Re-Evaluating Running Back Metrics

Re-Evaluating Running Back Metrics
Re-Evaluating Running Back Metrics
Photo: USA Today Sports Images

by Bill Barnwell

In the past, Football Outsiders has come under criticism from fans who think our core metrics -- DVOA and DYAR -- don't take into account the explosive nature of players whose astounding athleticism and ability to make defenders miss goes beyond the scope of statistics. While we haven't discounted the impact that those sort of players can have on an isolated play in a game, it's been difficult for both writers and readers both to reconcile themselves with the idea that players considered stars by the media and valuable assets by their teams are drastically overrated because of their "boom or bust" style.

Until now.

When Mike Tanier and I were discussing the likelihood of Reggie Bush accruing 45 rushing DYAR in the 2008 season, I knew it was an easy bet. I like to argue with Mike about something every August just to stretch my debating muscles before the season. It's an important part of FO training camp, right up there with getting the new Week In Quotes guy to jump on a table and recite famous Herm lines.

I was debating what to set the bar for Bush's rushing DYAR and counting my chickens when it hit me. I thought to myself, "The only way Bush even hits positive DYAR is if we start including..."

That was August 2008; seven months have passed, and with them, I've been hard at work putting together our revolutionary new statistic:

Horizontal Yards.

There have been plenty of long nights watching game tape, trying to figure out exactly what the baselines should be for things like stutter-steps above replacement. I've sent hundreds of e-mails to groundskeepers around this great nation trying to convince them that they need to draw horizontal yard-lines across the field without a single reply. I've spoken with several upstart football leagues and tried to convince them that they should consider the 6.124 yards between hashmarks the "Indie First Down" and allow teams a chance to at least replay the down if they can maneuver the ball from one hash to the other. I faxed each team a warning to prepare their scoreboards, adding space for Horizontal Yards and cautioning them to change their "yards" indicators to "Vertical Yards." Strangely, this resulted in the Raiders making me a job offer. I've even pitched Aaron on going back to a version of our old slogan: "Football To And Around The Hashmarks" is catchy, but I guess it doesn't fit well on a t-shirt.

More importantly, though, I've gone through every NFL game from the 2008 season, watching brilliant backs like Bush make initial tacklers miss before being brought down by the second wave of tacklers that caught up to him while he was making the initial tackler miss. I've seen Tim Hightower bounce a play designed to be off-tackle out to the industrial fans and Gatorade table. I've gazed in admiration as Dominic Rhodes ran the sweep like he was under the aegis of a Nintendo controller, operating on two axes. Finally, I thought to myself night after night, I understand how Adrian Peterson tap-dancing in the backfield like he was waiting for a storm to pass begat a successful NFL rusher. I scoffed at the failures of the Large Hadron Collider; by the time they'd found Higgs Boson, I'd know how many horizontal yards Mark Higgs and Cap Boso had accumulated.

As I finally compiled the final totals and began the calculations for horizontal yards last week, I began to shudder with the thought of what might be on the other side. Would I blow the lid off conventional wisdom, proving that DVOA and DYAR were right about boom-or-bust runners? Would horizontal yards prove to be an excellent indicator of performance that Football Outsiders had failed to incorporate into its projections and metrics?

The answer is yes.

Rank Team Horizontal Yards Rank Team Horizontal Yards
1 TEN 1198 17 BAL 747
2 MIN 1166 18 CHI 732
3 DET 1166 19 BUF 689
4 STL 1028 20 NO 673
5 IND 1002 21 DAL 679
6 SEA 996 22 PHI 635
7 NYG 992 23 DEN 583
8 ATL 984 24 HOU 559
9 GB 928 25 WAS 555
10 PIT 897 26 JAC 554
11 MIA 871 27 SF 492
12 CAR 869 28 NYJ 473
13 CIN 795 29 SD 468
14 KC 795 30 CLE 434
15 OAK 758 31 TB 411
16 ARI 744 32 NE 386

As you can see, horizontal yards bears a significant relationship to success by virtue of the fact that its top two teams -- Tennessee and Minnesota -- both made the playoffs. Included in the top ten are playoff teams like Indianapolis, the New York Giants, Atlanta, and the Super Bowl Champion Pittsburgh Steelers, who would have ranked even higher had Willie Parker been healthy for the entire season.

Of course, I know Willie Parker and horizontal yards seem like a match made in statistically significant heaven, so let's introduce you to the individual stats. For all rushing plays and completed passes, I've calculated both POV (percent over adjusted vertical yards), which is my percentage-based horizontal stat, and POS (parallel yards above adjusted standard), which is my cumulative stat.

We'll start with the backs. Here are the top 15 backs in the league with 30 carries or more by POV:

Player Team POV POS
44-A.Bradshaw NYG 17.9% 194
38-D.Rhodes IND 17.1% 492
23-C.Perry CIN 15.7% 292
30-A.Pittman STL 15.2% 318
34-K.Smith DET 15.2% 877
25-R.Bush NO 15.1% 313
27-K.Jones CHI 14.7% 116
28-C.Johnson TEN 14.3% 827
23-A.Hall DEN 14.3% 57
21-K.Smith KC 14.3% 116
20-M.Morris SEA 14.2% 401
42-T.Duckett SEA 13.6% 156
34-T.Hightower ARI 13.3% 454
32-R.Johnson DET 13.2% 265

I'm very excited that Ahmad Bradshaw of my very own New York Giants was the leader in POV for the 2008 season. Because of his speed and sideline-to-sideline ability, the cumulative effect of chasing Bradshaw around the backfield tires out defenses and creates opportunities for Brandon Jacobs.

Although Adrian Peterson had 838 horizontal yards on the season, he only registered a POV of 6.3% because he actually ran for about 890 adjusted vertical yards on the year. We'll have to see if Peterson can turn more of those vertical yards into horizontal ones next year.

One sleeper for 2009: 49ers tight end Delanie Walker. He only had two carries on the year, and they combined for -13 vertical yards, but Walker had an astounding 24 horizontal yards on the two plays, producing a ratio far above anyone else in the league. HAROLD, our horizontal fantasy football projection system (named after former Bengals back Harold Green), is seeing big things for him in the San Francisco offense come 2009.

The top 15 in POV on receiving plays (minimum: 50 targets) reveals the fruits of having a veteran presence on your roster:

Player Team POV POS
19-I.Hilliard TB 29.3% 304
82-J.Reed BUF 26.9% 342
86-D.Lee GB 26.0% 321
12-S.Smith NYG 25.6% 346
82-G.Olsen CHI 25.6% 474
84-T.Houshmandzadeh CIN 24.5% 697
83-H.Miller PIT 24.2% 396
47-C.Cooley WAS 23.9% 626
44-D.Clark IND 23.8% 590
88-D.Clark CHI 23.3% 414
87-J.Nelson GB 23.1% 207
23-M.Lynch BUF 23.1% 620
11-Re.Williams JAC 23.0% 332
88-J.Shockey NO 22.9% 423
21-M.Moore PIT 22.6% 353

Ike Hilliard's ability to stretch the field horizontally makes him an incredibly valuable player, even as he ages; that's something astute fans of the game already know from his accomplished past, though. Upon Hilliard's arrival in New York out of Florida in 1996, the Giants immediately went from 6-10 to 10-5-1 and into the ascendancy in the NFC East. After making it to the Super Bowl in 2000, the Giants began to phase Hilliard out as a starter in 2001, resulting in a 7-9 record; the year after, they returned him to the starting lineup and went 10-6, a clear sign of his importance to the team.

After the Giants went 6-10 in 2004, they magnanimously let Hilliard go to Tampa Bay; the Buccaneers went from 5-11 without Hilliard to 11-5 with him. The year after, Tampa Bay took him out of the starting lineup and immediately went 4-12. Duh. Hilliard became a starter again in 2007, and the result was a 9-7 season and an NFC South crown. Ike Hilliard, my friends, is a winner, and horizontal yards proves it.

Of course, the only way we might ever get to know the true value of every player -- his real worth to his team, the absolute quantifiable measure of his performance and skills -- is if we could somehow combine vertical yards and horizontal yards to create...

I've got some work to do.


71 comments, Last at 01 Apr 2010, 11:48pm

1 Re: Re-Evaluating Running Back Metrics

"Upon Hilliard's arrival in New York out of Florida in 1996, the Giants immediately went from 6-10 to 10-5-1 and into the ascendancy in the NFC East." was the last straw for me.

And "POV" is hilarious!

4 Re: Re-Evaluating Running Back Metrics

Also, I found it funny that a stat created just for Reggie Bush doesn't even let him be #1. That dude just can't get any love.


I think this is probably one of maybe 3 stats that New England finishes last in. That and "Playoff appearances by 11-5 teams."

19 Re: Re-Evaluating Running Back Metrics

That's because Bush was better running between the tackles than out wide. The best plays for him are draws and counters, something that Sean Payton realized just before he injured his knee this season.

Sports talk radio and sports message boards are the killing fields of intellectual discourse.

7 Re: Re-Evaluating Running Back Metrics

If you're going to combine horizontal and vertical yards, clearly that makes a hypotenuse.

Perhaps Defence/Opponent adjusted Hypotenusal yards, more handily refered to DOH, would be the ideal metric for future evaluations?

9 Re: Re-Evaluating Running Back Metrics

Maybe I'm just not very bright, but I'm having a difficult time wrapping my mind around this. Why, in football terms, are horizontal yards a good thing?

For all the 'good' teams on the list, I see teams like Detroit, Saint Louis, Seattle, and an Indy team with one of the worst run offenses in the game in the top 6.

Why is it good that Dom Rhodes was great at gaining horizontal yards? He was only mediocre at gaining vertical ones, and those are the ones that help you get first downs. Ike Hilliard anecdotes aside, what evidence is there that a player who gains horizontal yards is helping his team and not just running around?

I know I'm missing something here, can anyone help me out?

43 Re: Re-Evaluating Running Back Metrics

Mister Zombie,

You know... you really let all the 18to88 fans out there down today.... I just don't know what to say. Well, might as well get back to inventing oxygen.

After about a paragraph of the usual humor (the new TWIQ guy reciting Herm) I started to think WTF?!?! and bounced to the bottom hoping for an AHA! moment. When there was one, I had to actually read it to enjoy all the little aha's!

50 Re: Re-Evaluating Running Back Metrics

In my defense, I didn't get a lot of sleep last night. I knew I would feel like an idiot if expressed my concern, because there had to be a simple explanation. My faith in human nature is non-existent and I honestly feared that FO had gone all CHFF screwy on us.

I can't express my relief in words.

14 Re: Re-Evaluating Running Back Metrics


Bill, you have discovered the previously unknown value of all of those University of Florida wide receivers. I'm sure it will turn out that Reche Caldwell, Taylor Jacobs and CHad Jackson weren't really busts, just that their true value hadn't been measured. Maybe Travis Taylor can get another shot now. It's a little late for Jacquez Green.

18 Re: Re-Evaluating Running Back Metrics

I like Bill's Dr. Z-esque quest to get attention from the league and the teams for this long-ignored statistical need. Of course, given some team's propensity for employing this kind of RB, I think many of them already worship at the horizontal altar.

20 Re: Re-Evaluating Running Back Metrics

Boom/Bust Running backs are over valued in relation to equal cloud of dust RB's, but a "boom/bust" young QB who throws 3Td's with 3 interceptions is worth more than the "game manager" who throws 0Tds and 0 ints.

A RB taking a handoff and running to the los and then sideways to the sideline and getting 0 yards is better than running right into the line and being tackled for 0 yards, but listing the backs with the most horizontal yards " as if this is a great league leading stat" is incorrect.

Ike Hillard is a good possession WR than can read defenses and make catches in a WCO, but you went too far with the Giants stuff. Anytime you upgrade from Dave Brown, Kent Grahmn and Dannel Kannel to Kerry Collins (in what was then a weak NFC East), you are going to improve regardless.

and what's with this " my giants". Didn't you pick them to finish in last place in 2007? You sound like a huge Giants fan to me.

21 Re: Re-Evaluating Running Back Metrics

If you want the "king" of the horizontal offense, look no further than Jim Zorn and Jason Campbell taking snaps and throwing the ball sideways to his WR's in smoke screen fashion.

Not something I'd be proud of.

22 Re: Re-Evaluating Running Back Metrics

Has POV been calculated for punt returns? I'm guessing Antwan Randle El tops the list, after all, his running sideline to sideline on every punt tires out the opposing special teams, which proved to be the deciding factor in many of the Redskins games last year. I think.

I found Google to be pretty lame today, BTW.

23 Re: Re-Evaluating Running Back Metrics

"I faxed each team a warning to prepare their scoreboards, adding space for Horizontal Yards and cautioning them to change their "yards" indicators to "Vertical Yards." Strangely, this resulted in the Raiders making me a job offer."

This was fantastic.

29 Re: Re-Evaluating Running Back Metrics

I new that Chris wouldn't get the true greatness of this tremendous article. Sometimes you have to take your head out of your anus to truly understand the intricacies of greatness.

Good article Bill. Can you take another entire year to write something just like this for next season?

31 Well Said

Very nicely done Bill. Kudos to you sir.

32 Re: Re-Evaluating Running Back Metrics

Sigh. Philly at #22? You'd think Westbrook would be better at this sort of thing. No wonder we can't win a Super Bowl.

(Nice job, Bill. I read all the way to the first instance of the phrase "Horizontal Yards" before I got it (which, apparently, some people couldn't do even by reading the comments)).

34 Re: Re-Evaluating Running Back Metrics

So I'm assuming the numbers are made up, right?

I'm asking about the slim slim chance the numbers are legit, because I Would love to throw Dom Rhodes' massive number of Horizonal yards at all the Colts fans that are hating on Addai for "dancing behind the line" and not putting up the same production over the last 1.5 seasons as he did for his 1.5 seasons (which is a stupid argument because the biggest difference between Addai's first 24 games and his 2nd 24 was OL health.

2006+1st half of ’07
14 starts missed on the OL (1 by Diem, 2 by Ugoh, 0 by Saturday, 0 by Scott, 0 by Glenn, 11 by Lilja)

2nd half of ’07 + 2008
36 starts missed on the OL (4 by Saturday, 16 by Lilja, 3 by Pollak, 6 by Diem, 7 by Ugoh)

45 Re: Re-Evaluating Running Back Metrics

I assume all h-yards are invented--there's no real way to count them and no point, of course. Hours and hours of tape for what?

I love the Jaddai Master and wear his jersey, but the contrast with Rhodes is similar in my mind to the Edge/Rhodes contrast. Very similar. Dom is a hammer of a runner, not too subtle, not too patient (though that has improved over the years), fairly shifty. But a nice weapon to have. Edge and Addai are both more like swords--more subtle, flexible, patient. And Edge was insanely, brutally, crushingly patient behind his blockers and his consistency scores bore that out--he almost never lost yardage and almost never broke big ones, but was always a 3-6 yard guy and much more valuable. Addai is more like Edge IMO. But to do that and succeed you do need some decent blocking, unlike last year.

Look at us, having a serious discussion today! What an ass I am. Okay, back to my time machine: I'm curious to see if that play at Ford's Theater was really worth killing Lincoln over. Must have bene some bad play.

36 Re: Re-Evaluating Running Back Metrics

Bill says that he is a Giants fan, but the year they finished the season SB champs, he said they'd finish in last place. Some fan to me.

April Fools jokes aside, Having a RB run sideways and wear down defenses ( chasing LB's and D-lineman) is better than having a RB slam into a line. That was a strategy the Broncos had in the SB with the Packers where they wanted to wear down a stout Gilbert Brown in the middle of the defense who was gassed at the end of the game.

Sort of like how the baseball guys would factor in the # of pitches a batter takes, because more pitches = worn down starter which means it is more likely to bring in garbage relief pitchers.

For example, chasing Barry Sanders around the field for a 1 yard loss is worth more than Ron Dayne taking a handoff, running forward 4 yards and being tackled.

Parker - did you "know" that Chris is a dumb mean guy or did you "new" it?

51 Re: Re-Evaluating Running Back Metrics

Bill says that he is a Giants fan, but the year they finished the season SB champs, he said they'd finish in last place. Some fan to me.

1) Believing a team you're the fan of will finish in last place (in the division!) is the sign of a healthy fan. Believing your team will always finish in first place is the sign of a delusional fan (see: Raiderjoe).

2) The fact that the Giants run in 2007 was improbable doesn't cheapen it. It makes it that much more amazing. I'd bet that the Giants fans who enjoyed 2007 the most are the ones who thought the least of the team in the beginning.

61 Re: Re-Evaluating Running Back Metrics

"For example, chasing Barry Sanders around the field for a 1 yard loss is worth more than Ron Dayne taking a handoff, running forward 4 yards and being tackled".

That's very true...until the next play, when you have 2nd (or 3rd) and 6 instead of 2nd and 11.

38 Re: Re-Evaluating Running Back Metrics

I mean to say Dayne taking a handoff, running forward 4 yards, and also being tackled for a 1 yard loss.

Denver ran a lot of outside stuff early at GB in the SB to specifically wear down Gilbert Brown and it worked. The guy was almost impossible to run on inside, by the end of the game he wasn't so. Even though Mike Holmgren let Denver score that last TD...

Just trying to bring up some glory for some pissed off Jay Cutler/Denver fans.

41 Re: Re-Evaluating Running Back Metrics

Always sucks to be the last one in on the joke. I k/now I don't have to resort to grammer checking to have a relevant point.

46 Re: Re-Evaluating Running Back Metrics

My inital comment WAS saying that having a Barry Sanders type back rack up a 1 yard loss on 20 yards worth of run and chase is better than a Ron Dayne type having an equal run.

Denver in the Super Bowl running horizontally is actually a good example of how it does have value, when you have a 350 pound worthless Gilbert Brown at the end of the game who was sick and tired of chasing Terrel Davis towards the sidelines.

Maybe you'd like to call Ted Cottrell and Norv Turner idiots before you start asking what a 3 technique defensive lineman is too....

47 Re: Re-Evaluating Running Back Metrics

Great job in making the joke more legit with who you put on the list. The receivers list really reflected what you would call "horizontal receivers". We should incorporate this term next season when discussing WR who make worthless 6 yard catches on 3rd and 14.

58 Re: Re-Evaluating Running Back Metrics

Pat- but Barnwell didn't pick the Giants to finish in last place in the NFC, the genius known as Bill Barnwell picked the Giants to finish in last place in the NFL. Big difference.

59 Re: Re-Evaluating Running Back Metrics

Appreciate the kind words. The numbers are based on real stats with some multipliers and random numbers thrown in to keep people off the trail. I don't remember exactly what since I ran them a couple of months ago, but I believe that they're the following:

- Team stats are the number of times a team ran the ball for negative yardage; yes, Tennessee and Minnesota were 1-2.

- For RBs, POV is the percentage of each player's carries that went for negative yardage (although that doesn't seem right). POS is the yards they gained on those plays with a multiplier and random number added.

- I don't remember at all what the stats were for receivers. Maybe number of times they caught a pass and got negative YAC or something.

62 Re: Re-Evaluating Running Back Metrics

The funny thing is, that despite this being an April Fools article, they *should* re-evaluate their metrics. According to FO, the stats we would expect from a replacement level RB on the Titans is 1663 rushing in place of Chris Johnson. That would equate to a top 40 all time rushing season. I won't even go into some of the weirdness regarding Barry Sanders, Terrell Davis, and Emmitt Smith.

64 Re: Re-Evaluating Running Back Metrics

Actually, you are wrong about that. Yards - DYARs is more or less the expected number of yards of a replacement player given the same carries in the same situations. In fact here is an email that I received from Aaron.

My question:

I was hoping to get a little clarification on DYAR.

If a player rushes for 1000 yards, 10 TD, 2 fumbles on 250 carries and has a DYAR of 100. Does this mean that a replacement level RB would roughly be expected to rush for 900 yards, 10 TD, 2 fumbles on those same 250 carries?


His answer:

The measurements aren't really as exact as that, but yes, if those 250
carries came in the same situations against the same defenses, that
would be the expectation.

65 Re: Re-Evaluating Running Back Metrics

Here is the flaw in your reasoning.

A replacement level running back would be expected to get those yards if he was given the ball in the exact same situations as Chris Johnson. However, since a replacement level back would be so much worse, those situations wouldn't exist. Thus he would end up with fewer yards. I hope this helps.

68 Re: Re-Evaluating Running Back Metrics

Right, which means if we replace CJ with a replacement player you would expect X result. What CJ did previously has absolutely *no* bearing on what is addressed in the email. Of course CJ does better than a replacement otherwise he wouldn't have a positive DYAR, that isn't what is being discussed. There is no "flaw" in what I am saying. Your reasoning is flawed because you aren't viewing each carry as an individual event - as the FO system does.

69 Re: Re-Evaluating Running Back Metrics

I don't think I'm explaining myself very well, I'll try an example.

1st and 10 from the 50
Replacement back gets 4 yards
CJ gets 6

The next down is now either 2nd and 4 or 2nd and 6, which would have different baselines for replacement level. CJ created a situation for himself where he is expected to have more success than the replacement back is.

71 Re: Re-Evaluating Running Back Metrics

I understood exactly what you said from the beginning - but again that doesn't affect anything that I have posted. It doesn't matter if CJ runs for 6 yards and the replacement back runs for 4. The next carry is from where CJ ended his run when dealing with DYAR.

Basically, FO stats tell us that we would expect a replacement back to run for ~1660 yards if you subbed out CJ for each of CJ's runs this year. To me, this seems ridiculous. FO's model does not handle big plays and big play players well.