A change of scenery is a pay cut, not a promotion
What this means for your roster. The new address is not the opportunity. It is usually the receipt for losing the old one. In Washington, our board has Jacory Croskey-Merritt at RB39 and Rachaad White at RB41, and White is the name on every list, so buy the incumbent and let somebody else buy the arrival. Fade Keaton Mitchell at RB72 and Elijah Moore at WR99, the two bit-part movers our own engine marks up hardest. Take Romeo Doubs at WR49 at his Green Bay price and not a pick higher for the New England zip code. Hold Isaiah Likely at TE33, who moved into a real target share rather than toward one. And A.J. Brown at WR37 is where we sit furthest from the room, which is a price call about a 29-year-old, not a shot at the move.
What the consensus says
August has one genre and it never closes. RotoBaller puts Rachaad White in Washington on a prove-it deal in a backfield "devoid of quality running back options." FantasyPros has Romeo Doubs slotting straight in as the New England WR2 after four years of target trouble in Green Bay. RotoWire files Isaiah Likely to the Giants as a featured role waiting to happen. Bleacher Report and NFL.com run the same shape at a dozen more names.
The through-line is always the same and it is always about the destination. Nobody in that stack asks the boring question, which is what happened to the last several hundred guys who got the same phone call.
The claim, in plain English
A player is in the test group if he lined up for a different NFL team the next season than the one he finished this season with. That is it. It is a transaction, settled in March, and it contains none of his own production.
The control is the entire experiment. Movers are not a random draw, because teams let go of players who were
not producing, so an uncontrolled comparison just re-measures who was already good. So I fixed the size of
the role first: what share of the touches taken by his own position group on his own team did he account
for. One means he was the whole running back room. Under about fifteen percent means he was a bit part.
Then I compared movers to stayers inside each band, and only inside each band. The script is
scripts/h83_cameo_mover_change_of_scenery.py.
How I beat on it
Every running back, receiver and tight end season from 2015 through 2024 with at least four games and fifteen touches, paired against the following season, scored in half-PPR with the half-point tight end bonus. That is 2,725 pairs across 858 players, 730 of them movers. Another 505 seasons produced no following year at all and sit out of both sides, because you cannot change teams if you leave the league. Full framework in the autonomous research spec.
What the data actually said
First swing, no controls: movers came back at 5.29 points a game, stayers at 7.65. A 2.36 gap. That number proves nothing, because those movers were already 1.64 points worse the year before they moved. Aggregates lie, and this one lies in exactly the direction you would expect.
So I put them in bands and made them race their own kind.
| 2025-style room share | movers | stayers | Y+1 gap | gap after removing the head start |
|---|---|---|---|---|
| 0.60 and up, room alpha | 50 | 230 | -2.97 | -0.28 |
| 0.40 to 0.60 | 95 | 348 | -3.00 | -1.06 |
| 0.25 to 0.40 | 159 | 507 | -2.15 | -0.86 |
| 0.15 to 0.25 | 203 | 433 | -1.27 | -1.02 |
| 0.07 to 0.15, bit part | 162 | 332 | -1.29 | -1.06 |
| under 0.07, deep bit part | 61 | 145 | -0.56 | +0.08 |
Every band is negative. Not one of them is the breakout the lists promise. And the cleanest cells are the middle two, where movers and stayers were separated by roughly two tenths of a point the year before, which is nothing, and then by a point and a quarter the year after, which is not.
Across all 2,725 pairs, once you strip out the head start, moving cost three quarters of a point a game, with a bootstrap interval of 0.44 to 1.00 that never touches zero. In the bit-part bands specifically it cost 0.75, interval 0.23 to 1.26. The mechanism is not mysterious and it is not effort. It is touches. A bit-part back who stayed handled 65 touches the next year; one who moved handled 36. Receivers, 40 against 26. Tight ends, 25 against 15. The new team did not hand him the ball. It handed him a roster spot.
Which gets you the number that should end the genre. A bit-part player who changed teams reached 10 points a game the next season 2.7 percent of the time. The one who stayed exactly where he was hit 4.2. The change-of-scenery lottery ticket is not a better ticket. It is a worse one.
What the engine already figured out
Here is the part I did not enjoy. We bet the wrong way on this, in public, right now.
SignalTuned re-projects a team-changer's opportunity toward the league average for his position, on the theory that his old usage no longer describes him. For a player who was carrying a big role, that pulls him down, which is right. For a bit part, it pulls him up. On today's board, of the 86 non-quarterback movers, 51 get marked up. Split by how big their role was, it is worse than that: of the 47 movers who were bit parts last season, 91 percent get an upgrade, at a median of 1.25 times their production base. Of the seven movers who ran their room, none do. The rank correlation between how small your role was and how hard we push you up for leaving is 0.87.
We already fixed this once, at quarterback, where a cameo starter who changed teams was collecting the promotion and the move as two separate raises. That gate never got written for the other three positions. The cohort says it needs to be.
The 2026 names
Board ranks from the 12-team superflex half-PPR tight-end-premium export, engine v1.6.144. Room share is his 2025 share of his own position group's touches.
Rachaad White, RB41, moved Tampa Bay to Washington, 0.36 room share. Fade at his current price. He is the most-named back on the change-of-scenery lists and he lands in the band where movers gave back 0.86 a game against their own kind.
Jacory Croskey-Merritt, RB39, stayed in Washington, 0.47 room share. Buy, and buy him from the manager who just read that White is coming. He is the incumbent in the exact backfield the story is about and our board already has him ahead.
Tyler Allgeier, RB40, Atlanta to Arizona, 0.29 room share. Hold, do not chase. Same band as White, same discount, and our engine currently docks him only 6 percent for the move.
Keaton Mitchell, RB72, Baltimore to the Chargers, 0.15 room share. Fade. He is a bit-part mover our own board marks up 11 percent, and this study says that sign is backwards.
Elijah Moore, WR99, Buffalo to Philadelphia, 0.08 room share. Fade hardest. He takes the single largest mover upgrade on our board at 1.28 times base, out of the deepest bit-part band in the sample.
Romeo Doubs, WR49, Green Bay to New England, 0.29 room share. Hold at his old price. A real role that moved sideways, not a promotion, and the market is charging for the promotion.
Darnell Mooney, WR85, Atlanta to the Giants, 0.28 room share. Hold. Same band, same math, and the case for him is Malik Nabers missing time rather than anything about the move itself.
A.J. Brown, WR37, Philadelphia to New England, 0.42 room share. Hold, and know that this is our loudest disagreement with the room. Our board is low on him because he turns 29 with a 0.75 mover mark applied on top, not because we think the trade ruins him.
Isaiah Likely, TE33, Baltimore to the Giants, 0.28 room share. Hold. The one mover here I would defend paying for, because he arrives at a target share instead of at a depth chart, and our board takes only 4 percent off him.
Mike Evans, WR137, Tampa Bay to San Francisco, 0.16 room share. Fade, but for his age and his eight games, not for the flight. At 33 the move is the least interesting thing about him.
What to do about it
Ask which side of the transaction the player is on. A move is information the team is giving you about him, and the information is almost never good. The bet you actually want is the other guy: the incumbent standing in the room the new arrival just walked into, whose price drops the week the signing hits the wire, and who historically keeps more of the football than the guy holding the press-conference jersey.
On our side, this goes back to the engine as the non-quarterback version of a gate we already ship at quarterback. It is not landing as a quiet tweak today, because the change the data implies is a 6 to 21 percent move on affected players, which is far outside what one of these studies is allowed to push on its own, and because the ranking guard that grades this layer currently fails on the configuration we already ship, so nothing can be graded until that is repaired. Both go to Ryan with the receipts attached.
One honest caveat. A player traded partway through a season shows up in this data on the team he finished with, so a handful of real movers are sitting in my stayer group, which if anything makes the mover penalty here slightly smaller than the true one.
Receipts
Cohort output data/h_cameo-mover-change-of-scenery_1786727689.json. Test script
scripts/h83_cameo_mover_change_of_scenery.py. Board ranks from boards/rankings_dynasty_sf_half_ppr_tep.csv; mover factors computed from
metrics_xfp.compute_volume_factors against reference season 2025. Engine v1.6.144.