There's a lot of strong dynasty analysis out there, especially when compared to five or ten years ago. But most of it is so dang practical-- Player X is undervalued, Player Y's workload is troubling, the market at this position is irrational, and take this specific action to win your league. Dynasty, in Theory is meant as a corrective, offering insights and takeaways into the strategic and structural nature of the game that might not lead to an immediate benefit but which should help us become better players over time.

Better Living Through Betteridge

Betteridge's law of headlines states that "any headline that ends in a question mark can be answered by the word no." It's a commentary on the incentives of publishers-- readers are more likely to read about a surprising result than an unsurprising one, but surprising results are more likely to be wrong.

Is getting a big workload a bad thing for a running back? No. We want backs who score fantasy points. Backs can only score fantasy points when they touch the football. Therefore, touching the football more often is good, not bad. Duh.

Despite the obviousness of this observation, there have been many through the years who would have posited that touching the football more is actually kind of bad. After all, backs can only get injured and wear down when they get hit, and they only get hit when they touch the football.

Since this is a dynasty column, you're probably most familiar with the "touching the football is bad, actually" argument by another name: "mileage". The idea is that running backs are like cars-- they wear down with use and only have a certain number of miles available before everything starts to fall apart. Certain cars are built better than others and good maintenance can extend a lifespan, but all else being equal, if presented with two identical cars from the same model year, you'd expect to get more longevity out of the one that has driven fewer miles.

Sometimes this will be explained using another car-based analogy-- "tread on the tires". Tires have a certain lifespan, and while driving conditions can impact that (driving aggressively in start-and-stop traffic on bad roads will greatly decrease a tire's longevity), every mile they travel brings them closer to the end.

(Confusingly, people tend to get this analogy very wrong. Tread on tires is a good thing-- tires have the most tread when they're brand new, and that tread wears down with use. Yet people will often discount running backs with large workloads because they "have a lot of tread on their tires". It should be that they "have little tread left on their tires", or perhaps "they have a lot of wear on their tires".)

Either way, both analogies operate under the assumption that people work the same way cars work.

Are People the Same as Cars?

Betteridge's law of headlines: no, people are not the same as cars. In fact, they work quite differently.

I've seen a lot of studies over the years that sort running backs by career carries and track how much longer they had left. They all find that the more carries a running back has had, the less time he has left in his career. So mileage is bad, right?

Except if you don't very carefully control for confounding variables, such a study is meaningless. At the most extreme, you wind up in a situation where you're comparing Emmitt Smith at 500 career carries with... Emmitt Smith at 1,000 career carries. Of course, 500-carry Emmitt Smith had more career left in front of him than 1000-carry Emmitt Smith. (Without checking, I'm willing to bet he had exactly 500 more carries remaining after that point.)

For humans, the thing that ultimately predicts decline is age. The specifics vary depending on the field, but every endeavor has a point where participants age out. For something incredibly demanding, like Olympic gymnast, most careers are over by the mid-20s. For a much easier job, like President of the United States of America, we routinely see participants into their 70s or even 80s.

The NFL is no different; different positions have different benchmarks, but everyone eventually ages out. Given that accumulating carries takes time, "career carries" often winds up being a proxy for age, and when one finds that past mileage predicts decline, one is really just finding that older players are nearer the end of their careers than younger ones.

To guard against this, you need to design a study in such a way that you're controlling for age-- comparing players of similar ages but disparate past workloads to see who lasted longer. And every study I've ever seen that controlled for age found that past workload was meaningless (or even slightly positive) for predicting player longevity. I ran down the full list of evidence two years ago.

The inspiration for that article was Rhamondre Stevenson, who was wrapping up an RB8 overall finish in his second NFL season and had seen his dynasty value rise quite high as a result. By the end of the season, he had just edged past Josh Jacobs despite the latter finishing 21st, 8th, 12th, and 3rd in his four NFL seasons, outscoring Stevenson 330.3 to 251.1 in Stevenson's own breakout year.

"Dynasty managers prefer the younger RB" is an unsurprising story, but despite being in the league for two fewer seasons, Stevenson was only 12 days younger than Jacobs. Instead, the reason most managers were taking Stevenson over Jacobs was because they were worried about Jacobs' mileage. He had 1232 career touches to 426 for Stevenson.

Those who worried Jacobs' usage heralded a decline were right; he missed four games in 2023 and his point per game average fell to 14.1. But Stevenson missed even more games and his average was nearly 2 points lower. This year Jacobs has rebounded, ranking 2nd in the NFL in touches and 9th in points per game while Stevenson continues to disappoint. Fortunately, the dynasty community has learned from this error and will never again doubt a running back because he has a lot of mileage.

Just kidding.

RB Ashton Jeanty (Boise) entered today with 651 career carries.

He’s at 21 carries today and it’s only the first half. With at least two games remaining this season, it’s fair to assume he’s going to eclipse 700 carries in three seasons.

It’s not ideal, but not terrible. https://t.co/2B3vMeZ5mS pic.twitter.com/okOYwUIUID

— Dominic White (@DomWWhite) November 29, 2024

Are College Players Like Cars?

Betteridge once again: not only are college players not like cars, they're even less like cars than NFL players. There are fewer studies on college workload, but the ones that exist typically find that even without controlling for age, it predicts nothing about NFL longevity.

Many find this puzzling, wondering whether college tackles aren't as damaging as NFL tackles or whether teenagers are just better at recovering from the workload. Personally, I think it's more a function of how NFL longevity for players still in college is driven mostly by whether they are NFL-caliber or not (a question that no longer needs to be asked once a player is 24+ with 1,000 career carries), with a dash of "even real relationships are hard to find when you restrict the range". (The difference in ages between college players is much smaller than the same difference between NFL players.)

Ashton Jeanty is having the best season by a college running back since Barry Sanders. The fact that he is racking up a lot of carries while doing so doesn't matter because "mileage" is predictively useless.

Is Workload Useful in the Short Run?

Betteridge again: nope.

The idea that single-season workload was a negative indicator arose in 2006, when Football Outsiders wrote an essay on Shaun Alexander and the so-called Curse of 370 -- the tendency for running backs who receive 370 carries in a season to fall off precipitously in the following year. 

If your first thought on reading that is that 370 seems like a suspiciously arbitrary threshold, give yourself a gold star; it turns out that backs who received between 357-369 carries had some of the most productive follow-up campaigns in NFL history. In 2008, Brian Burke wrote the definitive investigation on the so-called curse and found the evidence lacking.

Most importantly, the idea that workload mattered originated in an era where running back workloads looked very different than they do today. From 2000-2008, there were ten seasons with 370 carries. In the fifteen years since, there have only been two.

Drop the threshold a bit and the story remains the same. From 2000-2008, there were 41 backs with 330 carries, an average of 4.5 per season. From 2009-2023, there were 13, an average of 0.9 per season. And two of those thirteen needed a 17th game to get there.

To the extent that huge workloads were genuinely concerning, the NFL has realized and adjusted to the point where it rarely gives them anymore.

What About In-Season?

Does workload in the first half of the season predict injury in the second? 

I took every back from the last five seasons who played in each of his team's first 6 games and recorded at least 80 touches. This gave me 113 running backs, or about 22.5 per year. I sorted them from most to fewest touches and divided them into four groups of roughly equal size, then tracked how they did over the remainder of the season. (To avoid any impact from teams resting their starters with playoff seeding locked up, I only tracked each back over the team's 7th-15th games, so the best a back could do was play in nine more games.)

Here are the results:

Some might point out that the highest workload in the sample belonged to Derrick Henry, who averaged 29.7 carries per game through six weeks in 2021 and only lasted two more games before he got injured. That's true. Derrick Henry also had the 3rd and 4th highest totals in the sample in 2020 and 2022, and he played all season in both years. Overall, injuries happen to backs at every workload level at pretty much the exact same rate.

So Workload Just... Doesn't Matter?

Is this the question that finally foils Mr. Betteridge? Even after all of this, the answer is once again no. If coaches gave De'Von Achane as many carries as Derrick Henry, he might disintegrate. And that's really the key thing to keep in mind-- workload is not assigned randomly. It is given out by coaches who theoretically know their players and want to put them in the best position to succeed.

Backs of all workloads have fairly similar injury rates and longevity because all backs tend to be used fairly optimally. And the one counterexample buried amid all the studies-- the one well-designed investigation that found workload predicted future decline or injury-- is illustrative.

If injury rates aren't correlated with workload because coaches are using running backs optimally, in what situations might we expect coaches to use their running backs suboptimally-- to push past what a running back could easily bear?

In 2011, Jason Lisk posited that coaches were most likely to push too far late in close games. He hypothesized that a running back who had 30 carries in a blowout was less likely to get hurt in subsequent weeks than a back who had 30 carries in a one-score game. And when he investigated, he found that this was the case, a fact I noted in Week 1 when Joe Mixon had 30 carries in a 2-point Texans win.

30 carries for Joe Mixon in a 2-point win. @JasonLisk once found that high workloads in close games were more likely to translate to injuries down the road than in blowouts-- with the theory being that coaches were more likely to give "too many" carries when the stakes were high.

— Adam Harstad (@AdamHarstad) September 8, 2024

Mixon had 9 carries for 25 yards the following week, was injured, and missed the next three games.

It turns out that those arguing that mileage matters were right. There is such a thing as "too many" carries. But they were wrong to assume that the values we observe qualify as "too much". To quote Jason Lisk: "If [coaches] are going to overdo it with a player, it will be just barely, with an extra few touches late."

I understand the appeal of the idea of mileage. I wish as much as you do that there was an easy filter we could apply to sort the good bets from the bad ones. Unfortunately, workload isn't such a filter.