Projecting Duke Johnson in Houston

The Duke Johnson trade carries a number of ramifications for other skill position players on both the Browns and Texans, but this post will try to zero in on Johnson’s own range of outcomes on his new team. That necessarily entails projecting Lamar Miller as well, so you’ll also have that to look forward to.

I should disclose at the start that I am a longtime fan of Johnson going back to his days at Miami, and I earnestly think he’s one of the best running backs in the league. I’m such a true believer that I rank Johnson’s talent above that of brief former teammate Kareem Hunt, and more or less believe he can do most things Alvin Kamara can. My interpretation of Johnson’s 2019 upside is probably more enthusiastic than just about anyone’s, so account for that bias however you will.

But I was also a Miller supporter for most of his career, even if I’ve become less interested as his Houston career stretches onward. I was among those irritated at how the Dolphins marginalized Miller, and I was an early investor when he made his way to the Texans. So as much as I might be a superfan of Johnson’s, it’s not as if I hold Miller in any sort of contempt.

With that said, I absolutely consider Johnson the better talent between the two, and I consider his arrival bad news for Miller. As much as I expect Miller to remain the technical starter at running back for Houston, I suspect Johnson’s immense talent advantage over Alfred Blue means that Johnson will earn a higher share of snaps and usage than any of Miller’s prior backups.

I’d like to otherwise start off with a comparison of their prospect profiles. The short story is that Miller has better athletic tools, but Johnson has better skills and production. Miller ran a 4.40-second 40-yard dash at 5-foot-11, 212 pounds, and his two 97-yard touchdown runs in the NFL are a testament to his long speed on the field. Duke ran only a 4.54-second 40-yard dash, and at a lighter 5-foot-9, 207 pounds. That’s a win for Miller, no doubt. Even today with Miller at 28 years old and nearly 10 pounds heavier, you’d have to pick him in a sprint.

On a football field, though, there’s reason to think Johnson is the greater cumulative threat. Both of these guys played college football at Miami (FL), where Miller went pro following his 2011 redshirt sophomore season and Duke, then the top-ranked running back recruit, succeeded Miller in 2012. In Miller’s 23 career games he totaled 1,918 yards (5.7 YPC) and 15 touchdowns to go with 28 catches for 181 yards and a touchdown. In 33 career games Johnson ran for 3,519 yards (6.7 YPC) and 26 touchdowns, adding 69 receptions for 719 yards and four touchdowns.

If Miller is supposed to be a better pure runner than Duke, wouldn’t we be anxious about a discrepancy of a full yard in their respective rushing averages? Particularly when Duke’s yardage volume was nearly twice that of Miller? The team context doesn’t do much to absolve Miller – his rushing production was a 43.78 percent share of Miami’s rushing yardage those two years, on a team averaging 5.12 yards per carry when adjusting for sack yardage. Duke did his damage on a much greater share of 53.35 percent on a team that averaged 5.5 yards per carry as a whole. It’d be reasonable to posit that the 0.38-yard difference between the two teams is accounted for merely by the two respective runners, but even if not, a 0.38-yard team distinction doesn’t absolve a runner for placing a full yard per carry behind another, especially when the higher YPC runner takes on a greater team burden and executes it over a substantially greater volume.

That brings us to their NFL production to date. Playing on some truly awful offenses in his first three NFL seasons, Johnson nonetheless boasts one of the best pass-catching backgrounds among NFL running backs, turning 303 targets into 235 receptions for 2,170 yards, good for a catch rate of 77.6 percent at 7.2 yards per target. That’s elite efficiency, and over a substantial volume level. In fact, last year was the first where Duke failed to exceed 500 receiving yards in the NFL. To Miller’s credit, he’s a non-liability as a pass catcher. After doing very little in that capacity in college, Miller turned 271 NFL targets into 209 catches for 1,565 yards (77.1 percent catch rate, 5.8 YPT), though the much lower YPT and the lower volume (Miller has never exceeded 400 receiving yards in his seven seasons) illustrate Duke’s emphatic lead in this category.

RotoWire has the best daily fantasy football tools on the web.
Try Our NFL Lineup Optimizer Now

If Miller is better than Duke at anything, other than the previously mentioned footrace scenario, it would have to be running between the tackles. Duke’s career YPC is 4.301, while Miller’s is 4.331. Miller did that over 1,354 carries – substantially greater volume than Duke’s workload of 299 carries. I’m not ready to concede that Miller is definitively better in this aspect, though, especially as he gets closer to age 30, and I base that mostly on Duke’s easily superior college production. It’s also worth mentioning that Duke averaged 3.6 yards per carry in his rookie season, but 4.65 yards per carry since then.

What makes the pairing interesting is that their skill sets are different enough that it won’t be an entirely zero-sum contest between them. That’s primarily because Duke can play wide receiver well. He played roughly ¼ of his snaps last year lined up in the slot or out wide, so the two could see snaps at the same time if the Texans go four wide or any of DeAndre Hopkins, Will Fuller, or Keke Coutee needs a breather. You couldn’t say the same about Alfred Blue, whose snaps and function almost necessarily occurred at Miller’s expense.

To close out this article I will try to put forth a rough projection for these two runners. To start that process, we’ll note that in Miller’s three seasons with Houston he averaged roughly 667 snaps per season, or 45.46 snaps per game. Over those three years Houston ran 3,138 plays from scrimmage, meaning Miller claimed 63.74 percent of the team’s snaps. Miller played 14 games in two of those three seasons, totaling 624 snaps in one and 619 in the other. In the other of the three seasons he played 16 games, and 757 snaps within it. So we can reason that Miller’s cap is about 760 snaps if he plays 16 games in 2019. I feel grim projecting injuries, so I’ll give him those 16 games. But because Duke is so much better than the likes of Alfred Blue, D’Onta Foreman, and Tyler Ervin, I’ll take the liberty of projecting Miller for 62 percent of Houston’s snaps rather than the 70 he might approach otherwise.

If we take the three-year average for Houston’s snaps from scrimmage and use that for the team projection, we get 1,046, which would leave Miller with roughly 650 snaps. Bill O’Brien prefers to use only two backs in a given game, so I think it’s reasonable to leave Duke with at least 360 of the remaining 396 running back snaps, while giving Duke another 120 snaps at receiver for 480 total.

Miller averaged 0.358 carries and 0.046 targets per snap over the last three years. I’d maintain the rushing rate and decrease the target rate to 0.04 or so, leaving us with roughly 233 carries and 26 targets. I’m going to project 4.0 yards per carry for Miller – his average in Houston is 3.97 if you take out his 97-yard touchdown run from last year – and that would give us 932 yards. A catch rate of 76.25 percent would leave Miller with roughly 20 catches in this scenario, which would leave him with 150 receiving yards if we apply his career average of 7.5 yards per catch. Miller has found the end zone on less than two percent of his Houston carries for whatever reason, but if we generously bump him up to two percent we get five rushing touchdowns. With a touchdown rate of roughly 5.5 percent on his receptions in Houston, I’ll apply that same rate here and get one receiving touchdown as a result.

So that would leave us with 932 yards and five touchdowns rushing, and 20 catches for 150 yards and a touchdown receiving for Miller.

Duke’s projection is tougher since O’Brien has never utilized a player like him before. It’s possible that not even O’Brien knows what he’ll do next. It’s also worth recalling that with Hopkins, Fuller, and Coutee present, Houston’s wide receivers demand a far greater share of Houston’s offense than Duke’s wideout teammates in Cleveland did. Instead of looking to Duke’s rushing usage trends from Cleveland, where he averaged 0.146 carries per snap, I’ll split the difference between his Cleveland per-snap rushing usage and that of Miller, giving us 0.252 per snap. That would project for 121 carries in this scenario. Duke averaged 0.148 targets per snap in Cleveland, but if only to justify the acquisition cost of a third-round pick, I’ll project 0.15 targets per snap with Houston, resulting in a 72-target projection.

Because I’m a Duke optimist, I’m going to take the liberty of projecting 4.7 yards per carry in this hypothetical, which would give him 569 yards rushing, while a modest touchdown rate of two percent would project for 2.4 rushing touchdowns. If we apply Duke’s career catch rate (77.6 percent) and YPT average (7.2), we get 56 receptions for 518 yards, and if he can score three times on 47 catches last year then I feel safe projecting another three receiving touchdowns in this context.

To summarize the results of this projection:

Miller: 932 yards, five touchdowns rushing + 20 catches for 150 yards and one touchdown

Johnson: 569 yards, two touchdowns rushing + 56 receptions for 518 yards and three touchdowns

In standard scoring, Miller would average 9.01 points per game while Duke would average 8.67. In 0.5PPR Miller would average 9.64 points per game compared to 10.42 for Duke. Based on this hypothetical, I’ve managed to convince myself that I clearly prefer Duke in PPR scoring, where Miller would average 10.26 points per game versus 12.17 for Johnson.

If you take the 194.7 PPR points I project for Johnson and compare it to last year, he’d rank just behind runners like Derrick Henry (201.4) and Chris Carson (201.4, 14 games) while finishing just ahead of Tevin Coleman (193.6) and Adrian Peterson (189.0). Keep in mind, this projection is assuming a 16-game season for Miller. If Miller should get nicked up, there should be upside for more from Duke than this projection implies.