Eight weeks ago, the Vikings were undefeated and ranked No. 1 in DVOA. Now they're 6-6 and probably done. What went wrong?
14 Apr 2009
Guest column by Mike Horn
At the end of Part I of this essay, we saw that the use of workhorse back and dual-back systems has been increasing since 1960. The obvious question is, if those two usage methods have gone up, what has gone away? For one thing, as discussed earlier, RB3s and other reserve backs get far fewer carries today. For another, the use of running back committees, at least to run the ball, has declined -- in fact, it has disappeared.
A Running-Back-By-Committee (RBBC) system is defined here as an offense which has either:
This eliminates teams that have, for example, three different tailbacks who each manage to play five or six games and each get about a third of the team's rushing attempts. The idea is to find teams that regularly played three (or more) backs and gave each of them a decent share of the load. These standards probably eliminate a few genuine committees but minimize false positives.
The next chart shows how RBBC has fallen into disuse:
The black line shows the linear trend, but that overstates the steadiness of the decline. There were a lot of RBBC teams in the '60s, a slight revival of the system in the mid-'70s after a period of disfavor (the RBBC used by the '72 Dolphins may have inspired some copycats, and also see the footnote about the Broncos), then a fairly steady use of RBBC from the liberalization of the passing rules until free agency. After 1994, RBBC has quickly disappeared, at least as far as the distribution of carries is concerned. The next table shows how many teams played in those three periods, the number and percentage of them that used an RBBC system, and the maximum and minimum percentages of teams using the system in a season in each period.
|Distribution of RBBC Seasons|
Until this past season, the 2003 Eagles were the last team to have an RBBC, with Duce Staley (96 carries, 16 games), Correll Buckhalter (126, 15), and Brian Westbrook (117, 15) sharing the load. Each back led the team in rushing attempts for at least three games, and there were weeks where they split the carries almost equally (8, 7, 9 and 6, 8, 6). Before that, no team had employed an RBBC since 1997 (Chiefs, Saints, and Giants). Curiously, three of those four teams made the playoffs and they combined for a 41-22-1 record, yet no one else attempted RBBC after the Eagles -- until the 2008 Ravens. Le'Ron McClain led Baltimore with 232 carries in 16 games. The other committee members -- Willis McGahee (170, 13) and Ray Rice (107, 12) -- missed some time while sharing the load. Again, all three backs led the team in carries in multiple games; although Rice only pulled that off in weeks when McGahee was out, he was often the RB2 when all three played.
Not everyone would define RBBC as a group of backs sharing the total rushing load. In fact, a common conception for a running back committee would be a lead back, a short yardage/goalline back, and a third- or passing-down back. In other words, a collection of specialists who together perform all the functions expected of a running back.
I'd like to trace the history of the short-yardage back. He'd be a player who was not the RB1 on his team but got at least half of his team's short yardage carries. Unfortunately, I don't have situational running back statistics for the entire period of this study. So I can't definitively describe how short-yardage backs have been used over time.
The other specialist often cited in a committee approach to the running back position is the third-down or pass-catching back. Again, it would be best to study this with detailed data. I'd like to know not only which running backs caught the ball, and the down-and-distance when they did, but who was thrown the ball and who was on the field in passing situations even if they were not thrown to. Without that data, the validity of any attempt to quantify third-down back usage is going to be limited.
We can at least look at how running backs have been used in the passing game since 1960:
Teams were very gradually increasing the use of running backs to catch passes from 1960 to 1976, although that tailed off in 1977. With the rule changes in the late '70s, there appears to have been some experimentation with throwing the ball more to backs, and their receptions shot up dramatically over three years, peaking in 1980. A secondary but not unimportant factor in this increase was Bill Walsh bringing his West Coast Offense to San Francisco, which alone accounted for about an additional 0.5 running back receptions per game in the league from 1978-1979. The trend overall since 1980 has been to throw to running backs less and less, bottoming out in 2005, although there was a brief revival in running back receptions in the early '90s.
Lead backs have seen their total number of receptions stay right around two per game, although it almost reached three in 1982 (strike year) and 2002. But as the total number of passes caught by running backs has declined, RB1s have seen their share rise over the last ten or so years. Up until the mid-'90s, RB1s consistently saw a share of around 37 percent of all running back receptions, and RB2s caught about 28 percent. There were even a couple of years where RB2 as a group caught more passes than RB1s. Then over the last decade, RB1s have increased their share to an average of 44 percent vs. 22 percent for RB2s. That gap narrowed considerably in 2007-2008, down to only 41 percent vs. 33 percent this past year, mostly a case of the RB2s increasing their share. Is two years a trend?
RB3s pretty consistently have caught about 17 percent of all running back receptions, although they hit post-1960 lows the past two seasons. The other reserves typically catch more as a group than RB3s and sometimes even more than RB2s, but their numbers have fallen off in the last few years too.
So where do pass-catching backs fit into this? First, my definition: A "pass-catching RB" cannot be a team's lead back. He must play often enough to have a clearly defined role in the offense; I arbitrarily set that at 2.5 touches per game. He can run the ball more than he catches it (he is a running back, not a receiver), but not a lot more, so his ratio of receptions to rushes has to be greater than 0.5.
I tested this definition on three players who I think of as pass-catching backs: Dave Meggett, Richie Anderson, and Kevin Faulk. Together they have played 32 seasons. Together, they met my qualifications of third-down backs 21 times. In five seasons they didn't have enough touches to qualify, and in three they ran the ball quite a bit (Faulk had more than 160 carries in both 2000 and 2003; Meggett had his career-high of 92 in 1994 while also having one of his lowest pass-catching totals). Faulk as a rookie only caught 12 passes with more than 60 carries; he wasn't a pass-catching back that year. That left two seasons (one each for Meggett and Anderson) when these runners were just under the reception/rush ratio. So out of 31 seasons, my definition possibly registered a false negative twice. I could live with that error rate.
Clearly the pass-catching specialist was not a big part of the offenses of the 1960s and 1970s. But as the rules encouraged more passing from 1978 on, there became a place for this type of back in the 1980s. Also, Bill Walsh's offense not only threw often to running backs, it tended to employ at least one back primarily as a pass-catcher every year, usually a fullback, the main exception being when Roger Craig was the lead back in the mid-'80s. Another coach who favored the pass-catching back was Bill Parcells. The influence of these two successful coaches and their protégés probably helped the trend along.
By the early '90s, more than half the teams in the league had pass-catching backs, but since 1996 there has been a steep decline in their use. As lead backs (and recently, RB2s) have picked up a greater share of the decreasing number of passes thrown to running backs, the pass-catching back has fallen out of favor. This probably also corresponds to an increased specialization among fullbacks, who have become more and more just blocking backs, and are less likely to have a pass-catching function. Whatever the cause, in recent years (there was an uptick in 2008) the use of pass-catching backs has declined to levels not seen since the early '80s or even the late '70s.
Now that we've looked (with reservations) at the usage of specialist pass-catching backs, what about the all-around back? The running back who is not only the lead back, but also leads his team's running backs in receptions and rushing touchdowns. I'll call this player the "featured back."
Featured backs have been a part of the NFL since 1960. They declined a bit in popularity from the mid-'80s to the free agency period, the age of the pass-catching back, but returned to common usage after 1993. There has been a slight drop again in the last four seasons:
|Distribution of Featured RB Seasons|
|Years||Teams||Feat RBs||% Feat RBs||Min||Max|
The graphs also show how workhorse back usage has changed, and that the recent drop-off in workhorse back usage has not been exactly matched by a decrease in teams using featured backs. By the definitions I'm using, not all workhorse backs are featured backs; when this happens, it is generally because the workhorse backs don't lead their teams' running backs in receptions. They typically score plenty of their teams' rushing touchdowns.
It is very possible to have both a featured back and dual backs; 51 teams with featured backs have also met my dual-back test, about one in 12. Those are teams running a hybrid featured back/dual-RB system, vs. a "pure" version of either. The next graph shows the breakdowns of teams running pure vs. hybrid forms of the featured back and dual-RB systems:
The use of a pure featured back hit its high point in 2003 and has declined steadily since then. Occasionally from 1960 to 2004 you could find two consecutive years where more than 10 percent of the NFL employed pure dual running backs, most recently as 1996-1997, but we've now seen four consecutive seasons where that has happened. It's still far from a dominant system (14 percent of teams have used it in the last four years), but it is more common than it has ever been.
Meanwhile, 2008 saw the hybrid system pass the dual-RB system and hit a new peak in usage. That has happened occasionally since 1985 (it happened more frequently before that), so it's impossible to say if this is a trend or an aberration. But the sharp upturn and the new high suggest that some teams are considering combining the featured back with a strong RB2 who will get substantial carries without being dominant in a receiving or goal-line role.
That's a lot of data and talk about a variety of different running back usages. Not all the running back types I've described are mutually exclusive -- for example, teams can use a workhorse back to run the ball and a pass-catching back to handle the receiving chores. Certainly a top-level analysis like this using only gross statistics has its limitations. But there are perhaps some useful generalizations that I can make.
The next chart puts most of the trends on one chart, although the data is simplified and approximated in an attempt to present it more clearly:
Five periods of transition or change in RB usage prior to 2000 are highlighted:
That brings us to the last several years. A number of trends appear to have changed direction again, although this depiction may exaggerate the changes. It may be too soon to definitively conclude this, but Coach Sparano's quotation at the beginning of Part I appears to be borne out: recently, both workhorse and featured back use has declined, while dual backs have become more common. In 2006, for the first time more than 20 percent of all NFL teams used dual backs. And in 2007 more teams used dual backs than workhorses for the first time since 1990.
From 2000 on, here's a breakdown of the type backfield every team has used, how many teams have used it, the average number of wins they've posted, and their average offensive and rushing DVOA:
|Success of Different RB Systems Since 2000|
|Type Backfield||Teams||Avg. Wins||Avg. Off. DVOA||Avg. Rush Off. DVOA|
|Dual RB w/Featured RB||15||9.2||6.52%||2.31%|
|Dual RB w/WH RB||1||9.0||-11.60%||-11.30%|
|Featured Workhorse RB||53||8.9||4.76%||2.17%|
I debated how to best sort this data and ended up doing it by wins. From that perspective, the variations on the dual-RB system were the most successful, followed by the workhorse backs systems, and then the others. Note that "undefined" teams failed to meet any of my definitions. These teams were probably trying to use a particular system, but either injury prevented it, or the crude nature of a usage-stats only methodology couldn't detect it.
Teams running a dual-RB system where one of the two backs had the featured role not only were near the top in wins, they had the most effective overall and rushing offenses by DVOA of any of the categories. There have been some high-profile teams using this method, including both the Colts and Bears in Super Bowl XLI, the Cardinals from this past Super Bowl and two 13-win teams (the 2005 Broncos and 2007 Cowboys, where Coach Sparano came from). There were also the five-win 2004 Bears and 2008 Jaguars, so clearly its not a panacea. And of course. 15 teams is a small sample.
Teams with a featured workhorse had the next best offenses, total and rushing only, according to DVOA. That didn't translate to more wins, possibly because the featured workhorse consumed a lot of cap space. The two 2008 users of this system, the Redskins and the Rams, are not great advertisements for it. The "pure" workhorse teams, the Falcons and the Vikings, had better seasons, and the workhorse back teams as a group had the third-best DVOAs for total and rushing offense.
The "pure" dual-RB teams had the most wins, on average, but average offenses and mediocre or worse running games. This may reflect a cap management strategy that results in team success at the expense of the rushing game, or it may be just a small sample size problem. The win numbers were increased by the four dual-RB teams from 2008 -- the Colts, Giants, Titans, and Steelers -- who all had 12 or more wins. Only the Giants had an excellent running game (19.0% DVOA), with the Titans having a pretty good one (7.5%), the Steelers an average one (0.7%), and the Colts a poor rushing attack (-7.5%). However, Indianapolis had an excellent offense overall (21.3%), as did New York (23.4%). I haven't done a cap analysis of those four teams, but although they have some first-round picks invested in running backs, none have broken their banks to keep expensive veteran runners either.
It seems clear that having just a plain-old featured running back doesn't help an offense nor a team's chances of winning -- and the same is true of teams with backfields that don't fall into one of my defined categories.
Looking more closely at those two 2006 Super Bowl opponents, first the Colts:
|Indianapolis' RB Systems, 2005-07|
|Year||Type Backfield||Wins||Off. DVOA||Rush Off. DVOA|
|2005||Featured Workhorse RB||14||26.9%||8.0%|
|2006||Dual RB w/Featured RB||12||33.8%||7.5%|
They went from a featured workhorse back in Edgerrin James in 2005 to a dual-RB system featuring Joseph Addai in 2006, and then to using Addai in a featured role in 2007. There is not a great deal of difference among the three teams. The dual-back 2006 version had the best overall offense, the least wins, poorest rushing DVOA and the championship. When the 2007 team went to a featured back, it improved its rushing numbers. Its hard to say the any particular type of backfield was the most successful, although the pundits tend to focus on the 2006 team because it won the Super Bowl with a two-back system (no doubt some of those same pundits had predicted doom for the Colts when they let their featured workhorse walk in free agency).
|Chicago's RB Systems, 2004-07|
|Year||Type Backfield||Wins||Off. DVOA||Rush Off. DVOA|
|2004||Dual RB w/Featured RB||5||-36.5%||-14.9%|
|2005||Featured Workhorse RB||11||-16.8%||-6.9%|
|2006||Dual RB w/Featured RB||13||-3.9%||2.8%|
The pundits also took note of the 2006 Bears' use of two backs on the way to the Super Bowl. But they had used their running backs the same way in 2004 (with Thomas Jones as the featured back both seasons, swapping Anthony Thomas for Cedric Benson in the supporting role) and were terrible. In between, in 2005 the Bears improved from 2004 by using a featured workhorse system and then improved further in 2006. It seems unlikely that the continued improvement had much to do with the way the backfield was employed. The loss of Thomas Jones in 2007 and a variety of injuries contributed to the undefined nature of the Bears' backfield last year -- none of which helped the offense or the rushing game.
Note that neither team tried to stick with the formula from 2006, and both let one member of their backfield duo depart. So clearly they did not think that the two-back system was critical to making their offenses go or getting them to the Super Bowl. Both teams felt cap or other considerations were more important than sticking blindly with a particular type of backfield. Still, other teams seem, at least publicly, to favor adopting the two-back system over the workhorse back model.
It remains to be seen if finding two pretty good backs is easier than acquiring one workhorse. If this apparent trend continues, though, there will be fewer big contracts for star running backs as teams will spread their salary cap risk across a couple of medium-priced runners rather than tie up large amounts in one player who might just get hurt or decline rapidly (Shaun Alexander and Larry Johnson being the poster children). Running back career length may increase if fewer are subjected to workhorse loads. There may be fewer Hall of Fame running backs in this coming era as there may be fewer backs with big numbers -- although the backs who do post outstanding stats will seem even greater. And the workhorse backs of the late '90s and early '00s, with their large stat totals, will seem even better than the runners who followed them.
In 1977, the Denver Broncos were coming off the best record in their history, 9-5, but still had never made the playoffs. They had a new coach, Red Miller, who would take them there for the first time and, in fact, lead them to the Super Bowl, followed by three more seasons at or above .500. This was easily the franchise's best four-year period at that time. They did it with one of the more idiosyncratic running back offenses in NFL history.
Since 1960 there have been 25 times where a team's top four running backs all had between 13 and 29 percent of a team's running backrushing attempts. Only one team has done that two years in a row: Miller's Broncos. They did it four years running.
|Denver's RB Usage, 1977-80|
Two of the pass-catching running backs that sprung to my mind to test my definition were Bill Parcells favorites: Dave Meggett and Richie Anderson. They followed him through coaching moves while continuing to play a defined role in his offense. Although Bill Walsh and his West coast Offense are known for using running backs to catch short passes, Parcells may be the coach who used the most pass-catching backs.
In 19 seasons as a head coach, Parcells employed pass-catching backs 14 times, 74 percent of the time, compared to 42 percent of all teams from 1983-2006. Not only that, but Parcells some years had two players who fit the pass-catching back definition, so he actually had 19 pass-catching backs in 19 seasons.
However, he had no pass-catching backs in his last three seasons. In 2004, Richie Anderson just missed the criteria. But in 2005 and 2006, Parcells had no back that even looked like a pass-catching specialist. He shifted to more of a dual-back philosophy, although Marion Barber and Julius Jones didn't quite make the cut by this study's definition (Barber had 29 percent and 32 percent of the running back carries, just under the one-third cutoff). If Parcells, of all people, deliberately stopped using a pass-catching back, then the day of that specialist has definitely passed.
Mike Horn writes a stats-based column for FantasyGuru.com. In real life he is a retired Army officer working as a military defense analyst in the Washington, DC area. He occasionally comments as MRH on Football Outsiders discussions and has written two previous guest articles on tight end usage and how teams used draft picks over a five-year span. If you are interested in writing a guest column for Football Outsiders, please send an idea or rough draft to info-at-footballoutsiders.com.
18 comments, Last at 15 Apr 2009, 1:50pm by Mr Shush