The 'OC usage fingerprint' making the rounds is mostly a roster fingerprint. I tested it.
Ignore coordinator usage cards for RB and WR. A coach's lead-back rate and target spread do not travel with him. When you see "this OC rides one back" or "this OC splits targets," you are looking at a description of his last roster, not a prediction about his next one. The one place the coach genuinely shows up is tight end usage, and even there the team's own recent TE share already carries the signal.
What the consensus says
Coordinator "usage fingerprint" cards are making the rounds in the dynasty community: one-pagers stamping an OC with a lead-back carry share, a target HHI, an alpha-target rate, presented as a stable coaching identity you can project onto whoever joins his depth chart. The pitch is intuitive. Play callers have systems; systems dictate touches; buy the players whose skill set matches the system. It is the same instinct that makes "Shanahan RBs" and "Reid TEs" feel like investment categories.
The intuition deserves a real test, because if it holds, it is draft-day gold: you could fade a committee back landing with a bellcow coach and buy the WR2 whose new OC funnels targets. If it does not hold, those cards are describing rosters and billing it as coaching.
The claim, in plain English
If usage concentration is coach DNA, it has to satisfy one falsifiable condition: it follows the coach. His profile at team A should predict his profile at team B better than chance, and retaining him should make a team's concentration stickier year over year than replacing him, after you account for whether the top player stayed.
How I beat on it
Two studies, both pre-registered before any result was computed. First, team-level persistence: concentration fingerprints (lead-back carry and touch share, backfield HHI, WR1 and WR top-2 target share, team target HHI, TE1 target share) for all 320 team-seasons 2015-2024, joined to a hand-curated coordinator database, then two tests: same-OC vs changed-OC year-over-year autocorrelation with roster controls and bootstrap CIs, plus a movers test across the 63 coordinator team changes in the window. Second, player-level value: 2,505 player-season transitions, walk-forward 2018-2024, asking whether adding the incoming coordinator's career usage profile to player priors and team priors improves next-season target-share projections at all.
What the data actually said
The movers test is the clean kill. A coordinator's lead-back carry share correlates at r=0.16 with his own profile at his previous stop. His WR top-2 target concentration comes back at r=-0.21, negative, and team-wide target HHI at -0.19. He does not bring the usage tree with him. He inherits the one the roster hands him. Within teams the same story: retaining the OC adds nothing measurable to RB or WR concentration persistence once you control for whether the incumbent top player returned. Every RB and WR confidence interval straddles zero.
The player-level study closes the loop. Adding coordinator career profiles to a baseline of player priors plus team priors moved projection error by roughly nothing at every position, and on the exact subgroup where the feature was designed to earn its keep, teams changing coordinators, it made RB projections significantly worse. Kevin Stefanski's first year calling plays in Cleveland in 2020 is the type specimen: his prior profile inflated the projected RB target share to 0.107 against an actual of 0.036. The coach's history imports noise from rosters he no longer has.
And the roster explanation fits the extremes. The most concentrated backfield in the sample is 2019 Carolina at an 0.880 lead-back carry share. That was not Norv Turner's system. That was Christian McCaffrey.
The exception: tight ends
One dim survived everything I threw at it. Teams that keep their coordinator repeat their TE1 target share at r=0.56; teams that change him drop to 0.22, a gap whose CI excludes zero by a wide margin, robust to roster controls and to deleting any single team from the sample. Tight end usage is play design in a way RB and WR pecking orders are not. But before you draft on it: the player-level test shows the market-facing version is already priced in, because a team's own recent TE usage carries the coach signal whenever the coach stays, and when he moves, his TE history is too roster-contaminated to beat the new team's own baseline.
What this means for your roster
Read rosters, not coordinator cards. Incumbent talent and the team's own recent usage predict the touch tree; the coach's resume does not, and at RB it actively misleads on coaching changes. Full receipts for both studies, including every confidence interval and the named cases, are in the study docs: team-level persistence and player-level walk-forward.