564.1585.8%95.3%22
Steven JacksonSTL201027330124163.764.1890%84.8%23
Thomas JonesNYJ200729310111913.614.1686.7%81.4%24
Earl CampbellHOU19822715753823.433.8489.2%78.5%24
Willis McGaheeBUF20062525999063.824.291.1%70.4%25
James StewartDET2000293391184103.493.9987.5%89.2%25
Trent RichardsonCLE201222267950113.564.2484%76.7%26
Curtis EnisCHI19992328791633.193.8982%84.2%26
Jamal AndersonATL200028282102463.633.9991%88.7%27
Errict RhettTAM1995253321207113.643.9791.6%91.5%27
Earnest JacksonPHI198526282102853.654.1587.8%74%27
Eric DickersonIND19913116753623.213.9681.1%75.9%28
Jamal LewisCLE200829279100243.594.2584.5%80.6%30
Reuben DroughnsCLE20062822075843.454.282.1%70.3%30
Clinton PortisWAS200423343131553.834.1891.6%78.9%31

Of course, once you see the bottom of the list, it’s easy to form the counter-argument. Perhaps the most logical interpretation is that if a running back is going to have a poor YPC average, it’s more impressive if he does it on a bad offense like Trent Richardson or Clinton Portis. After all, if a running back is facing stacked boxes because of his poor quarterback play, it seems easier to forgive them. On the other hand, it strikes me as much fairer to wonder why a bad offense gave so many carries to a running back who had a poor rushing average.

I’m not sure where I stand on this. So I thought I’d produce the data and see what you guys have to say. How would you use yards per carry and points per drive?

  1. Query: Did the 3.74 yards per carry average in 1994 lead to the passing explosion in 1995? []
  2. OPPED is the basis for Approximate Value and eliminates some of the problems associated with just using points scored. []

{ 12 comments… read them below or add one }

Danish September 10, 2013 at 3:43 am

Well the bad offenses may not have better options than 3.8 yards per play. Monta Ellis is an inefficient scorer, but he’s likely better at shooting than anyone else on the Bucks offense.

I’m not sure about this. Perhaps for HoFers like Curtis Martin and LT on abslutely devastating offenses, but I don’t think its unreasonable to think that, say, the 2002 Titans would have been able to improve on that 10th rating by limiting George a little. It seems to me that it’s very possible to have a top 10 offense with at bad RB that you are using too much – I mean did Atlanta really have to give the ball to Michael Turner 222 times last year?

Reply

Chase Stuart September 10, 2013 at 9:37 am

Yeah, in any one season, I think you’re correct.

But in some ways, your examples prove my point. Atlanta moved on from Turner after the season. The Titans moved on from George after ’03. Teams do have limited resources during the season — see whoever ends up getting a bunch of carries for the Steelers this year — but that changes after the year. The fact that teams continued to make George and Martin key cogs in their offenses is a strong sign of at least one thing: the coach thought they were very good players. That, to me, holds some weight.

Reply

Leroy September 10, 2013 at 9:08 am

A team with an excellent offense might find itself in situations where burning clock is a priority over yardage. Of course, the defense would be stacked up to stop the run. This could be part of it.
I don’t understand why you just assume these coaches don’t make mistakes though. Jimmy Johnson won Super Bowls, doesn’t mean his draft chart is sensible.

Reply

Chase Stuart September 10, 2013 at 9:40 am

Yes, I think poor fourth-quarter runs is a part of this. Curtis Martin in ’98 averaged 3.0 YPC in the 4th quarter of games in ’98:

http://espn.go.com/nfl/player/splits/_/id/775/year/1998/curtis-martin

Re: coaching mistakes, that plays a part, I’m sure. But on average, I think if a team continues to feed a running back 300+ carries a year, that’s very telling.

Reply

sn0mm1s September 10, 2013 at 1:52 pm

I place YPC much higher on the list than any other single RB stat (given a threshold of carries).

I think some of the issue is coaching philosophy. How many of those RBs have coaches that are known to like the workhorse RB? It wasn’t that long ago that establishing/stopping the run was (and for some coaches/talking heads still) considered the most important ingredient in the recipe to win a football game.

Also, I think QB play is a significantly larger factor in points per drive. I would be curious to see a similar table with RBs averaging 10 or 20% above league but also have the team QB rating and team ANY/A (and have that added to this table as well).

I would be willing to bet that rating and ANY/A correlate a lot higher to points per drive. If we want to eyeball a YPC effect add those high YPC RBs to the above table along with team QB stats. Sort on QB stats and see if there is a significant difference in Points/drive when YPC is very different but QB play is similar.

Reply

Chase Stuart September 10, 2013 at 4:18 pm

No doubt that quarterbacks significantly impact the team’s OPPED. Your idea for a study is interesting; I will put it on the to-do list.

What do you think of George and Martin?

Reply

sn0mm1s September 10, 2013 at 5:05 pm

It depends. As RBs they were obviously good compared to the vast majority of RBs in NFL history – but I don’t consider them HOFers. I think that I have mentioned before that both of their career YPC were below league average. As far as their teams’ successes they generally had solid, but unspectacular QB play paired with above average defenses. To me, the most impressive thing about Martin is the fact that he rarely fumbled.

To follow up on my previous post here are the ratings and any/a for the top 7 RBs on the list:
2008 SD – 105.5, 8.04
1998 NYJ – 94.0, 7.02
1995 GB – 97.6, 7.07
2003 TN – 100.3, 7.36
1997 MIA – 79.2, 5.97
2006 CIN – 94.1, 6.81
2002 TN – 83.9, 5.92

Here are the bottom 7 (ranked highest to lowest)
2000 ATL – 66.8, 3.76
1995 TB – 60.3, 3.83
1985 PHI – 65.0, 4.35
1991 IND – 70.2, 3.62
2008 CLE – 54.8, 3.32
2006 CLE – 69.8, 3.67
2004 WA – 70.0, 4.03

Reply

Red September 10, 2013 at 9:54 pm

According to a Brian Burke study, NFL coaches are trying to maximize Success Rate in the running game, not YPC or total yards. There are variations of how exactly to measure SR, but Burke concluded that rushing in the NFL is used for situational football, compared to passing which is centered around scoring points. In other words, the best rushing attack is one that can consistently convert 3rd-and-short, run out the clock in the 4th quarter, or gouge the defense when they’re expecting a pass. Success rate has very little correlation with either YPC or total rush yards.

I think Burke’s study affirms your stance that a high number of carries is a better measure of skill than YPC. If a coach continues to give carries to the same RB, he must be doing well in regard to whatever that coach considers important to the team’s success. The definition of “success” may change from one coach to another, given his style of offense and the circumstances surrounding his team.

One thing’s for sure: rushing is becoming less and less important within the framework of the modern offense, so the value of 80′s and 90′s RB’s was probably higher than almost anyone in today’s NFL. In week 1, teams averaged 23 PPG and 347 YPG, despite a meager 3.6 YPC. Sometimes I can’t help but think the only real purpose of rushing is to set up play action passing.

Reply

Chase Stuart September 11, 2013 at 8:53 am

Yes, I’m quite familiar with Burke’s study http://www.advancednflstats.com/2010/10/how-coaches-think-run-success-rate.html ;)

I agree that if we had success rate going back historically, that would be great info. Actually, Aaron Schatz might have that for the entirety of the careers of George and Martin, so I should ping him.

Reply

Red September 12, 2013 at 12:09 am

DVOA now goes back to 1989, so Aaron would definitely have the success rates calculated for Martin and George (and Barry Sanders, which would make an interesting comparison). Unfortunately, FO doesn’t have career tables or ranking for players, so you’d have to go through each season page and then calculate a weighted average based on number of carries per season. Hopefully Aaron will be nice enough to give you special treatment and bypass that laborious process :)

On a related note, I think it would be fascinating to see if the optimum pass/run ratio has changed in recent years with the increase in passing efficiency. I’ll admit I have no clue how to calculate such a thing, but I bet you do!

Reply

sn0mm1s September 12, 2013 at 12:10 pm

Wasn’t hard to do with Excel. I do have a lot of issues with FO’s success rates/DYAR/DVOA but here are their numbers.

Eddie George 43%
Curtis Martin 46%
Barry Sanders 47%

Reply

Elias September 11, 2013 at 5:41 pm

Interesting data. I would be interested to know the percentage of total offensive plays that went to each RB, and not just the percentage of rushing plays.

I also think that voluminous scoring would lead to more conservative play calling, and thus more rushing. The more predictable your play calling the more likely it would be that the defense would stack the box, and the YPC would come down. It is not just running out the clock, but protecting the football as well. As was mentioned above, Curtis Martin could be relied upon to maintain possession because he rarely fumbled, so if I am a 12 and 1 team, I have had a bunch of leads to protect.

Conversely, if I am a bad team that cannot throw the ball (like the Jets with Sanchez), and teams stack the box, I could be afraid of turnovers caused by blitzing defenses. Perhaps calling a game just to keep yourself competitive causes a drop in scoring efficiency secondary to a drop in YPC. Of course the quarterback matters- but in the NFL, teams seem to prescribe to identities. Some teams will philosophically run more often because they cannot philosophically throw more often and vice versa.

Perhaps this is just evidence of GMs and coaches sticking to preordained play calling despite the progression of the game. Hence the skins riding Clinton Portis all the way to the least scoring efficient offense in the league. They were probably just realizing that Patrick Ramsey was never going to be their solution at QB. 2008 was also Tomlinson’s first year of real regression after posting 1400 and 1800 yard seasons the two years prior. 1100 rushing yards was the lowest sum of his career at that point, but teams were probably still playing him like the Tomlinson of 2006.

Reply

Leave a Comment

You can use these HTML tags and attributes: <strong> <em> <pre> <code> <a href="" title="">

{ 4 trackbacks }