Writers of Pro Football Prospectus 2008

Most Recent FO Features

CoxMar15.jpg

» Varsity Numbers: Honing in

Bill Connelly again looks at which college football teams the F/+ ratings are sure about, and which teams remain a mystery (led by Appalachian State).

15 Aug 2013

FO 10th Anniversary: Best Running Backs

by Danny Tuccitto

In our 10th anniversary conversation about DVOA and DYAR records, it's time to start talking about running backs. If you haven't read the first two parts in the series, here are the links:

Mr. Bettman was unavailable to proctor, so there's no pop quiz today. However, there is one question I would like you to answer before we start. Below is a list of six DVOA-era running backs that are not yet in the Hall of Fame, four of which have already failed once in balloting. Please rank them from most to least deserving of enshrinement, and set aside your answer for reconsideration later.

  • Tiki Barber
  • Jerome Bettis
  • Terrell Davis
  • Priest Holmes
  • Edgerrin James
  • LaDainian Tomlinson

Last week, FO newcomer Scott Kacsmar made a strong case for why Davis deserves to be in the Hall of Fame, at least compared to Bettis and Roger Craig. When I set out to put today's column together, I was not planning on it being a re-evaluation of or rebuttal to Scott's argument, nor a specific argument of my own in favor of a different back. However, after gathering all the stats, it ended up going in that direction. As you'll see, there are two running backs on the above list that both DVOA and (especially) DYAR deem to be no doubters and another that should be getting stronger consideration among voters than he has so far: One of these three may or may not be Terrell Davis. And besides, the tables are so dominated by a certain back already in the Hall, that a narrative other than "man, Marshall Faulk was really good" seemed necessary to keep your attention.

To wit, here are the best rush DVOA seasons by a running back since 1991:

Best Rush DVOA, Season, 1991-2012
(min. 100 rushes)
Year Player Team Runs Yards TD DVOA
2000 Marshall Faulk STL 253 1,359 18 36.6%
2010 Jamaal Charles KC 231 1,466 5 32.9%
1999 Stephen Davis WAS 290 1,407 17 32.4%
2003 Onterrio Smith MIN 107 579 5 30.9%
1995 Charlie Garner PHI 108 588 6 30.8%
2011 Pierre Thomas NO 110 562 5 30.0%
1993 Gary Brown HOIL 195 1,002 6 29.9%
2002 Priest Holmes KC 313 1,615 21 29.3%
2006 Marion Barber DAL 135 654 14 29.0%
2004 Larry Johnson KC 120 581 9 29.0%
Best Rush DVOA, Season, 1991-2012
(min. 200 rushes)
Year Player Team Runs Yards TD DVOA
2000 Marshall Faulk STL 253 1,359 18 36.6%
2010 Jamaal Charles KC 231 1,466 5 32.9%
1999 Stephen Davis WAS 290 1,407 17 32.4%
2002 Priest Holmes KC 313 1,615 21 29.3%
2012 C.J. Spiller BUF 207 1,249 6 27.6%
2010 BenJarvus Green-Ellis NE 229 1,008 13 26.7%
1998 Terrell Davis DEN 392 2,008 21 26.5%
2002 Clinton Portis DEN 273 1,508 15 25.5%
1999 Marshall Faulk STL 253 1,381 7 25.4%
1997 Barry Sanders DET 335 2,053 11 25.3%

Man, Marshall Faulk was really good!

If we use the standard Football Outsiders minimum of 100 carries, our list of the top rushing DVOA seasons runs into the same usage problem we had with quarterbacks. So we also decided to run a table with the top 10 DVOA seasons given a minimum of 200 carries, although this requires going down to No. 22 in the rankings based on the 100-carry minimum.

As you proceed further down the rankings, the legit high performers continue to separate themselves from the stems and seeds even more. No running back produced two seasons in the top 10, and only two backs have multiple appearances in the top 25 (Faulk and Holmes), but 11 backs repeat in the top 50 and those 11 account for 25 of the 50 seasons: Charlie Garner, Pierre Thomas, Larry Johnson, Brian Westbrook, Emmitt Smith, Faulk, Davis, Sanders, Holmes, Bettis, and Tomlinson. And who is the cream of this cash crop? Well, although Sanders will probably join them when we finish adding 1990 to our database, as of right now only Faulk and Holmes had at least three seasons in the top 50 DVOAs. One of that pair actually had four, and his name is not Marshall Faulk.

It probably doesn't surprise you, however, that only two of the top 10 seasons occurred in the past six years (also only six of the top 25 and nine of the top 50), and that they came from players who have constantly left fans asking, "Why don't they run him more often?" Regardless of whether or not coaches are consciously driving it, offense has shifted more and more towards a preference for efficiency over volume in the running game. And to me, that's a tacit vindication -- and admittedly self-serving observation -- of football analytics, which has been saying for the past 10 years that teams need not establish the run. Todd Haley may have lost everything in Kansas City, but you never know. One day, he might find a double-die Denver mint penny underneath one of Pittsburgh's team buses.

With that idea in mind -- volume vs. efficiency, not Todd Haley qua panhandler -- let's see if any of the best rush DVOA seasons make an encore when we switch to DYAR. The top 10 rush DYAR seasons are displayed below, along with those for receiving DYAR:

Best Rush DYAR, Season, 1991-2012
(min. 100 Rushes)
Year Player Team Runs Yards TD DYAR
1998 Terrell Davis DEN 392 2,008 21 602
1999 Stephen Davis WAS 290 1,407 17 526
1997 Terrell Davis DEN 369 1,743 15 526
1995 Emmitt Smith DAL 375 1,770 25 505
2000 Marshall Faulk STL 253 1,359 18 501
2002 Priest Holmes KC 313 1,615 21 497
2005 Larry Johnson KC 335 1,741 20 488
2003 Priest Holmes KC 320 1,420 27 485
1994 Emmitt Smith DAL 368 1,484 21 461
2006 LaDainian Tomlinson SD 347 1,813 28 460
Best Receiving DYAR, Season, 1991-2012
(min. 25 Passes)
Year Player Team Recs Yards TD DYAR
1999 Marshall Faulk STL 87 1,048 5 418
1998 Marshall Faulk IND 86 908 4 417
2002 Charlie Garner OAK 91 941 4 376
2001 Marshall Faulk STL 83 765 9 357
1995 Larry Centers ARI 101 962 2 354
2000 Marshall Faulk STL 81 830 8 342
1991 Thurman Thomas BUF 62 631 5 290
1993 Terry Kirby MIA 75 874 3 289
1996 Larry Centers ARI 99 766 7 288
2011 Darren Sproles NO 86 710 7 279

(Ed. Note: Receiving numbers may differ slightly from numbers on our stats pages because of a recently fixed error regarding receiver numbers on interceptions; we'll have all our stats pages fixed soon.)

As far as the receiving DYAR table goes, there's really only one thing to say: Man, Marshall Faulk was really good!

Game Rewind: Relive every NFL moment…subscribe to Game Rewind.

Of the four top 10 rush DVOA seasons featuring 200 or more carries, three carry over to the list of top 10 rush DYARs, with Jamaal Charles' 2010 campaign being the only omission. Charles 6.38 yards per rush represents a level of efficiency that remains the highest since 1991 (and second-highest since 1978), but a 2.2 percent touchdown rate (ranked 599th out of 964 qualifying seasons) drops his total value in 2010 to 20th overall.

Getting down to brass tacks, though, only eight-tenths of a DYAR keep Terrell Davis from having the two most valuable rushing seasons since 1991. More than any stat you will see today, that one distills the argument for Davis' Hall of Fame candidacy down to its most justifiable premise: All careers have peaks, but no peak was higher than his 1,128 rush DYAR in 1997 and 1998. Even the best of Emmitt Smith falls nearly 200 DYAR short. If we add in Faulk's 1999 season, his two-year peak ends up 300 short of Davis'. At this point, it doesn't take Fred Edelstein to correctly predict the next sentence in this paragraph. There's one other running back in the table besides Davis whose two-year peak produced more rushing value than the aforementioned Hall of Famers, and his name is not LaDainian Tomlinson.

But since this is Davis' high-water mark in the piece, it's worth citing a couple of stats that should go a long way towards closing the casket on a long-standing argument against him. As Scott revealed in his piece last week, Davis had 1,829 rush DYAR during a seven-year career spent entirely under the tutelage of Mike Shanahan. (That's not one of the fatal stats I'm referring to.) In three seasons with Shanahan, Clinton Portis had 783 rush DYAR (that's not it either), but we'll see later that Sheriff Gonna Getcha was plenty great without help from Boss Hogg. The other 40 running backs who ever carried the ball during Shanahan's stints in Denver or Washington amassed 2,721 rush DYAR combined. (That's the one!) That means Davis was nearly as valuable in seven seasons as the 85 combined seasons of every other Shanahan-era running back for the Broncos and [Redskins] not named Clinton Portis.

If that doesn't do it for you, here's another way of looking at things. During Shanahan's tenure with the Broncos (1995-2008), the cumulative 4,862 rush DYAR by running backs ranked No. 1 over that 14-year span. Remove Davis from the total, and they drop to third. Remove Davis and Portis, and they drop out of the top 10 entirely. Obviously, two numbers don't represent a nail in the coffin, but continuing to argue the Shanahan angle starts to resemble death throes when you consider that two great runners account for such a large portion of all the value that Shanahan backs have ever produced.

That ends the brief pro-Davis portion of this column. We now return to regular programming. Here are the top 10 games since 1991 according to total DYAR (box scores linked up in the "Week" column; asterisks mean the team lost):

Best Total DYAR, Game, 1991-2012
Year Week Player Team Runs Run
Yards
Run
TD
Recs Rec
Yards
Rec
TD
DYAR
2002 12 Priest Holmes KC 23 197 2 7 110 1 152
2006 12 Joseph Addai IND 24 171 4 3 37 0 145
1997 15 Corey Dillon CIN 39 246 4 2 30 0 135
2005 3 LaDainian Tomlinson SD 21 192 3 7 28 0 134
1991 13 Barry Sanders DET 23 220 4 4 31 0 131
1998 13* Marshall Faulk IND 17 192 1 8 75 1 131
2000 17 Marshall Faulk STL 32 220 2 8 41 1 123
2007 3 Brian Westbrook PHI 14 110 2 5 111 1 117
1991 1 Thurman Thomas BUF 25 165 1 9 103 1 114
2000 7 Marshall Faulk STL 25 108 1 10 78 0 114

Man, Marshall Faulk was really good!

Admittedly, this list looks like it's skewed a bit towards running backs that were valuable as receivers; glorified Roger Craigs, if you will. And indeed, the games ranked sixth through tenth would not qualify for the top 10 according to rush DYAR. However, there's a danger in overestimating how much the nature of a running back's skill set matters here. First, the flip side of the previous statement is that the top five do rank among the 10 best games of rush DYAR, and none of those five backs were slouches when it came to receiving ability. Even Corey Dillon (46th in receiving DYAR since 1991) and Sanders (100th) were above-replacement pass catchers during their careers. Second, it's also true that the backs in the bottom of the table weren't replacement-level runners. I've belabored how good Faulk was, Thurman Thomas is in the Hall of Fame thanks in part to his 14th-most rushing yards of all time, and -- spoiler alert! -- Westbrook's career rush DYAR is among the top 22 since 1991.

So if we can agree that both rushing and receiving matter for running backs, then perhaps the most interesting about the table is how many of the greatest games since 1991 in terms of yards from scrimmage aren't in the top 10 of total DYAR. Chief among them is Adrian Peterson's 315-yard game in Week 9 of his rookie season, which ranks only 39th on our list (94 total DYAR). The reasons why it's not higher are five-fold: (1) It came against a 2007 Chargers run defense that finished 23rd in DVOA, (2) 145 rushing yards and a touchdown came while Minnesota was protecting a fourth-quarter lead, (3) he fumbled during that fourth quarter, (4) all that extra yardage on his long runs produced diminishing returns in terms of value, and (5) he only caught two passes for 17 yards.

Still, Peterson's performance that day resulted in the most scrimmage yards in a game since 1991, so it remains impressive no matter what total DYAR thinks. You know which running back had a game in 2002 that sits just behind Peterson's on the scrimmage-yard list and at the top of our total DYAR list? His name isn't Joseph Addai.

But before I end the suspense, and finally reveal what this running back's name actually is -- god forbid you haven't figured it out already -- below are the top 22 careers of the past 22 years according to rush DYAR and receiving DYAR:

Best Rush DYAR, Career
(Debuted 1991 or Later)
Player Years DYAR
Priest Holmes 9 2,006
LaDainian Tomlinson 11 1,831
Terrell Davis 7 1,829
Marshall Faulk 12 1,791
Jerome Bettis 13 1,722
Curtis Martin 11 1,542
Edgerrin James 11 1,478
Corey Dillon 10 1,455
Fred Taylor 13 1,362
Clinton Portis 9 1,350
Player Years DYAR
Adrian Peterson 6* 1,292
Stephen Davis 11 1,262
Ricky Watters 10 1,192
Tiki Barber 10 1,151
Brian Westbrook 9 1,134
Charlie Garner 11 1,126
Ahman Green 12 1,114
Robert Smith 8 1,073
Shaun Alexander 9 1,040
Maurice Jones-Drew 7* 972
Garrison Hearst 10 959
Terry Allen 10 892
Best Receiving DYAR, Career
(Debuted 1991 or Later)
Player Years DYAR
Marshall Faulk 12 1,909
Tiki Barber 10 1,113
Kevin Faulk 13 973
Brian Westbrook 9 969
Charlie Garner 11 941
Darren Sproles 7* 940
Ricky Watters 10 847
Amp Lee 9 832
Terry Kirby 10 817
Warrick Dunn 12 755
Player Years DYAR
Dorsey Levens 11 748
Maurice Jones-Drew 7* 715
William Henderson 12 697
Priest Holmes 9 674
Eddie George 9 636
Edgerrin James 11 593
Duce Staley 9 576
Steven Jackson 9* 543
Pierre Thomas 6* 513
Ray Rice 5* 497
Jamal Anderson 8 493
Richie Anderson 12 448

Man, Marshall Faulk was really good! Not only has he absolutely blown away the competition when it comes to receiving value, but he also ranked in the top five in rushing value. I wonder what that means for where his career ranks according to total DYAR. (/sarcasm)

So, of course, the running back I've been alluding to this entire time is Priest Holmes. Let's recap what he's accomplished so far in this column. His rush DVOA in 2002 was the eighth-best since 1991, and he's the only back with four seasons in the top 50 -- more than Marshall Faulk, Emmitt Smith, and Terrell Davis. Two of those seasons also featured top 10 run DYARs, a feat Smith and Davis matched, but Faulk didn't. His two-year peak in 2002 and 2003 was slightly lower than Davis', but higher than both Smith's and Faulk's. At his absolute best, he had a more valuable single-game performance than Davis, Smith, or Faulk ever did. Now, we find out that he had the best rushing career ever (so far), with the second-best career being closer to fifth place than it is to first. Someone, for the love of all that is holy, why was Priest Holmes not even a semifinalist in Hall of Fame voting this year?

I've waited until now to discuss the elephant in the room because the career rush DYAR table is the one to make this skeptic a true believer. I had thought that, in the grand scheme of things, Holmes' connection to Dick Vermeil made him a poor man's Faulk, and so more than the usual proportion of his value was wrapped up in receiving. I had also thought that he couldn't have been a better rusher than Tomlinson. But here I come to find out that he ranks No. 1 in career rush DYAR, well ahead of Tomlinson, and even farther ahead of rich man's Priest Holmes. For me, those are game changers.

Before proceeding any further, though, let's see the top 22 in career total DYAR according to a simple sum, a weighted sum, and an average of the back's six best seasons (asterisk means the player is still active):

Best Total DYAR, Career
(Debuted 1991 or Later)
Player Years DYAR
Marshall Faulk 12 3,700
Priest Holmes 9 2,679
Tiki Barber 10 2,264
LaDainian Tomlinson 10 2,189
Brian Westbrook 9 2,103
Edgerrin James 11 2,071
Charlie Garner 11 2,067
Ricky Watters 10 2,039
Curtis Martin 11 1,903
Terrell Davis 7 1,892
Player Years DYAR
Jerome Bettis 13 1,777
Corey Dillon 10 1,717
Fred Taylor 13 1,703
Clinton Portis 9 1,699
Maurice Jones-Drew 7* 1,687
Stephen Davis 11 1,554
Adrian Peterson 6* 1,500
Ahman Green 12 1,479
Steven Jackson 9* 1,342
Kevin Faulk 13 1,304
Garrison Hearst 10 1,284
Robert Smith 8 1,268
Best Total DYAR, Weighted Career
(Debuted 1991 or Later)
Player Years DYAR
Marshall Faulk 12 3,275
Priest Holmes 9 2,466
Tiki Barber 10 1,993
LaDainian Tomlinson 10 1,987
Edgerrin James 11 1,912
Brian Westbrook 9 1,880
Charlie Garner 11 1,837
Terrell Davis 7 1,773
Ricky Watters 10 1,742
Curtis Martin 11 1,603
Player Years DYAR
Jerome Bettis 13 1,581
Clinton Portis 9 1,568
Fred Taylor 13 1,496
Maurice Jones-Drew 7* 1,488
Corey Dillon 10 1,461
Adrian Peterson 6* 1,381
Stephen Davis 11 1,367
Ahman Green 12 1,323
Steven Jackson 9* 1,211
Garrison Hearst 10 1,211
Kevin Faulk 13 1,158
Robert Smith 8 1,154
Best Total DYAR, Six-Year Average
(Debuted 1991 or Later)
Player Years DYAR
Marshall Faulk 6 550
Priest Holmes 6 436
LaDainian Tomlinson 6 344
Edgerrin James 6 343
Tiki Barber 6 335
Brian Westbrook 6 325
Terrell Davis 6 313
Charlie Garner 6 307
Clinton Portis 6 290
Ricky Watters 6 285
Player Years DYAR
Maurice Jones-Drew 6* 274
Jerome Bettis 6 260
Adrian Peterson 6* 250
Fred Taylor 6 243
Curtis Martin 6 239
Garrison Hearst 6 230
Ahman Green 6 225
Corey Dillon 6 225
Stephen Davis 6 225
Shaun Alexander 6 221
Steven Jackson 6* 211
Pierre Thomas 6* 211

First thing's first. At this point in the column, we're all aware of how great Faulk was, so it shouldn't surprise anyone that he's the No. 1 running back of the past 22 years no matter how you measure it. That said, the margin between Faulk and Holmes in the table on the left is even more impressive than the lead Peyton Manning has over Tom Brady in total career DYAR for quarterbacks. In both cases, the leader's total is 38 percent higher than that of the runner-up. However, in Faulk's case, No. 2 isn't still active, so he literally doesn't stand a chance of closing the gap.

Furthermore, when you consider the pace that Maurice Jones-Drew and Peterson would have to maintain, how much longer they would have to maintain it, and the NFL's aforementioned trend towards efficiency over volume in the running game, I think Faulk's total DYAR record will be broken by a back who started his career before 1991 rather than by a back who will start his career in the future. Who might that older back be? Well, looking at the standings after removing our 1991 debut restriction, I doubt it will be any of the following Hall of Famers:

  • Emmitt Smith: 2,636 total DYAR, one season played before 1991.
  • Barry Sanders: 1,923 total DYAR, two seasons played before 1991.
  • Thurman Thomas: 1,935 total DYAR, three seasons played before 1991.

Of the other 123 running backs in our DYAR database who started before 1991, the only one that seems to even have a chance is Marcus Allen, who sits at 1,400 total DYAR, and had five Pro Bowls seasons from 1982 to 1990, averaging the following per year over that span: 245 carries, 4.1 yards per carry, 53 receptions, 9.4 yards per reception, and 11 total touchdowns. Join us, won't you, for when we unveil the answer during 20th-anniversary festivities in 2023.

Let's finish up by going back to the question at the beginning of the column. After seeing all of these stats, here is how I would rank the six running backs: (1) Priest Holmes, (2) LaDainian Tomlinson, (3) Terrell Davis, (4) Tiki Barber, (5) Edgerrin James, (6) Jerome Bettis.

And here's why I would rank them that way...

When you look at DVOA and DYAR in a light more favorable to Davis, it's clear that the argument for him making the Hall of Fame before Priest Holmes boils down to Davis' Super Bowls. I say "Super Bowls" instead of "playoff performances" because, well, we wouldn't cite a single iota of postseason evidence about Davis if he had never made the Super Bowl -- except of course to point out that he had never won a Super Bowl. Say, off the top of your head, what were Eric Dickerson's career stats in the post season? Cris Carter's? Heck, what were Priest Holmes' postseason stats? I can tell you one thing, though. None of them ever made the Super Bowl.

UPDATE:As has been pointed out in the comments, Holmes won a Super Bowl as a part-timer with the Ravens in 2000. I think the fact that I forgot this kind of helps the more general point I was trying to make. If Steve Young had lost Super Bowl XXIX, the only Super Bowl-related aspect of his candidacy that people would have brought up was the fact that he never won one; which is funny since he actually won two prior to 1994. They would have been as worthless to his Hall of Fame argument as Holmes' ring seems to be.

So with "Terrell Davis, great champion" out of the way, I think Holmes has the better case. He produced more value as a receiver and more values as a runner. Those things remain true even if you discount credit for Holmes' extra two years or look at both of their six-year peaks. And while I'm going to postpone a lengthier, more philosophical discussion of "peaks" until the comments section, my rebuttal to, "Davis had the best two-year (and three-year and four-year) rush DYAR peak," is that Holmes had the best five-year (and six-year) rush DYAR peak. He also had a better total DYAR peak for all possible peak lengths. Hell, he also happened to have the best one-game peak in total DYAR and a better one-game peak than Davis in rush DYAR. Really, what's the definition of "peak" anyway? Why does no one talk about Tiki Barber's peak when half his 10-year career was spent as a punter returner and RBBC-member?

Regarding the rest of my rankings, I have Tomlinson ahead of Davis for basically the same reasons that I have Holmes ahead of Davis. Many of those reasons apply to the careers of Barber and James, but they can't thank me for my support because they didn't appear in any of the season or game tables. Finally, Bettis was like the anti-Holmes of researching stats for this column. Before it, I was squarely in the "lean yes" camp. Afterwards, I don't think Bettis comes close to stacking up. And yet, somehow, he's been a Hall of Fame finalist?

That's all for today; the worst running backs will be revealed on Monday. Now, let's have at it in the comments!

Posted by: Danny Tuccitto on 15 Aug 2013

151 comments, Last at 22 Aug 2013, 5:31pm by skd

Comments

1
by jklps :: Thu, 08/15/2013 - 5:07pm

Really nice to see how amazing that 1999 Stephen Davis (WAS) season was.

Game 2 vs the Giants when he bounced off three players in a pile for a TD was just the best.

2
by Andreas Shepard :: Thu, 08/15/2013 - 5:14pm

It's worth pointing out that Holmes got injured late in the 2002 season, so he only played in 14 games. His raw numbers were so good that it's easy to forget he didn't play a full season (313 attempts for 1615 yards and 21 TDs, and 70 receptions for 672 yards and 3 TDs)

If you pro-rate his performance to a 16 game season, he'd have accumulated 568 rush DYAR, which would be 2nd all time since 1991. He also would have had 297 receiving DYAR, good for 7th. Plus is 24 total TDs would pro-rate to 27, exactly what he achieved when he broke the TD record in a full season the next year.

From 01-03, Holmes had at least 2100 yards from scrimmage each season and accounted for 61 total TDs. The case against him is that he only had one other season over 1000 yards rushing (and then only just barely). But if you're going to base your HoF case on 3 great years, that's a damn fine 3 years to rely on.

12
by Dean :: Thu, 08/15/2013 - 6:34pm

And in that 1 season, something like 40% of his yards came in 2 games against the Bengals, who at the time were awful even by Bengals standards.

67
by Eddo :: Fri, 08/16/2013 - 12:10pm

When DYAR is being discussed, quality of opponent shouldn't matter.

70
by Dean :: Fri, 08/16/2013 - 12:43pm

I was specifically referring to this part of his statement: " The case against him is that he only had one other season over 1000 yards rushing (and then only just barely). "

In the early 2000s, most NFL teams were not considering DYAR or even any of their own in-house proprietary metrics when evaluating free agent RBs. But they would evaluate soemthing like a guy who just barely got 1000 yards, and 40% came in 2 games against a crappy defense. Additionally, he was considered a backup who wasn't nearly as good as the starter (Jamal Lewis - who usually had terrible metrics but yet was very visually impressive as a back).

If given a choice, AT THAT TIME, between Priest Holmes and Jamal Lewis, NOBODY was picking Holmes.

Conventional wisdom at the time (and I don't recall FO debunking it) was that it was a terrible signing. But put Priest behind that KC line and feed him a ton of carries and suddenly he's a totally different back.

112
by Raiderjoe :: Sat, 08/17/2013 - 2:34am

Hmm..do not remember that. If happened could jsve only been in 2001 due yo scheduling format started on 2002. Chiefs could jot play Bengals 2x in regular season anymore.

113
by Shattenjager :: Sat, 08/17/2013 - 2:52am

He's talking about 1998, when Holmes was with the Ravens. He had 400 of his 1008 rushing yards in two games against the Bengals that year.

114
by Raiderjoe :: Sat, 08/17/2013 - 3:12am

Yes, realized error when taking dump a few minuets ago. Just made long post noting error made (along witj multi parapgrap comments on QBs [differrnt subject]) but it didn't take. Would do over if not so tired.

3
by RickD :: Thu, 08/15/2013 - 5:23pm

"That means Davis was nearly as valuable in seven seasons as the 85 combined seasons of every other Shanahan-era running back for the Broncos and [Redskins] not named Clinton Portis."

Sure, if you think 1829 is close to 2721. I think 2721 is nearly 50% bigger than 1829. (And it's only 2721 because you dropped Portis, which is really hard to justify doing.)

We get it, TD had a great peak.

But it didn't last long.

8
by SorryThisIsBad (not verified) :: Thu, 08/15/2013 - 6:13pm

Yeah, "if you drop Davis and Portis out Denver drops out of the top 10 altogether". Meaning the system turned nobodies like Mike Anderson and Orlandis freakin' Gary into not only starting-caliber NFL running backs, but above average (half of 32 is 16, dudes) NFL starting running backs. And this argues FOR?? putting Davis into the Hall of Fame?

Maybe he does belong, but such awful, awful analysis.

25
by D :: Thu, 08/15/2013 - 9:34pm

I didn't like this either

During Shanahan's tenure with the Broncos (1995-2008), the cumulative 4,862 rush DYAR by running backs ranked No. 1 over that 14-year span. Remove Davis from the total, and they drop to third. Remove Davis and Portis, and they drop out of the top 10 entirely.

If you are going to do that you need to drop everybody else' top 2 backs and then rerun the numbers.

27
by Danny Tuccitto :: Thu, 08/15/2013 - 9:45pm

Quite selfishly, I'm totally fine with you guys making Davis look worse in relation to Holmes.

11
by Danny Tuccitto :: Thu, 08/15/2013 - 6:22pm

That's fair. The reason I said "nearly as valuable" is because we're talking about one guy vs. 40 guys. If you remove Mike Anderson too, then Davis is virtually tied with the other 39, but I didn't see a reason to exclude Anderson since I think we can all agree that Portis was a great back in his own right, while Anderson wasn't.

Even accepting your point, though, I still think that the "Davis was a Shanahan back" card is massively overplayed by his detractors. It's somewhere that I agree with Scott, so wanted to argue Holmes vs. Davis as honestly and gracefully as possible.

17
by Led :: Thu, 08/15/2013 - 7:21pm

It's not just that Davis was a "Shanahan back," it is that Davis was a Shanahan back and his peak happened in the years Denver otherwise had the best talent on offense during the Shanahan era. Elway, Shannon Sharpe, the criminally underrated Rod Smith, a very good O-line. Portis averaged his 5.5 yards a carry for two years with Brian Griese and Jake Plummer at QB. I'm very much in favor of what I would call a "sustained peak" over career totals, so I'm sympathetic to the argument for Davis. But context matters. Davis's short peak happened to overlap with probably the most favorable environment for a RB to put up big yards that we've seen in the last 20 years. Maybe KC under Vermeil is comparable. So the argument is not a slam dunk for me.

4
by D :: Thu, 08/15/2013 - 5:23pm

When you look at DVOA and DYAR in a light more favorable to Davis, it's clear that the argument for him making the Hall of Fame before Priest Holmes boils down to Davis' Super Bowls. I say "Super Bowls" instead of "playoff performances" because, well, we wouldn't cite a single iota of postseason evidence about Davis if he had never made the Super Bowl -- except of course to point out that he had never won a Super Bowl. Say, off the top of your head, what were Eric Dickerson's career stats in the post season? Cris Carter's? Heck, what were Priest Holmes' postseason stats? I can tell you one thing, though. None of them ever made the Super Bowl.

This may just be awkward phrasing, but I should point out that Holmes made the Super Bowl with the 2000 Ravens.

9
by Danny Tuccitto :: Thu, 08/15/2013 - 6:15pm

Holy christ, you're right. I had totally forgotten that since he wasn't a big piece of their puzzle. But, I guess my forgetting it in and of itself helps the point I was trying to make. No one talks about Holmes' Super Bowl ring with BAL, but rings seem to be the centerpiece of why Davis should get in over him.

Thanks for pointing this out.

132
by wardh2o (not verified) :: Sun, 08/18/2013 - 4:34pm

It seems that playing well in and winning playoff games and Super Bowls has a strong correlation with football fame. In considering who should be in the Hall of Fame (not the Hall of the Statistically Best Players), playoff and Super Bowl performance is an important factor.

148
by Silm (not verified) :: Tue, 08/20/2013 - 2:31pm

Priest Holmes was a bit more valuable to the ravens than he gets credit for.

I think its much different than say Steve Young who obviously just watched from the sidelines. Holmes contributed enough to realistically get SOME degree of credit, if not a large one, for helping the Ravens get that ring without a stellar passing game.

I think a fair analogy would be arguing whether Bernard Pierce deserves some semblance of credit for the 2012 Ravens winning it. I think the answer would be clearly yes, but just not as much as Rice.

5
by Karl Cuba :: Thu, 08/15/2013 - 5:29pm

Roaf, Waters, Weigmann, Shields, Tait. Plus Gonzo.

The defense rests.

16
by Danny Tuccitto :: Thu, 08/15/2013 - 7:16pm

KC OL ALY rankings from 2001-2003: 4, 3, 3

DEN OL ALY rankings from 1996-1998: 5, 1, 1

Your honor, the prosecution rests its rebuttal.

18
by Karl Cuba :: Thu, 08/15/2013 - 8:02pm

It hurts both cases, it doesn't remove the fact that Holmes ran behind one of the best lines in recent history, or even all NFL history.

Roaf and Shields should easily make the hall, Waters has a very good case for a guard, Weigmann was a really good center and after playing on the right for the Chiefs, Tait went on to play very well on the left for the Bears. It was the type of line that prepubescent running backs dream of running behind. Yes, that Denver line was also a thing of beauty but it doesn't stop a reasonable observer from thinking that Holmes would not be the same back playing for last year's Bears.

I vaguely remember a game where Holmes and Larry Johnson both ran for four touchdowns, how can anyone fail to knock Holmes for romping through the holes that group opened? (It's like Matt Miller suggesting that TJ Yeldon is equal to Adrian Peterson because he galloped through the chasms forged by last year's Bama line.)

48
by Thok :: Fri, 08/16/2013 - 10:21am

I'm pretty sure you're somewhat misremembering: the game you're thinking of is probably the 2004 KC-Falcons game, where Holmes and Derrick Blaylock scored 4 TD's each

79
by Karl Cuba :: Fri, 08/16/2013 - 1:44pm

Ah yes. The beast, the monster, the unstoppable force that was Derrick Blaylock.

21
by Scott C :: Thu, 08/15/2013 - 9:06pm

ALY != OL quality. Its one of the weakest FO stats for measuring anything other than how the team overall executes in the run game.

The KC line could pull in the run game play the screen better than anything I've ever seen before or after.

26
by Danny Tuccitto :: Thu, 08/15/2013 - 9:43pm

My first thought is "you have a better one?" Besides your eyes? Not that eyes don't matter, of course.

But I'll answer my own question. How about Open Field Yards? Presumably, that's OL-independent YPC.

DEN 1996-1998 rankings: 4, 12, 3.
KC 2001-2003 rankings: 12, 11, 19.

Bet that's more to your liking.

Of course, OFY isn't adjusted for SOS or game situation, so here (at least) are the offensive SOS's (mind you, these really should be run O SOS's, but I don't have that):

DEN 1996-1998 rankings: 11, 17, 4.
KC 2001-2003 rankings: 11, 21, 25.

Man, guess you guys are onto something here, so I concede the point: Holmes rushing stats benefited more from his OL than Davis' rushing stats. So what's a reasonable relative proportion to reduce Holmes' rush DYAR by? Was his OL worth 10% more to him than DEN's was to Davis? 20%?

The thing I won't concede is the screen game point. For that to make any meaningful difference for how the two rank in total DYAR, you'd have to assume that all KC ran was screens, then ignore that DEN was running screens all day long in the WCO, and then come up with a relative proportion of around 80% (e.g., KC's OL meant 90% to his receiving DYAR, while DEN's OL meant only 10% to Davis') to reduce Holmes' receiving DYAR by.

32
by Independent George :: Thu, 08/15/2013 - 11:01pm

I don't ever hold it against a RB if he played with a great OL, but I will credit him if he played with a bad one.

The same holds true if you replace 'RB' with 'QB', and 'OL' with 'WR'. Thus, Barry Sanders and Stubbleface get extra points in my book, but Priest Holmes and Dan Marino don't lose any.

38
by Scott C :: Fri, 08/16/2013 - 3:18am

Where did I mention Davis at all?

In the comments on this site over many, many years, ALY have been one of the most picked-apart stats. It is a great _team_ metric, but not one that isolates OL from RB (or QB! running QB threats help RBs!).
Just like how a QB strongly affects adjusted sack rate, RB affects ALY. Two RB's on the same team can differ from each other quite a bit in open field skills, as well as turning 1 yard runs into 2 yard runs.

Likewise, a good OL helps cause more open field yards! You aren't going to be getting many 50 yard runs if your blockers never make it to the second level.

The short of it is, I have seen no strong evidence on this site or elsewhere, and much to the contrary, that ALY == OL and everything else is RB. I don't think you can say that either Davis or Holmes got more help from their OL than the other by looking at ALY or OFY and comparing ranking numbers. (rankings? 1st vs 5th might be a smaller absolute difference than 5th to 30th, or 10x larger)

86
by Danny Tuccitto :: Fri, 08/16/2013 - 2:24pm

Where did I mention Davis at all?

You didn't. I was just piggybacking off your screen point about Holmes to compare it to Davis since my focus here has been comparing Holmes vs. Davis. That's all.

60
by Aaron Brooks Go... :: Fri, 08/16/2013 - 11:16am

My first thought is "you have a better one?" Besides your eyes? Not that eyes don't matter, of course.

But I'll answer my own question. How about Open Field Yards? Presumably, that's OL-independent YPC.

DEN 1996-1998 rankings: 4, 12, 3.
KC 2001-2003 rankings: 12, 11, 19.

Bet that's more to your liking.
----------

If two teams have the same ALY, but one team has the better OFY, doesn't that imply the other team has the better line?

Both teams perform the same in the running game (ALY), but one team has better running backs independent of line play (DEN). Therefore the other team (KC) has better linemen.

6
by RickD :: Thu, 08/15/2013 - 5:45pm

On the whole, I agree more with this analysis than with Scott's (though you both have TD ahead of Bettis, and you don't bring up Roger Craig at all, which raises the question: why do I agree more now? You're asking better questions.)

Your first table on the right shows how DVOA really is not the best measure for RB contribution. Law Firm is 6th on the list for his second-to-last season on the Patriots and they didn't even make a serious effort to retain his services. (I don't remember if they bothered to make an offer - if they did it was at best half-hearted like their Welker offer.) DVOA rates his 1008 yards rushing as better than TD's 2008 yards rushing? That's the problem with using an average measure.

I hope this analysis makes the rounds for the next Hall of Fame voting. It would be nice to have the historical record corrected a bit: Priest Holmes was a much better RB than Jerome Bettis, for starters. And this needs correcting:

" I can tell you one thing, though. None of them ever made the Super Bowl."

Actually, Priest Holmes did win a ring with the Ravens. Admittedly this was as Jamal Lewis's backup, but hey...he had 4 carries for 8 yards!

10
by Danny Tuccitto :: Thu, 08/15/2013 - 6:17pm

Yeah, massive oversight that actually furthers my point about selectively focusing on rings for certain players but not for others.

61
by Aaron Brooks Go... :: Fri, 08/16/2013 - 11:18am

You can claim that, but you're wrong.

TD is acclaimed for being instrumental in two SB wins, whereas everyone forgot Holmes even played in one. His team won anyway, despite his presence being a non-entity.

109
by Raiderjoe :: Sat, 08/17/2013 - 2:00am

Holmes technicslly started SB 35 rven though J. Lewis was furst string.

7
by DenverCheeze (not verified) :: Thu, 08/15/2013 - 6:04pm

"Peterson's 315-yard game in Week 9 of his rookie season...(94 total DYAR)...(1) It came against a 2007 Chargers run defense that finished 23rd in DVOA"

So you all really think that the second stringer (replacement) would have gained 220 yards on that day against the Chargers defense? Something seems wrong there.

22
by Whatev :: Thu, 08/15/2013 - 9:28pm

Well, first, not quite, because the line is 314 yards and a fumble. Then, the argument is that the second stringer would've gained 220 yards less the difference in value between the actual fumble and the replacement's statistical fraction of a fumble, given the number of carries in the same situations. But of course, the situation and number of carries are themselves related to the ability of the back--a second stringer wouldn't be used as often, and wouldn't be running in as favorable situations (and, of course, isn't as good). So if we say the second stringer makes 25% less yardage per touch and gets 25% fewer touches, which I think aren't entirely implausible, then he generates 56.25% of the yards from scrimmage.

In sum, do I think the second stringer could've made, say, 120 yards from scrimmage? Yes, I do.

50
by sn0mm1s :: Fri, 08/16/2013 - 10:25am

In a nutshell - yes, that is exactly what it means given the same touches, fumbles, and TDs.

This is one of my #1 complaints about DYAR/DVOA - it often fails the eyeball/common sense test. I mean, does anyone really think that a replacement level RB would've led the league in rushing last year taking over for Peterson on the Vikes? That is what the numbers from last year suggest.

I have brought this up before - DVOA/DYAR is great for predicting team success... but it severely devalues game breaking players/plays. It isn't really good for judging individuals.

54
by Will Allen :: Fri, 08/16/2013 - 10:33am

The Vikings may not have won 5 games last year, with a replacement level back taking Peterson's spot.

56
by sn0mm1s :: Fri, 08/16/2013 - 10:36am

Sure, but that is tangential to what is being critiqued.

63
by Will Allen :: Fri, 08/16/2013 - 11:30am

I believe the entire point in gathering statistics on football teams and players is to gain insight as to what wins football games. If an attempt to measure running back performance is pretty much whiffing on the fact that a single running back's performance may have made a 5 win difference for a team in a single season, I wouldn't think that is tangential.

73
by sn0mm1s :: Fri, 08/16/2013 - 1:26pm

Yes and no.

In the DYAR/DVOA world a RB, that starts on his own 20, that has 4 consecutive 10 yard rushes is worth more than another RB that has a single 79 yard rush from the 20 to the opponent's 1.

I think everyone would take the 79 yard rush - but the problem is that long rushes aren't predictable so they are penalized in the DYAR/DVOA world. They are treated as outliers (despite some RBs being great at breaking off long runs).

A team that can rip off 4 consecutive 10 yard rushes implies that the team is a good running team that can predictably, and repeatedly run - this is why DYAR/DVOA is good at judging team success. However, all of us would take the 79 yard run which is *better* but not as predictable.

This is why guys like Peterson, Sanders, and CJ are severely penalized in the DYAR/DVOA. Being able to break long runs hurts you in FO's methodology.

77
by Ryan D. :: Fri, 08/16/2013 - 1:42pm

Let's play another round of Chicken vs Egg.

I remember some talking head last year say that the reason AP breaks so many long runs is becuase if he can break through the first level against a stacked box, he's only got 1 guy to beat, the deep safety, to score.

Much like Barry Sanders before him (and to a lesser extent, Chris Johnson now), do teams stack the box against AP because they know he can break a long run more than most other backs, or does he break long runs more often because the box is always stacked against him?

AP only had 5 carries last year that went over 50 yards, and four of them were scores. Assuming they were all against an 8-man box, would a 7-man box with an extra deep safety have really helped prevent the really high yardage on those plays? Or, would it have allowed for him to continuously grind out slightly longer runs over the course of the game? Which is better or worse for the defense?

84
by Will Allen :: Fri, 08/16/2013 - 2:03pm

Oh, I think the defenses are being very rational, given the Vikings passing attack. Peterson's best day as a pro came against a defense which had an idiot, Ted Cottrell, as a defensive coordinator, who refused to put 8 in the box. He rushed for 300 yards.

With the Vikings pretty incompetent at passing downfield, you'd be nuts to let Peterson have 10 to 20 yard runs all day.

83
by sn0mm1s :: Fri, 08/16/2013 - 1:57pm

It is a good question. I think FO has the data to give some sort of clarity over the past few years with its game charting. I don't think it goes back very far though.

80
by Will Allen :: Fri, 08/16/2013 - 1:50pm

I tend to agree that DYAR/DVOA penalizes long runs in a way that obscures some running backs' value. This gets at the value of symbiosis again. Yes, a conssistent running attack will obviously help a passing attack, and vice versa. A running attack or passing attack which is consistently good, albeit perhaps less so than a very consistent output, but is also a real threat to score a td on every play, helps the other attack, run or pass, a bit more. At least that is my suspicion. If we could turns back time, and make Barry Sanders and Emmitt Smith switch teams, we might gain some insight. My guess is that the Lions would not have been as successful with Smith, and the Cowboys may actually have been better, but very likely no worse, with Sanders.

13
by Dean :: Thu, 08/15/2013 - 6:38pm

Before I read this, I wrote down your answers 1) LT, 2) TD 3) Edge 4) Bus 5) Priest 6) Tiki - and of those I would only actually vote in 1 and 2 and maybe 3. Now? I still put LT 1 and Davis 2, but I'll move Priest up to a tie for #3 in the borderline category. He still stunk when he wasn't behind an all-world OL.

Oh and "Join us, won't you, for when we unveil the answer during 20th-anniversary festivities in 2023." - you got it!

14
by Jerry :: Thu, 08/15/2013 - 6:48pm

One thing DVOA has never handled well is clock management late in games. There's value in kneeling out the last two minutes of a close game, and DVOA doesn't capture it.

As some guy named Kacsmar wrote: "One of the great stats of Bill Cowher’s career was his record with an 11+ point lead (102-1-1)." The NFL apologized for a bad call in the loss. One reason they were so exceptional in that circumstance was the ability to ride the Bus into eight- and nine-man boxes. No, Bettis couldn't do what Faulk could in space. But Faulk couldn't do what Bettis could in heavy traffic.

24
by D :: Thu, 08/15/2013 - 9:31pm

No, Bettis couldn't do what Faulk could in space. But Faulk couldn't do what Bettis could in heavy traffic.

I agree to an extent, but it should pointed out that during his peak from '99-'01 Faulk finished lead the league in success rate twice and finished second once, so it's not like he couldn't pick up consistent yardage. Admittedly teams were generally too scared to put safeties in the box, but Faulk could move a pile if he needed to.

52
by sn0mm1s :: Fri, 08/16/2013 - 10:32am

Faulk was really only great when he was surrounded by HOFers/borderline HOFers. He was a pretty poor runner on the Colts.

In his 5 years on the Colts he totaled 374 DYAR on 1389 carries.

In his first year on the Rams he totaled 339 DYAR on 253 carries.

Huge difference. If anyone is knocking Holmes, Davis, Smith etc. etc. for surround talent they should do it just as much for Faulk.

57
by Will Allen :: Fri, 08/16/2013 - 10:39am

Gosh, I had forgotten that Faulk was there 5 years.

62
by Lance :: Fri, 08/16/2013 - 11:26am

He was great in college, though, playing for a totally non-elite team. I think Faulk was pretty great, regardless.

65
by Will Allen :: Fri, 08/16/2013 - 11:35am

Anybody who doesn't think Faulk was great is nuts, but it is interesting how his receiving value leaped with Peyton Manning's rookie year, and his overall value leaped again when he arrived on a team loaded with offensive talent.

68
by Independent George :: Fri, 08/16/2013 - 12:17pm

It makes me think about how many other players weren't fortunate enough to find a team that could take advantage of their talents. Steve Smith and MJD immediately come to mind - they've had incredible success in spite of the dearth of talent around them.

I actually do think Steve Smith does belong in the HoF, but he'd be a slam-dunk case in better circumstances.

76
by Will Allen :: Fri, 08/16/2013 - 1:35pm

When ya' make Jake and The Delhommes look like, at times, a headliner act, I'd say you were a pretty great member of the band. I was curious, so I looked back at the 2001 draft, and saw that the Patriots took Seymour and Light prior to Smith coming off the board, and those were two pretty darned good picks. The Packers had four picks before Smith went to the Panthers, however, and none of them amounted to much. If Stubbleface had Smith to throw to, I think he would have avoided his mini-slump in the Mike Sherman era, Sherman would have been looked upon as a much more successful coach (but since he was GM, this doesn't say much), and Smith would now be viewed as a first ballot HOFer. Those two guys would have been scary together; Smith always believed he could get separation from anybody, or outfight them for the ball, and The Jeans Model would have been squeezing the ball in to a guy that made The Jeans Model seem less reckless, because he'd make the play.

Luck plays a significant role in how these careers are viewed, in case anybody hasn't figured it out yet.

89
by DisplacedPackerFan :: Fri, 08/16/2013 - 3:15pm

Stubbleface + Smith + Driver with Ahman Green at the beginning of his peak, that would have been fun. There was a lot of "who is the other receiver on this team?" from 01 - 05. The #1 - #2 by numbers were in order Schroeder - Freeman; Driver - Terry Glen; Walker - Driver; Walker - Driver; Driver - Antonio Chatman. I would have loved a Smith - Driver or Driver - Smith, and if they still took Walker you would have had a Driver - Jennings - Nelson or Nelson - Jones - Cobb like trio to play with the old man. Of course Sherman imploded as a coach when he got the GM duties, he couldn't work as a coach when he had picked the player so many stories of him simply walking away when approached with a problem instead of dealing with it. I think he could have been a decent coach, but he was such an awful GM and he could not wear both hats. But yeah Smith would be a clear HOFer in that situation even if Sherman had kept both hats. I mean Freeman made a career out of making the catch in a tight window because the QB was going to throw it there. Smith could actually get open and be a real threat in other ways too.

Ah what could have been. I still can't gripe, 20+ years of a HOFer at QB and counting.

97
by Independent George :: Fri, 08/16/2013 - 6:24pm

Oh hell - but for some bad luck, Stubbleface + Sterling Sharpe would have been an all-time great combination.

111
by Raiderjoe :: Sat, 08/17/2013 - 2:25am

S. Smith one of vest wide outs all tume. .

Up thrre with all yime greats which include
Rice, Biletnikoff, Warfield, R. Moss, Berry, Alworth,
Hutson, largent, Maynard, Fitzgerald, calvin johnson and
some othrrs.

Believe Steve Smith one of top 20 greatest wife receivers all time

116
by Whatev :: Sat, 08/17/2013 - 4:34am

I'm surprised they were willing to talk about it.

117
by Will Allen :: Sat, 08/17/2013 - 7:09am

You did that on purpose, Rj.

119
by Raiderjoe :: Sat, 08/17/2013 - 7:56am

?

Typos were vest and wife

Supoosed to be best and wide

Not aware of multiple wives for Smith. Know guy is married to one lady. If had more wives not aware of it. Guy played at Utah but not Mormon. Was from California

120
by Raiderjoe :: Sat, 08/17/2013 - 7:58am

Also tume suppose to be time

15
by DA (not verified) :: Thu, 08/15/2013 - 7:03pm

Here was my list prior to the article:

Tomlinson
Terrell Davis
Tiki Barber
Priest Holmes
Edgerrin Jame
Jerome Bettis

**Priest Holmes is a very tricky case. His numbers were great but he also was behind an amazing Offensive Line. It isn't a coincidence that Larry Johnson in 2004 and 2005 also shows up on these lists.

19
by Scott C :: Thu, 08/15/2013 - 9:00pm

How much _passing_ DVOA does Tomlinson have? How does that compare to the others?

8 for 12 with 7 TDs for 143 yards, 0 INTs, mostly on third down.

His pass statistics are better than any other RB in the last 30 years, I think. Until NORV stopped calling the halfback option plays. One less TD than Walter Payton, but Payton had 6 INTs and was 11 for 34.

Passing statistics don't come up often when comparing RBs, but for the few that are trusted on trick plays, if they execute well add a LOT of value.

As for Holmes, he was very good. His offensive line may have been the best in 30+ years.

20
by Scott C :: Thu, 08/15/2013 - 9:02pm

It "feels" like Tomlinson had about 200 or more DYAR passing. I doubt anyone else in the top ten has more than 25. His passing TD's were parts of come-from-behind 4th quarter wins even.

39
by CBPodge :: Fri, 08/16/2013 - 4:34am

I don't think Tomlinson is getting into the HoF on his passing skills!

144
by Scott C :: Mon, 08/19/2013 - 5:10pm

Every aspect of a player is important. No, he won't be getting in because of that, nor will he do so because his pass protection and blitz pickup was very good.

But those skills are important, and lead to wins.

Why do we even look at pass catching ability? Because there are stats to measure it. There are stats to measure trick play / pass ability. However, I have seen none published to rate these RBs in pass protection.

In my mind, I could easily shuffle my list based on some of these other qualities. After all, if I wanted to assemble the best all-time team I could with a time machine, such abilities would be important to consider!

23
by Dan :: Thu, 08/15/2013 - 9:28pm

"When you look at DVOA and DYAR in a light more favorable to Davis, it's clear that the argument for him making the Hall of Fame before Priest Holmes boils down to Davis' Super Bowls. I say "Super Bowls" instead of "playoff performances" because, well, we wouldn't cite a single iota of postseason evidence about Davis if he had never made the Super Bowl -- except of course to point out that he had never won a Super Bowl."

This is a bad argument for ignoring postseason performance. Yes, a lot of non-stats-oriented people do a bad job of accounting for postseason performance, but that doesn't mean that we should throw it out altogether. 8 games with a rushing line of 204/1140/12 is very impressive. Doing it in the playoffs makes that more impressive, not less, so we should be giving those games at least as much weight as regular season games (rather than zero weight).

29
by Danny Tuccitto :: Thu, 08/15/2013 - 10:08pm

No, no. I'm not arguing that playoff performance shouldn't matter. I'm arguing that the way playoff (and more so) Super Bowl performance is used in HoF discussions is totally out of whack. Re the SB, it's either "Did he win one or didn't he?" Then, it's, "Did he play in one?" Then, if either of those are yes, the next question is, "Was he good in the SB?" If that answer's yes, then people bring it up. And then -- and only then -- do people delve into non-SB playoff performance whatsoever.

My point is just that the way playoff performance enters the argument is madness. Of course TD's great SB performances should count, and so should his other playoff performances. I just think that it's intellectually dishonest to cite playoff performance only in very specific circumstances for very specific types of players, and then ignore it for everyone else except for the purpose of using "they didn't win an SB" against them.

33
by MarkV :: Thu, 08/15/2013 - 11:38pm

I think that there is a difference between TD and most running backs when it comes to playoff performance.

Most players have either no playoff performance, or they have one game a season. Variability from that ridiculous sample size should make any self-respecting stat lover quiver. Davis had a large sample in a small time frame, which makes those performances in the aggregate more worthy of statistical inclusion. Especially so when those 7 games make up a relatively high percentage of his career total.

40
by Dan :: Fri, 08/16/2013 - 6:46am

A lot of people do a bad job of taking players' playoff performance into account. When you brought up playoff performance in the article, you took a jab at those people and then didn't say much more about it. I prefer to ignore those people, look at the player's actual playoff performance, and figure out how we should take it into account.

The simplest way to account for playoff performance is just to add playoff DYAR to regular season DYAR in the tables. I suspect that that would be enough to move Terrell Davis pretty far up the rankings, into the top 5 on each list (and possibly as high as 3 on some of them), although he wouldn't catch Holmes. I'd be inclined to give more weight to playoff performance than to regular season performance, perhaps by multiplying playoff DYAR by 2 or by giving players an additional bonus for playoff awards or elite games, which would help Davis further.

41
by nat :: Fri, 08/16/2013 - 7:26am

It think you set up a straw man. Around here, when people talk about DVOA and DYAR for the playoffs, they tend to be sensible about it.

DYAR certainly could be included in career totals, and in single game lists without changing their meaning. Seasonal DYARs, and the weighted and six best years lists would still work, but would give a bonus for making the playoffs. I can see good reasons to include or exclude it for those.

DVOA could include playoffs for career totals, and for single game lists, too. Playoffs could be included in seasonal totals, and doesn't suffer from a playoff bonus problem. It's a per play average.

You could fairly give career playoff DYARs separately, since those already reward playing more games. Career playoff DVOA would include a lot of one-and-done careers. You might limit those by a minimum number of games or plays. Seasonal playoff DVOA lists would have the problem of excluding a lot of players, or turning into single game lists.

Really, when looking for "best" players, it's lame to exclude playoff performances entirely. The only excuse is that you haven't compiled those stats in a usable way yet. Your articles should just say that, instead of pretending there are good reasons. We'd respect you for it.

85
by Danny Tuccitto :: Fri, 08/16/2013 - 2:20pm

It think you set up a straw man. Around here, when people talk about DVOA and DYAR for the playoffs, they tend to be sensible about it.

I wasn't talking about people "around here." I'm talking about the larger football punditocracy, the larger football fandom, etc. Of course FO readers engage in more intellectually consistent arguments than the rest of the football-consuming population. Wasn't talking about you guys at all.

121
by nat :: Sat, 08/17/2013 - 8:08am

But why ignore the playoffs? The antidote to idiots focusing only on the Super Bowl isn't to ignore the most important production of a player's career. It's to include ALL of his career.

Include the playoffs, or don't claim to be measuring the 'best'. Or at least say in the article that you are judging regular season play, and whether players belong in the Hall of Regular Season Fame.

28
by deflated (not verified) :: Thu, 08/15/2013 - 9:50pm

Out of interest what were Terrell Davis' total DYAR and DVOA for his post-season games? It would have to be pretty big numbers for half a season of work.

30
by Danny Tuccitto :: Thu, 08/15/2013 - 10:22pm

All I have is the top 25 from 1995 to 2010. Davis has two of the top 25 (#1 and #7) for a combined 169 total DYAR (92 + 77). 25th had 60 total DYAR, so Davis' other 6 games have to be worse than that, which would make his max over 8 games 523 total DYAR. Even if he averaged 30 in those other 6 games, that's 349. Either way, pretty damn good.

p.s. Holmes has the #13 playoff game (70 total DYAR).

31
by Independent George :: Thu, 08/15/2013 - 10:56pm

I'm just going start by cutting & pasting a comment from the 'No Carries for Old Men' thread.

If he hadn't turned into such a massive douchewaffle at the end, this would be moment where Giants fans start extolling the greatness of late-career Tiki Barber

Except now I've had some time to think about it, and realize it's stupid to hold personal grudges against a guy I don't even know. Tiki Barber was a very good RB for a long time, and a great one at the end of his career. His 2005 season was better than MVP Shaun Alexander's (though I still think the award belonged to Steve Smith). He carried the Giants offense during the "Bad Eli" years (and, lest we forget, they were truly terrible). He had remarkable durability, especially for a small back. He could block. If you flipped his career around so that 2006 was his rookie season, and 1997 was his last, I don't think anybody would care that he was a part-time back at the end.

I don't think he belongs in the hall, but it's a close call for me, and he's a far more deserving candidate than Jerome Bettis or Roger Craig (both of whom I like as people far more than Tiki).

Also, that Marshall Faulk guy was pretty good, eh?

42
by Karl Cuba :: Fri, 08/16/2013 - 9:57am

I think I agree with most of this. Barber might have been a bloviating gasbag but he was one hell of a football player. The only real knock on his on-field performance would be the fumble issues early on in his career but to his credit he did fix that.

49
by Will Allen :: Fri, 08/16/2013 - 10:24am

Yeah, once his fumbling diminished, he was tremendous, but he was 8 years in before he solved the problem. One of the reasons I think Curtis Martin gets underrated by some folks is that they don't fully appreciate what value there is in a rb who gets a ton of touches, is darned productive, but never fumbles. Coaches like Parcells are keenly aware of it, however.

55
by sn0mm1s :: Fri, 08/16/2013 - 10:34am

Curtis Martin's YPC was also below league average for his career - I think him being in the HOF is him being overrated by most folks.

58
by Will Allen :: Fri, 08/16/2013 - 10:40am

Yeah, I think that view underrates the value of ball security somewhat.

34
by Will Allen :: Fri, 08/16/2013 - 12:21am

Let it be noted up front that this is not a criticism of Marshall Faulk. Repeat.

What even advanced stats have yet to capture is running back performance, within the context of the effectiveness of his team's passing offense. Brother, rushing the ball effectively, within the context of MVP level Kurt Warner leading the passing attack, or even rookie year Peyton Manning, was not the same task as doing so when Tavaris Jackson, Brooks Bollinger, Gus Ferrotte, Kelly Holcomb, Christian Ponder, Joe Webb, or Brad Johnson's AARP card were slingin' the ball over 9-10 guys (no hyperbole} within 5 yards of the line of scrimmage. This is not to say that Adrian Peterson even belongs in the same area code as Faulk in the passing game. He doesn't.

Perhaps a more illuminating comparison can be made with another Viking rb who appears in this piece, Robert Smith. Smith in my view was and is an under appreciated player. However, the disparity between his rushing performance and Peterson's is much larger than the numbers indicate, simply because Robert Smith never had opposing defensive coordinators who had the luxury of spending an entire week on nothing more than slowing down Robert Smith.

36
by dmstorm22 :: Fri, 08/16/2013 - 1:10am

Isn't it a chicken and egg dilemma with Faulk in STL. Many people, including Belichick in his Super Bowl XXXVI gameplan, considered Faulk the centerpiece, the nucleus, the engine of that offense. To me, passing the ball with Faulk as your running back becomes easier. Trent Green is really good, but that offense barely skipped a beat when Warner got hurt in 2000.

I do agree that what Peterson has done with the QBs in mindblowing, but there is little evidence that YPC is positively correlated with YPA from your passing game. Total numbers may go up or down, but great backs usually don't get a giant bump from playing with good offenses around them, or at least less so than what people conventionally think.

37
by justanothersteve :: Fri, 08/16/2013 - 2:40am

Trent Green was never the #1 QB for the Rams except in preseason 1999. Warner took over at that point and was the starting QB until he had his bell rung too many times on the bad TWA Dome turf, at which point Mark Bulger took over. Green never became a star until he went to KC.

43
by Sakic (not verified) :: Fri, 08/16/2013 - 10:00am

I'd have to go back and look at the numbers to be certain but I'm almost sure Trent Green made a handful of starts in 2000 for STL and I don't recall their offense dropping off at all.

51
by Dean :: Fri, 08/16/2013 - 10:26am

He made 5 starts that year. And at first glance, it appears that there is a bit of a dropoff. But I won't pretend I dug all that deep.

http://www.pro-football-reference.com/teams/ram/2000.htm

71
by mrh :: Fri, 08/16/2013 - 12:43pm

Chiefs fan, so I'm biased.

2000 - Warner 923 DYAR on 369 passes, 28.0% DVOA
Green 610 DYAR on 263 passes, 28.6% DVOA

I'd say they were about exactly the same; Warner's DYAR was higher because he played more. Green played a tougher schedule - his DVOA is about 1.5 percentage points higher than his VOA, while Warner's DVOA is about 4 points lower than his VOA.

I'm convinced that if Green hadn't been hurt in preseason 1999 that he would be thought of as a near-HOF quality QB.

Holmes did benefit from a great line. But having watched a lot of his runs, he had a great if not unique ability to take advantage of their blocking on sweeps or stretch plays outside the box, patiently waiting to make the right cut. And his receiving value was not just screens, but from leaking out down the middle of the field and finding a hole and then generating a lot of YAC on checkdowns. You can credit the line for keeping Green upright long enough to find Holmes but his ability to check for blitz pickups and then get to an opening downfield was remarkable - it always looked like easy yards for the Chiefs but if it was easy everyone would do it.

44
by Will Allen :: Fri, 08/16/2013 - 10:06am

The reality is that the more good offensive players that there are on a roster, the easier it is for those who throw, catch, or carry the ball to perform effectively. I agree that it is hard to capture this with metrics; that was also my point.

As good as Peterson has been (with some deficiencies in the passing game, to be sure) what he did last year in rushing was on an entirely different plane. To have his ypc leap to 6 yards per carry, on 348 attempts, on a team which won 10 games (no piling of yards while down 14 in the fourth quarter), against stacked boxes, because his team was pretty much wholly incompetent at the forward pass, is truly remarkable. With a replacement level back, the Vikings may not have won 5 games. This season, however, cannot crack the top 10 for the last 20 years. Something is being missed.

72
by Ryan D. :: Fri, 08/16/2013 - 1:15pm

He had two different 82-yard TD runs last season, with two more scores from 64 and 61 yards away, according to NFL.com's game log. How much value could that possibly cost him though?

74
by sn0mm1s :: Fri, 08/16/2013 - 1:28pm

It can hurt him a lot in the DYAR/DVOA measurement considering there are diminishing returns for everything over a 10 yard carry.

94
by dbostedo :: Fri, 08/16/2013 - 4:32pm

I assume there's no "quality of passing game" factor in rushing DYAR. Or maybe call it a "men in the box" multiplier.

I think that's what missing. DYAR I think is assuming the average play of each of the defenses that Peterson faced. Not the actual play only against the Vikings - which was likely designed specifically to stop Peterson.

So assuming the Vikings used a replacement runner against the same schedule as Peterson, AND that the teams they played were exactly average in how they played the run, AND the replacement back got the same opportunities as Peterson, THEN maybe the replacement numbers (or Peterson's seemingly too-low DYAR) makes more sense.

Or maybe I'm just rambling...

95
by Intropy :: Fri, 08/16/2013 - 4:47pm

No, you absolutely must not do that. If you did then Bettis might look like a Hall of Famer.

88
by Jimmy :: Fri, 08/16/2013 - 3:08pm

You could make a similar argument for Walter Payton's early to middle career.

35
by Bowl Game Anomaly :: Fri, 08/16/2013 - 1:00am

Here are a few reasonable statements about these various RBs' HOF cases:

1. Both Holmes and Davis had short careers, and so did Barber as a feature back.
2. Both Holmes and Davis played in very favorable circumstances (arguably the most favorable of any HOF candidates, certainly in the discussion with Emmitt Smith and Jim Brown).
3. Holmes, Davis, and Barber would each be locks to make the HOF if they had accumulated a few more productive (i.e. Pro Bowl) seasons, maybe as little as 1 season each, but certainly with 2.
4. None of those 3 has been a HOF finalist, but Davis has come closest. This is because of his playoff performance, which was as good as any RB ever. Davis is not Timmy Smith. He didn't just have one big game. Elite playoff production does and should matter for the HOF. If he had merely league-average playoff production, Davis probably has the worst case of the 3.
5. I can't quite wrap my head around the fact that Holmes is #1 on the rush DYAR list, #2 on the total DYAR list, and that Barber is higher than Tomlinson on the total DYAR list. Conclusion: Holmes and Barber were really, really good, and maybe Tomlinson was not quite as good as I thought.
6. The only true standout career, meaning the only player who truly separated from the pack according to these (regular season) career numbers, was Faulk, with an honorable mention to Holmes.

My final ranking of the 6 players listed in the article:
Tier 1: Tomlinson
Tier 2: Davis, Holmes, Barber
Tier 3: James, Bettis
Explanation: Tomlinson is the only one of the 6 who has both peak and longevity, plus he had less help than Davis or Holmes. Davis's playoff performances lift him above Holmes and Barber, who are basically all 3 indistinguishable as players who had massive peaks but short productive careers. James and Bettis do not compare to the others. (I know that James ranks higher than Davis on all 3 versions of the career list but so does Westbrook and even Garner nearly pulls it off. Suffice it to say that the career lists are not the reason why Davis is being considered.)

45
by Johnny Socko (not verified) :: Fri, 08/16/2013 - 10:07am

Hey Bill Polian - are you reading this?! I will never forgive Bill for breaking-up the Manning/Faulk tandem. Yes, Edge was a fine replacement, but the Colts could have (should have) payed Faulk what he deserved and used the 4th pick in the 1999 draft to take Torry Holt or Champ Bailey. Instead, Polian traded one of the best RB's of all time for 2nd and 5th round picks. What a disgrace.

47
by Will Allen :: Fri, 08/16/2013 - 10:20am

Peyton Manning with Faulk, Holt, and Marvin Harrison woulda' been something to behold.

87
by Jimmy :: Fri, 08/16/2013 - 3:07pm

I'd go with Bailey. The best corner in the league along with the existing Colts offense of the time would have been unstoppable.

64
by Independent George :: Fri, 08/16/2013 - 11:34am

It had to be done, and both teams benefited from it. Wasn't it the last year of his contract? Faulk had three more years playing at HOF-levels, and then tailed off due to injuries and age - exactly as you would expect. They were building a team around their young QB; they wouldn't truly be contenders until 2003, when Faulk began his decline.

66
by Will Allen :: Fri, 08/16/2013 - 11:39am

Yeah, teams were not yet as sophisticated in managing the cap 16-17 years ago, unless they were doing some cheating, Broncos-Style or Niners-Style.

78
by Karl Cuba :: Fri, 08/16/2013 - 1:42pm

What I thought was a little silly at the time was that he traded Faulk because he wanted to avoid the big contract but then had to give James a pretty big deal because he was a high draft pick. Didn't make a lot of sense to me.

81
by sn0mm1s :: Fri, 08/16/2013 - 1:55pm

Of course, James had a better season as a rookie than Faulk ever did on the Colts.

82
by Will Allen :: Fri, 08/16/2013 - 1:57pm

With Manning no longer a rookie, of course. Yeah, these things are hard to sort out.

90
by Bobman :: Fri, 08/16/2013 - 3:44pm

It was a holdover from the team getting burned by overpriced older guys before--Dickerson being notable. It was the safe move, though it seemed somewhat crazy at the time, and a line in the sand for all future negotiations. It was bigger than just a personnel move, IMO.

105
by Mxpx (not verified) :: Fri, 08/16/2013 - 9:35pm

So you think it was a good idea to trade Faulk and Champ Bailey for E.James and Mike Peterson? If so, I've got some swamp land in Florida...

46
by Thok :: Fri, 08/16/2013 - 10:12am

Man, if only those early 2000ish Kansas City teams could have pretended to have a defense.

53
by Anonymous37 (not verified) :: Fri, 08/16/2013 - 10:33am

This just reminds me why I don't think DYAR does a very good job with running backs when I look at Quick Reads rankings during the season.

In that first DYAR chart, best rush DYAR for a season, 1998 Terrell Davis has 602 while 2006 Tomlinson has 460 - a difference of roughly 30 percent. This despite Tomlinson having 45 fewer carries, trailing by less than 200 yards, and actually having 7 more TDs than Davis.

I'm well aware other factors enter into DYAR (I thought turnovers would be one in this case but they each had only 2 fumbles) but I still think the discrepancy in their season totals is too large to be reasonable. Frankly, if we were able to swap the two players to each other's teams for those respective seasons, I think LDT would have a better chance of outperforming Davis's numbers than vice versa.

59
by Aaron Brooks Go... :: Fri, 08/16/2013 - 11:11am

" then perhaps the most interesting about the table is how many of the greatest games since 1991 in terms of yards from scrimmage aren't in the top 10 of total DYAR."

That's to be expected. DVOA has always punished volume runners, and intentionally punishes long rushes. It's a flaw that's treated as a feature. You shouldn't expect any runners with large amounts of rushing attempts to be near the top -- that some are is indicative of how amazingly good they were.

98
by Duff Soviet Union :: Fri, 08/16/2013 - 7:27pm

Er, no. DVOA doesn't "punish volume runners" at all. There are lots of guys who are mediocre on a per carry basis who end up with good DYAR figures just because they carry the ball a lot. It doesn't punish guys who carry the ball a lot unless most of those carries are bad. I don't see what's wrong with this.

It also doesn't "punish" long runs at all. Repeat after me, "long runs do not have negative value". What it does punish is guys who end up with 20 carries for 100 yards....with 90 of them coming on one play. That one play is awesome. Repeat: AWESOME. It doesn't make up for the fact that his other 19 carries gained 10 yards. And nor should it.

Look, I know you're upset that your favourite player is treated as merely excellent by Football Outsiders stats rather than as the Golden God you see him as, but that doesn't mean they're wrong.

99
by Will Allen :: Fri, 08/16/2013 - 7:44pm

Look, I know you think DYAR is a perfect measuring tool, which, with 100% accuracy, captures the value of a running back's performance, thus making it important to have a semantic battle over the term "punish", but that doesn't mean a condescending tone is somewhat inadvisable.

100
by Duff Soviet Union :: Fri, 08/16/2013 - 8:20pm

It's hardly a "semantic battle". He said (and has said repeatedly in the past) that FO stats penalise runners who carry the ball a lot and penalises long runs. Both of these statements are patently wrong.

I never said DYAR is perfect. There's the obvious player interdependence issue and I'm not sure that receiving by backs is properly valued, but yeah, I think it's pretty good. It's better than any other statistical analysis of running back performance I've seen. If anyone can point me to something better, I'll happily read it.

118
by Will Allen :: Sat, 08/17/2013 - 7:17am

I also think it is pretty good, while also thinking it undervalues (no punishing!), a little, the type of runner who is a greater threat to go the distance with each carry. As I posted elsewhere in this thread, if I had a DeLorean modified for time travel, I'd like to go back a couple decades and a couple more years with Sanders and Smith, and have them change teams.

103
by sn0mm1s :: Fri, 08/16/2013 - 8:37pm

It does punish long runs. Diminishing returns begin with everything longer than 10 yards.

104
by Duff Soviet Union :: Fri, 08/16/2013 - 8:52pm

I was under the impression that it basically stopped counting yards after 40, so a 90 yard run is scored the same as a 40 yard run.

Maybe this is getting semantic, but I think there's a difference between "diminishing returns" and "punishing long runs". The former says that yards 6-10 are more important than yards 31-35, while the latter says that yards 31-35 have negative value.

This isn't just a Fooball Outsiders thing either. Advanced NFL stats which uses a system that I think is pretty similar to FO (I think with different success rate baselines) has CJ Spiller comfortably ahead of Adrian Peterson in it's Estimated Points Added metric despite the fact that Peterson had almost twice as many carries at the same average basically due to Spiller's much higher percentage of runs that were actually good (success rate) and greater receiving value. They don't have data on Barry Sanders' career, but I reckon it would tell a pretty similar story about him. Excellent, but not as great as his yardage and yards per carry would indicate.

107
by sn0mm1s :: Fri, 08/16/2013 - 10:23pm

AFN is quite a bit different. They don't have diminishing returns. Also, they mix their rushing and receiving stats so I wouldn't be surprised if there are some cases of Simpson's paradox in their results.

FO continues counting after 40 yards - the yards are just worth much less than the first 10. Which makes little sense in the way they value ALY. They basically state that the Oline is more important in the negative to 4 yard range while anything 10 yards or more is based more on the ability of the RB and the Oline has little to do with it. If that is the case why put significantly more weight on the runs heavily influenced by the Oline and not RB ability when ranking individual RBs?

115
by Jerry :: Sat, 08/17/2013 - 4:14am

An offense runs the exact same play twice, and the defense reacts exactly the same way each time. In both cases, the back pops wide open and goes in for the touchdown. Is the play that started at the offense's 20 twice as good as the one that started at the defense's 40?

122
by sn0mm1s :: Sat, 08/17/2013 - 1:52pm

Yes, (at least as far as yardage is concerned).

What you are alluding to here and what FO more or assumes is that if you are going to break it 40 yards you are going to go the full distance no matter where you are on the field. In reality that is no where near the case.

Since 1999, there have been exactly 10, 40 yard rushing TDs while there have been 14, 80 yard rushing TDs. However, the the number of total rushes from the teams' own 20 are 7155 while from the opponents' 40 it is 2274.

If breaking it from the 40 is just as likely as breaking it from your own 20 then we would expect ~31 80 yard rushing TDs over that same time period. We have less than 1/2 of that. Meaning is more than twice as likely to score a 40 yard rushing TD than an 80 yard rushing TD.

106
by Danny Tuccitto :: Fri, 08/16/2013 - 9:43pm

I'm going to give the benefit of the doubt and assume your acronym sleight of hand was intentional here, but DVOA's only influences DYAR to the extent that it's what establishes replacement level. DYAR explicitly rewards volume runners insofar as they're also efficient (See DYAR tables as opposed to DVOA tables). Volume runners get penalized by DVOA because DVOA is efficiency per play, but they're helped by DYAR because DYAR values volume (See what happens to the DVOA table when it goes from a 100 to 200 minimum).

123
by Aaron Brooks Go... :: Sat, 08/17/2013 - 10:21pm

My argument has always been that DVOA over-rewards change-of-pace 3rd down backs (hello Eagles and Patriots) by granting them a disproportionately high per-play value, which is not recaptured on a volume basis by DYAR.

Basically, this: A scatback who is incapable of performing the 100-150 dives and off-guard rushes inherent to an offense (rushes with low DVOA potential) is not more valuable than a well-rounded running back who can perform those scatback rushes and also perform the 100-150 dives per year that the offense still needs, even though they aren't high-efficiency plays.

My argument has also been that there's no sound reason for two 20-yard rushes to be worth more than one 40-yard rush. There's also something of a path-dependence to DVOA. It wants a certain progression of yardage, even though that progression in and of itself, has no meaning within the game. Consider that a RB who generates three yards per rush, every rush. Consider him to be RoboHoard. This runner is unstoppable, and will generate a TD every drive. DVOA hates this runner. Consider another runner, who loses five yards on one rush, gains 5 on the next, and gains 15 on the third. This cycle then repeats. Consider him to be RoboSanders. This runner is also unstoppable, and DVOA also hates him.

124
by Jerry :: Sat, 08/17/2013 - 11:44pm

My argument has also been that there's no sound reason for two 20-yard rushes to be worth more than one 40-yard rush.

One 40-yard run is a higher DVOA than two 20-yard runs. 100 40-yard runs are a higher DVOA than 200 20-yard runs.

Of course, two 20-yard runs are a higher DVOA than a 0-yard run, and 200 20-yard runs are a higher DVOA than 100 0-yard runs.

It's when you combine the zeroes and forties to come up with "equivalent" rushers that you get to the point you're describing.

128
by sn0mm1s :: Sun, 08/18/2013 - 11:28am

But the two 20 yard rushes are worth more DYAR.

130
by Arkaein :: Sun, 08/18/2013 - 2:54pm

One 40 yard run only gets you one first down, which means it only can make up for a small number of failed offensive plays before punting or a turnover on downs.

Two 20 yards runs gives you two first downs, and can make up for potentially twice as many failed offensive plays while maintaining possession. Not if the plays happen back-to-back of course, but in at least some game situations two 20 yards runs will lead to a TD, where a 40 yard run by itself would not have.

There are a few other minor arguments in favor of the current DYAR system, although they aren't explicitly part of the calculation. Teams trying to run out the clock will benefit more from two 20 yards runs. Teams with a defense needing rest will similarly benefit from two first down plays rather than one long one.

131
by sn0mm1s :: Sun, 08/18/2013 - 4:06pm

A 40 yard run makes up for just as many failed offensive plays. You end up in the exact same place as 2 20 yard runs. In fact, it could be argued that the 40 yard play is better since it eliminates the need to convert another set of down and eliminates potential turnovers/turnover on downs while moving the same distance as 2 20 yard runs.

There is no situation where 2 20 yard runs lead to a TD where a 40 yard one wouldn't - they are the same distance.

In the DYAR world 4 consecutive 10 yard runs generates more DYAR than 2 20 yard runs which generates more than 1 40 yard run. However, the 40 yard run puts you in the exact same place and effectively eliminates the need to convert more sets of downs to move the ball. In every measurable way the 40 yard run is better.

140
by Scott C :: Mon, 08/19/2013 - 4:38pm

"There is no situation where 2 20 yard runs lead to a TD where a 40 yard one wouldn't - they are the same distance. "

Easy to debunk!
Start at your 30 yard line after receiving a kickoff.

Case A: -15yd personal foul, 40yd run, 7yd pass, -15yd personal foul, 10yd run, punt.

Case A: -15yd personal foul, 20yd run, 7yd pass, -15yd personal foul, 20yd run, 7yd pass, 45yd Pass TD.

In the long run, two 20 yard runs randomly distributed in the game is MORE VALUABLE than one 40 yard run because it can cover up two mistakes and sustain drives. The only case it is not and your argument holds is if the runs are back-to-back.

I can construct other examples where the 40 yd case leads to one TD in two drives, but the two 20yd runs lead to two TDs in otherwise identical drives. Large numbers of medium sized plays (10 to 25 yards) correlate with offensive success more than frequency of very long plays (40 yards+).

145
by sn0mm1s :: Mon, 08/19/2013 - 10:54pm

1) I think it is pretty clear that we are talking about consecutive runs.
2) such contrived examples aren't really useful.
3) not sure why the team would punt on 3rd down in your example.

I can make similar contrived examples showing why a 40 yard run is better.

146
by Jerry :: Tue, 08/20/2013 - 2:31am

You may be talking about consecutive runs, but DVOA definitely isn't.

147
by sn0mm1s :: Tue, 08/20/2013 - 10:16am

Yeah, and that makes it even worse. What is better on 1st and 10? A 40 yard run from the 30 to the 30 or two twenty yard runs from the 30 to 50?

In DYAR, the two 20 yard runs are much better.

AFN, whose expected points are based off of actual game DnD situations, states that the 40 yard run is worth 2.42 points while the two 20 yard runs are worth 2.28 points (each worth 1.14 pts individually).

From your own 10 the 40 yard run is 2.25 pts while the two 20 are 2.22 pts.
From your own 20 the 40 yard run is 2.32 pts while the two 20 are 2.2 pts.
From your own 30 the 40 yard run is 2.42 pts while the two 20 are 2.28 pts.
From your own 40 the 40 yard run is 2.64 pts while the two 20 are 2.44 pts.
From the 50 the 40 yard run is 2.7 pts while the two 20 are 2.56 pts.
From opp 40 the 40 yard TD run is 4.34 (assuming PAT) pts while the two 20 are 2.84 pts.

The 40 yard is always better in terms of scoring points than two non consecutive 20 yard runs from the same place. The only reason that the two 20 yard runs are better in the DYAR world is that shorter runs are more predictable so they weight them more.

149
by Bowl Game Anomaly :: Wed, 08/21/2013 - 11:28am

It's not that they are more predictable, exactly. It's that it is easier to say with confidence that a 20 yard play is the result of a quality offense than it is to say that a 40 yard play is the result of an offense that's twice as good. DVOA is a measure of offensive quality, and DYAR is built from DVOA.

Let's take it to the other extreme. Let's say that in Week 1 the Lions pass 60 times and only rush 1 time, but that run is 99 yards for a TD. DVOA will say (incorrectly) that, based on that single play, the Lions have the best rushing offense in the league. Do they really? DYAR will correctly downgrade that play. It is the the most valuable run of the week, but probably not more valuable than the performance of another back who runs 5 times for 5 first downs. The Lions can only score a maximum of 1 TD from that 99 yard play, while the other team (the Steelers) could potentially score up to 5 TDs, because run plays are not always consecutive and those runs could have come on different drives. The Lions back scored on his 1 play, but the Steelers back gave his team 5 chances to score.

If you still prefer EPA as a measure of individual player production, more power to you.

125
by Duff Soviet Union :: Sun, 08/18/2013 - 8:38am

If RoboHoard gains three yards on every play, won't his team end up punting on their first set of downs unless they're coached by someone with balls? If RoboHoard gained 4 yards on every single play, I think DVOA would love him and his name would be changed to RoboEmmitt.

If RoboSanders' three rushes come consecutively, I imagine DVOA would actually love him because he's consistently gaining 15 yards on 3rd and 10, which is a pretty good play. If on the other hand, RoboSanders is dependent on his QB (RoboMitchell?) consistently converting 3rd and 10's to keep drives alive and give him the chance to gain 15 yards on his next carry, then I don't blame DVOA for being sceptical of him. Who's really carrying the offense in this example?

As for the point about 2*20 is greater than 1*40, I think it's more a case of two twenty yard runs is better than a 40 yard run followed by a 0 yard run, which makes obvious sense to me. Runner A has given his team a first down 40 yards down the field from where they were two plays ago. Runner B has also got his team 40 yards down the field, but they're now facing 2nd and 10 instead of 1st and 10. I think it's pretty logical that Runner A is slightly more valuable than Runner B.

127
by Will Allen :: Sun, 08/18/2013 - 10:10am

I think what DYAR misses is that a player who is the significantly greater threat to get a 60 yard run, and a td, makes a defense do things that the defense would rather not do, which makes other avenues of attack more likely to succeed. It's really hard to test this theory, however, which is why I want to put Smith and Sanders in my time raveling DeLorean, and then force a trade.

129
by sn0mm1s :: Sun, 08/18/2013 - 11:30am

No, in DYAR two 20 are worth more than one 40.

133
by dbostedo :: Mon, 08/19/2013 - 2:26am

Is this getting back into the "predictive" and "explanatory" dichotomy of DYAR/DVOA?

I.e. the 40 yard run IS just as good, in a explanatory, what-did-happen way. But in order to also make DYAR/DVOA more predictive, you have to discount longer runs because on average across all running backs for baseline purposes, 40 yard runs are slightly less predictive. So they then don't count for as much as multiple runs that end up with the same result.

Not sure if that's correct, but it could make sense if that's the issue.

134
by Will Allen :: Mon, 08/19/2013 - 8:58am

That is what gets missed; the baseline of all backs is not terribly representative of the sort of back who is a threat to score a touchdwn every time he takes a handoff.

141
by mrt1212 :: Mon, 08/19/2013 - 4:46pm

What is the rate at which

a. these long rushes happen
b. how predictive are they across the board

you dont make special models for outliers.

142
by Scott C :: Mon, 08/19/2013 - 4:53pm

Right. Adrian Peterson and Ladanian Tomlinson (and others in the 90's) faced 8 men in the box all the time.

Peyton Manning's RB did not. Barry Sanders did not often have a passing attack to divert defenders attention, but Emmet and Davis did.

This doesn't only affect RBs, it also strongly affects the passing game and adjusted line yards.

Adjusting for the quality of the defense doesn't cut it either, because the defenses play differently.

DVOA/DYAR help view these players in new ways, but they certainly don't account for all of the context.

135
by sn0mm1s :: Mon, 08/19/2013 - 9:10am

Yes, that is true - which is why I don't like DYAR as a stat to rank individuals. DYAR is fine tool to predict team success. DVOA is slightly better at ranking individuals but it still suffers from diminishing returns.

136
by Aaron Brooks Go... :: Mon, 08/19/2013 - 9:31am

"If RoboHoard gains three yards on every play, won't his team end up punting on their first set of downs unless they're coached by someone with balls?"

Sure. And RoboManning only converts a two-minute drive if RoboFox doesn't always take a knee. Even DYAR cannot account for bad coaching.

But RoboHoard gets 3 yards every time, whether it's 1st-10 or 4th-1.

126
by Will Allen :: Sun, 08/18/2013 - 10:03am

Would a decent baseball analogy be a metric which declared Mo Rivera the best pitcher ever, period, without recognition that getting to the ninth inning with the lead is pretty important?

137
by Aaron Brooks Go... :: Mon, 08/19/2013 - 9:43am

Rivera is a good one. He was a bad starter in 1995, who became a middle-innings/setup guy for the end of 1995 and the beginning of 1996, who took over closing in 1996 and never looked back. DVOA-like metrics love him.

Consider him in comparison to John Smoltz. In his few years of closing, while recovering from an injury, he became a closer every bit as dominant as Rivera, but unlike Rivera, was also a borderline HOF starter. He was as good as Rivera at Rivera's job, and better than Rivera at Smoltz's job. But according to efficiency metrics, Rivera is the better pitcher.

This sort of thing is the flaw w/ scaled DVOA metrics -- there's a fundamentally flawed assumption that player roles are perfectly interchangeable.

As an aside, here's a fun one -- in 2007, Joba Chamberlain turned in the single greatest season, on an adjusted ERA basis, in major league history. As the rest of his career showed, efficiency metrics reward small sample sizes.

138
by Bowl Game Anomaly :: Mon, 08/19/2013 - 12:18pm

To be fair to Rivera, there are a lot of pitchers who have been dominant closers for a few years (like Smoltz) but hardly any who were dominant closers for a long career. And Smoltz was a bit of a freak too, because there has probably never been anyone who was as good as he was at both starting and closing.

139
by Aaron Brooks Go... :: Mon, 08/19/2013 - 12:49pm

Eckersley was a pretty good starter for a bunch of years before beginning his more famous career as a closer.

Trevor Hoffman basically had Mariano's career, he just dwelled in secrecy down in San Diego.

143
by Scott C :: Mon, 08/19/2013 - 4:59pm

As a Padre fan, I remember talking to a Yankee fan about the two around 2002. He said "What makes Rivera so special is he doesn't rely on a 100 mile an hour fastball, compared to guys like Hoffman". Ignorance is bliss.

Yeah, their careers were more similar than most know. Hoffman barely reached 88mph on the fastball in the prime of his career, and his killer pitch was his change-up.

Perhaps there is something to longevity in a closer if they don't rely on speed and instead rely on a killer off-speed pitch + command.

150
by DSFC (not verified) :: Wed, 08/21/2013 - 1:29pm

Is this what happens when football guys try to analyze baseball? Stick to what you know.

Trevor Hoffman - career ERA of 2.87, career ERA+ of 141. bWAR of 28.0 and fWAR of 23.0
Mariano Rivera - career ERA of 2.22, career ERA+ of 204. bWAR of 55.7 and fWAR of 39.5

Hoffman cracked 200 in ERA+ only twice (not counting his injury-shortened 2003 when he threw 9 innings): 265 in 1998 and 226 in 2009. Mariano's top ten: 316, 308, 267, 262, 257, 252, 241, 240, 239 and 233. Hoffman had ONE season that was on the level of Mo's peak. His career ISO allowed is .079. No other pitcher EVER - not Hoffman, not Pedro Martinez, Tom Seaver, Sandy Koufax, Lefty Grove, no one - has one under .100.

Regarding Smoltz....I don't think he was was quite equal to Mo at his best, but he was in the ballpark. Still, what truly sets Mariano apart is the fact that he's never had a bad season (other than as a starter as a rookie). There have been guys like Eric Gagne or Jonathan Papelbon or Craig Kimbrel who have matched Mo's best seasons, but none of them have been able to do it for more than a few years. Hoffman was very good for a very long time, but he was never on Mo's level of dominance. When you combine the brilliance of his seasons indivdually with his consistency and longevity, there is no one anywhere near him. Wayne Gretzky is the only guy in team sports I can think of who so outdistances his peers.

69
by chance47 :: Fri, 08/16/2013 - 12:41pm

I recall Texas coach John Mackovic saying after the NFL did not draft Holmes that "He is going to make some team a fine running back". Understatement! How was he undrafted? Priest was lost in the shadow of Ricky Williams back then and had lost a year to injury.

75
by Ryan D. :: Fri, 08/16/2013 - 1:34pm

As a bitter UNC fan, I still dislike Holmes for his 161-yard, 4-TD day against Carolina in the 1994 Sun Bowl.

92
by Bobman :: Fri, 08/16/2013 - 3:47pm

Man, that's quite a grudge you are holding. We must be related.

101
by Ryan D. :: Fri, 08/16/2013 - 8:27pm

I also like Peyton Manning. Does that help? :)

96
by chance47 :: Fri, 08/16/2013 - 5:49pm

Was Mack Brown the NC coach in that game? A little extra irony. Of course, one could argue losing Mack to Texas wasn't so bad for NC...

102
by Ryan D. :: Fri, 08/16/2013 - 8:35pm

He was. Personally, I was very disheartened when Mack Brown left UNC for Texas, as he was the best coach and recruiter UNC had ever seen. No one that has coached since he left has coached or recruited nearly as well as Coach Brown did. His mid-90's UNC defenses were loaded with NFL talent. He had a few 1 and 2 loss seasons, back when FSU was basically unbeatable.

Once he left for Texas, I hated everything about Texas so much that I pulled for USC in the Rose Bowl, if only out of spite against Mack Brown. I also disliked Vince Young, solely because he played for Mack Brown at Texas. I still think Matt Leinart is a better QB, probably because of this.

Yes Bobman, I hold a grudge. :)

91
by Bobman :: Fri, 08/16/2013 - 3:47pm

Their numbers here are very close, but whereas noted NFL talent evaluator Peyton F Manning is on record as saying Edge was the best teammate he ever had (and his lunchpail work ethic was legendary despite being Mr. Laidback when not on the job), I don't know of ANYBODY not named Ronde who might have ever said Tiki was the best teammate I ever had.

93
by Will Allen :: Fri, 08/16/2013 - 3:58pm

I'm pretty sure the former Mrs. Tiki Barber agrees. Heck, I'm not sure that, if there is a former Mrs. Edgerrin James, that SHE doesn't agree! Hell, I'm not even all that sure that Ronde doesn't agree!!

The Tikimaster is about about on par with Congress, and various herpes virus afflictions, in terms of popularity.

108
by Raiderjoe :: Sat, 08/17/2013 - 1:52am

Ranling six guys up to list beginning of article.

Based on geratness-
1. P. Holmes
2. tomlinson
3. T. Davis
4. Barber
5. James
6. Bettis

110
by Raiderjoe :: Sat, 08/17/2013 - 2:13am

After read aryivle will krep list same.

Found Holmes to be really good at Texas. Suprirsed not drafted. Guy was good with crappy Ravens tema and then was good backup on Super Bowl cjampion ravens team. Go to Chiefs and was great.

Best 4 Raiders divisional opponents RBs all tome
Curt Warner
P. Holmes
Tomlinson
T. Davis

Tomlinson has longevity over Holmes and Davos but all 3 close in peak year greatness. Just gooing talent, have Holmes ahead of other two.

James really good but didn't have extra juice. Not many long runs. Veryy goof player. Good pass catcher.

Barber freat late in career. Ho him early. Jerk of person.

Bettis very overrate d. Media pumped guy up bebcasuse play for Notre Dame and then steelers and was nice guy.

Would vote for Holmes , Tomlinson for hall. Davis short carrer do go back and forth with him. Would not vote for Natber pr James for hall but would not be mad if any of them get in.

Think Bettis in Hall is most questionable thong but guy play for Steelers, singlehandledly win Super Bowl 40 in hometown and has crappy overrated nicknaem that many peiple think is great, so will get in

151
by skd (not verified) :: Thu, 08/22/2013 - 5:31pm

Facing 8 and 9 men in the box actually make it easier to break off long runs. Once you get past the first line the defense is playing catch up. When Edge played for Indy, the defense was almost exclusively in two or more deep zone coverages, with line backers zoning underneath. With people lined up all over the field you have to use much more patience and intelligence to gain 7 and 8 yards like edge typically would. Rather than man on man 8 man fronts where you have more of an opportunity to use your speed change of direction when the defense over pursues. A good example of this is Frank Gore. He spent most of his career running into 8 man fronts with bad passers behind him ala Adrian Peterson. He would break a lot of outside runs and cutbacks. In the last few years you really see him using his patience and vision to set up interior defenders, a skill no one really one he had because of the lack of interior space early in his career. I guess my point is it takes two completely different styles and mentalities to run against the vastly different defenses you see when playing with a great and a below average qb. One looks more spectacular but to diminish what Edgerrin James did on an every down basis mean you didn't consistently watch him and you just punched some stats into a formula.