The cliff isn't in his legs. It's in the depth chart.
What this means for your roster. Stop selling Jonathan Taylor at RB13 because he turns 28 and start asking who else is in that backfield, his age is already the biggest discount on his row. Buy Colston Loveland at TE3 and Harold Fannin Jr. at TE6, tight end is the position where a growing role is worth the most and both already hold a real target share. Pay knowingly for Kyren Williams at RB6, he is our sixth-best back on a workload the Rams have publicly said they are still splitting with Blake Corum at RB22.
What the consensus says
The sell-the-veteran genre had a big August. RotoBaller ran two separate sell lists. FantasyPros filed bust candidates and a trade-now column. Roto Street Journal built a whole roster-equity framework on peak ages, and Apex put a number on the peak: 24.8, with 93.8 percent of peak back seasons landing before 29.
They all reach for the same mechanism. Fast-twitch fades. The cuts get slower. He loses a step, so he loses points. Jonathan Taylor gets named specifically, as an age-28 back to move at peak value.
Notice what that mechanism is: a claim about the player's body. Nobody checks it. It gets asserted, and then the ranking moves.
The claim, in plain English
A fantasy total is two numbers multiplied together. How many chances you got, and what you did with each one. Touches per game times points per touch. That is the whole thing.
So the consensus claim is testable, and it is narrow: if decline is a body story, then when a player falls off, the points-per-touch half should be where the damage lives. If it is a job story, the damage lives in the chances half. I split every season into those two halves and asked which one moved.
The control is the whole game here. I compared every player only against peers who held the same share of their team's targets the year before, so a bell cow is never being measured against a backup.
How I beat on it
Every wide receiver, running back and tight end season from 2015 through 2024, paired against that same player's following season. That is 1,826 pairs across 608 players, scored in a 10-team superflex half-PPR dynasty format with a tight end reception premium. The following season tops out at 2025, so the locked 2026 holdout is never touched. Uncertainty is a cluster bootstrap resampled by player, so one guy with eight seasons cannot vote eight times. Full method and the script are here, and the framework is the same two-gate design every Lab study runs.
What the data actually said
First swing, everybody: across all 1,826 pairs the average player loses 0.68 points a game year over year. Of that, 0.52 is chances he no longer gets and 0.13 is what he did with them. Three quarters of the drift is the job.
Because aggregates lie, I ran it again on the 1,112 seasons that actually went down. Those players lost 2.80 points a game. Lost opportunity accounts for 1.91 of it. Lost efficiency accounts for 0.89. That is 68 percent job, 32 percent player.
| Who fell off | Points a game lost | From lost chances | From lost efficiency |
|---|---|---|---|
| Every decline (n=1,112) | 2.80 | 1.91 (68%) | 0.89 (32%) |
| Wide receivers (n=512) | 2.69 | 1.72 (63%) | 1.00 (37%) |
| Running backs (n=347) | 3.16 | 2.19 (69%) | 0.97 (31%) |
| Tight ends (n=253) | 2.49 | 1.93 (78%) | 0.56 (22%) |
Then the archetype test, the one the consensus mechanism actually predicts. Speed-dependent receivers, the guys whose game is supposed to be the first thing age takes, should decline differently from everyone else. They do not. At the same target share, the 131 speed receivers in the sample separate from the other 554 by 0.04 points a game, with a 95 percent range running from minus 0.48 to plus 0.51. That is a shrug with a confidence interval on it.
One group did separate, and it is the group our own error map had flagged as overpriced. Players 23 and under beat their same-role peers by 1.15 points a game, and 0.93 of that is opportunity, ranging from plus 0.66 to plus 1.22. Their role grows in all five target-share bands. Youth is not a bet on better legs. It is a bet on a bigger job.
What the engine already figured out
Our board would not have made the fast-twitch bet either, and the receipts here say we are still paying part of the tab. The engine carries an archetype knob that decays speed-dependent receivers faster than everyone else, and this test says that knob is measuring nothing. It also charges Jonathan Taylor an age factor of 0.729, the single biggest discount on his row, for a risk that is 68 percent his depth chart.
The trap fooling the takesmiths is that both stories predict the same headline. A back turns 28, his points fall, and "he lost a step" and "he lost 90 carries" fit the box score equally well. Only one of them survives being separated. Our committee flag is the tell: it fires on Blake Corum and not on Kyren Williams, even though they are two halves of a split the Rams have said out loud is continuing.
What to do about it
Value the job, then the player. In that order.
| 2026 name | Our board | Call |
|---|---|---|
| Jonathan Taylor | RB13 | Hold, 378 touches in 2025 and the age tax is already paid |
| Colston Loveland | TE3 | Buy, 82 targets as a rookie and the role is still growing |
| Harold Fannin Jr. | TE6 | Buy, 107 targets already, at the position where role is worth most |
| Mason Taylor | TE16 | Buy, 65 targets in 13 games, same shape one tier cheaper |
| Luther Burden III | WR33 | Buy, the speed-threat label costs him nothing we can measure |
| Kyren Williams | RB6 | Pay knowingly, our RB6 on a workload being split |
| Blake Corum | RB22 | Fade, 159 touches is the half of that split you do not want |
| Rhamondre Stevenson | RB65 | Fade, and it is the job, not the 28th birthday |
| Christian McCaffrey | RB38 | Hold, 440 touches at 30, we already have him declining |
| Derrick Henry | RB47 | Fade, we have him RB47 and declining, this study does not rescue him |
The free trade is the one your league-mate is already offering you. He is selling a 28-year-old with 378 touches because of a birthday and buying a 24-year-old with 159 touches because of one. Take that trade every time, then go read the depth chart behind both of them.
One honest caveat: a decomposition tells you where the damage landed, not what caused it. A back who genuinely slows down gets benched, and that benching shows up in my opportunity column, not my efficiency column. What this study can say for certain is that the archetype and age labels the market trades on carry no separable signal once role is held fixed, and that every group I could measure crossed zero except the young one, which went the wrong way for the people selling it.