On Monday I walked through historical data showing that *most* high-volume RBs regress the following season. These regressions are sizable enough to make even the most savvy managers anxious. I'm not even entirely comfortable with my research. Last summer, I still had Jonathan Taylor as the overall RB1, despite his 372 touches. I tepidly warned last July 1st that he was a "great bet for at least a 10% regression."
Cool. But he was still my RB1, because I couldn't justify putting anyone above him. His points per game ended up plummeting 39%. I followed the data and responded to the data, but didn't *embrace* the data.
I've got companion research to the 350+ touch regression piece: RBs coming off a 600+ offensive-smap campaign. This is intended to bolster conclusions. RBs who show up in both sets are deemed the highest risks, for hopefully obvious reasons. All three running backs coming off 350+ touch campaigns entering last season also were coming off 600+ offensive-snap campaigns: Taylor, Najee Harris, and Joe Mixon.
We know what happened. Taylor (as pointed out above) tanked even if we exclude the time he spent hurt. Harris (preseason RB5 ADP) finished as the RB20 in points per game. Mixon (RB7 ADP) was better than expected (RB6 in points per game) -- though it should be said that 23% of Mixon's production came in one week, while he finished outside the top 18 in half the weeks he played. And he still dipped 5% in points per game compared to the previous year.
Those of you who drafted Mixon in the first round can say if you were happy with the decision. The point is, he's an outlier -- a "success" -- compared to others found in this research. And at best, only barely.
So what does this research show? 77% of RBs who earn 600+ offensive snaps (regular season and playoffs) score fewer fantasy points the following season. That percentage increases to 80% for 700+ snaps and to 81% for 800+ snaps. Interestingly, the percentage levels out to 79% for 900+ snap RBs.
The net fantasy-point decline for 600+ snap RBs is 63 year-over-year. That jumps to 83 for 700+ snap RBs, and then 100 for 800+ snap RBs. 900+ snap RBs average a 120-point decline.
And age matters here. Older RBs (defined as 28+ years old) face significantly larger declines than younger RBs (21-23 years old). For 600+ snap running backs, it's an average 45% decline for the old guys, and an average 13% decline for the young guys. Interestingly, that gap tightens as we isolate for higher snap counts. And by the time we reach 900+ snaps, the young guys (22%) experience bigger declines than the old guys (18%).
Perhaps this is a survival-of-the-fittest concept. Old guys who earn 800+ or even 900+ snaps probably have done it before, and they're still in the league -- still playing huge roles for their respective team. But young guys haven't yet demonstrated that they can handle massive snaps. There's more variance in their results.
This data goes back to 2012, which is the first year snap counts starting being tracked online. In the 10 seasons from 2012 to 2021, an RB has hit 600+ snaps 146 times. But only 14 of those times has an RB cracked 900 snaps. It's that rare. In fact, the 1,000-snap plateau has been cleared only four times -- most recently Najee Harris in 2021, and before that, CMC in 2019. We know what happened next.
A whopping 22 RBs had 600+ snaps last year. The top four were Saquon Barkley (955 snaps), CMC (914), Dalvin Cook (891), and Josh Jacobs (821). I've been publicly fading Saquon, Cook, and Jacobs all summer. The data shows they are excessive regression risks. That's why Cook signing with the Jets didn't impact him much on my draft board. Yeah, he moved up, because he has a home. But he's nowhere near his ADP.
And that brings us to CMC. He's like Jonathan Taylor last summer. I can't bring myself to trust the data enough to knock him down. I need to square this circle in a hurry. Combined with his 381 touches (including the playoffs), he's a no-brainer regression candidate. I'm giving myself until the end of this week to make a final determination. Trust the research, or make an exception?
Looking forward to your input.
---