With the same focus I had last week in contextualizing production at the wide receiver position and relating it to a team's ability to win, I've also created a method to measure running back impact.
My model uses seven seasons of data (2011-2017) to "learn" how games were won. It also compares the results from the past to the trends we're seeing today. To oversimplify how the model works: I measure how each drive got a team closer to or further from a victory.
As the game has evolved and teams have advanced in their personnel, matchup and play-calling strategies, the running back position has changed the most. A team's production at the running back position has been created in a number of ways over the past seven seasons: Some via committee backfields, some from using the back in the passing game and some from true every-down backs. Individual running back contributions therefore have become more situational.
For example, through seven weeks this season, 53 receiving touchdowns had been scored by running backs, tied for the third-most since 1970 -- and the other three years were 1983 (56), 1989 (54) and 1986 (53). The point I want to hammer home is that there are many ways to get production out of this position, and strategies vary from team to team.
Here's a quick overview of what I did to determine running back impact:
-- Using 10 seasons of historical data, I layered spatial data on top of traditional stats using computer vision. By adding what happened when a back had the ball (e.g. number of tackles broken, yards after contact, yards after the catch), I was able to better contextualize traditional stats like scrimmage yards and touchdowns.
-- Each drive was also evaluated with tracking data about what backs did when they didn't touch the ball. Actions like blocking on passing downs, and situations like facing defenses who stacked the box or drawing extra defensive attention in the red zone showed how the backs created opportunities for their teammates.
-- To define the connection between wins and the historical profiles, I enlisted the help of eight coaches and players to make sure the results had true football context.
-- In each team's system, the weighing of "on-" and "off-" ball actions of backs varies (because what is asked of them varies) but they can all be compared in terms of how much they impacted first downs and touchdowns that resulted in game-relevant scoring.
Through Week 7, here are the five most valuable backs in terms of impact, followed by a bonus five as well as five fast risers, for extra measure.
Gurley is the archetype for this metric. Not only can you can see it when you watch him play but the traditional and advanced metrics, even without context, match the eye test and the wins follow. Part of the reason the eye test works here is because his 24.1 touches per game pace the NFL, just like his 14 touchdowns. Gurley leads the league in rushing first downs per game (5.7) and 44 percent of his receptions have moved the sticks (11 of 25). Then, add the context of how he changes space second-most often of any back on downs when he's not the ball carrier or targeted (as measured by how many defenders are on him and how close they stay to him). The idea behind this metric is to measure how related a player is to wins, and no running back -- or any other offensive player who is not a quarterback -- rates higher through seven weeks.
**2018 traditional stats:** 7 games | 144 att | 686 rush yds | 4.8 ypc | 11 rush TDs | 25 rec | 270 rec yds | 3 rec TDs
</content:power-ranking>
This metric measures the chances a player gives his team to win, and both Elliott and Saquon Barkley rate at a level that far exceeds their teams win totals.
Elliott's 13 rushes of 15 or more yards lead the league and his 4.6 first downs per game rank third among backs. Defenses have accounted for Zeke significantly more than any other back, which isn't surprising if you've watched even a few snaps. Between being utilized on the highest percentage of rushes on first down (69.7 percent) in the NFL and being tied for the most passing targets (15) on third down among qualified backs (11 receptions, which ranks second), it starts to become clearer that the offense's dependence on Elliott has been huge ... and predictable for opponents. One example of how Elliott has greatly helped the Cowboys' potential to win despite unfavorable conditions is that 12 of his 13 long rushes have occurred on first down, which is four more than the next-closest back.
**2018 traditional stats:** 7 games | 132 att | 619 rush yds | 4.7 ypc | 3 rush TDs | 25 rec | 175 rec yds | 1 rec TD
</content:power-ranking>
Barkley has avoided 43 tackles this season (19 as a runner and 24 as a pass catcher), per Pro Football Focus. This is the greatest combined total and most in the receiving game for any running back. It makes sense that his 14 receptions of 10 or more yards and 60.6 receiving yards per game both pace backs. The context that my model adds shows that the broken tackles and his use in the passing game, specifically, have kept the Giants from dipping below 35 percent in terms of in-game win projection for 3-4 extra drives in each outing. This might sound confusing, but think of it like this: There are about 11 offensive drives per game and Barkley's impact alone gives the Giants a better chance to score -- up to 42 percent -- on 3-4 drives each week. This is how a RB on a 1-6 team ends up as the third-most impactful back as related to wins.
**2018 traditional stats:** 7 games | 98 att | 481 rush yds | 4.9 ypc | 5 rush TDs | 49 rec | 424 rec yds | 2 rec TDs
</content:power-ranking>
No back has broken more tackles in the run game than Hunt. The Chiefs have the most balanced offense in terms of combined impact rating at the wide receiver, tight end and running back positions, which shouldn't be surprising given their 6-1 start. Hunt's impact, especially in the run game both on and off the ball, has helped the Chiefs' defense. Last week's game against Cincinnati kind of embodies how his overall impact helps this team. Hunt's 5.73 yards per carry and 7.1 yards per touch helped keep Kansas City's defense off the field -- the Chiefs allowed the third-fewest yards (239) of any team in Week 7. This game also showed his on-ball red-zone impact (OK, it's not reasonable to assume three touchdowns is sustainable, but the fact that one was on the ground and two were receiving shows how they efficiently diversify his use in the red zone). Think of it like this: Hunt helps score points and keep the ball in the Chiefs' hands, which drives his fourth-highest running back impact.
**2018 traditional stats:** 7 games | 118 att | 542 rush yds | 4.6 ypc | 5 rush TDs | 15 rec | 226 rec yds | 4 rec TDs
</content:power-ranking>
Through seven weeks, Kamara ranks first in percentage of rushes that result in a first down among qualifying running backs this season (31.6 percent) and is tied for the third-most receptions that have resulted in first downs (14). Of the eight backs who average over 100 scrimmage yards per game, Kamara has the third-highest per-touch average (6.1) on the second-fewest touches per game (19.8). Kamara has forced 10 missed tackles as a rusher and 11 as a receiver, per PFF. Adding the contextual notes that factor in what was happening in the game when he earned his stats shows that his rushes were less about deep gashes than steadily moving the chains (only seven rushes of 10-plus yards, tied for 28th). It also shows that his ability to draw defenders into coverage on passing downs and force them to miss tackles has been the biggest driver of his impact rating. This is going to sound counterintuitive at first, but Kamara projects to be more impactful on a per-drive basis the more Mark Ingram is used because of how defenses will account for him in the passing game.
**2018 traditional stats:** 6 games | 79 att | 363 rush yds | 4.6 ypc | 6 rush TDs | 40 rec | 362 rec yds | 1 rec TD
</content:power-ranking>
AND NOW, THE NEXT FIVE ...
**2018 traditional stats:** 6 games | 91 att | 466 rush yds | 5.1 ypc | 6 rush TDs | 30 rec | 279 rec yds | 3 rec TDs
</content:power-ranking>
**2018 traditional stats:** 6 games | 78 att | 378 rush yds | 4.8 ypc | 0 rush TDs | 40 rec | 289 rec yds | 1 rec TD
</content:power-ranking>
**2018 traditional stats:** 6 games | 103 att | 453 rush yds | 4.4 ypc | 7 rush TDs | 26 rec | 257 rec yds | 0 rec TDs
</content:power-ranking>
**2018 traditional stats:** 5 games | 84 att | 386 rush yds | 4.6 ypc | 2 rush TDs | 16 rec | 100 rec yds | 1 rec TD
</content:power-ranking>
**2018 traditional stats:** 7 games | 68 att | 445 rush yds | 6.5 ypc | 2 rush TDs | 11 rec | 90 rec yds | 1 rec TD
</content:power-ranking>
NOTABLE FAST RISERS ...
**2018 traditional stats:** 6 games | 69 att | 444 rush yds | 6.4 ypc | 1 rush TDs | 15 rec | 89 rec yds | 0 rec TDs
</content:power-ranking>
**2018 traditional stats:** 3 games | 41 att | 249 rush yds | 6.1 ypc | 1 rush TDs | 4 rec | 39 rec yds | 1 rec TDs
</content:power-ranking>
**2018 traditional stats:** 7 games | 53 att | 305 rush yds | 5.8 ypc | 0 rush TDs | 19 rec | 233 rec yds | 3 rec TDs
</content:power-ranking>
**2018 traditional stats:** 6 games | 95 att | 422 rush yds | 4.4 ypc | 4 rush TDs | 4 rec | 31 rec yds | 0 rec TDs
</content:power-ranking>
**2018 traditional stats:** 7 games | 34 att | 253 rush yds | 7.4 ypc | 3 rush TDs | 0 rec | 0 rec yds | 0 rec TDs
</content:power-ranking>
Follow Cynthia Frelund on Twitter @cfrelund.
