Depth charts — 2026

The published order beside the measured one. Where they disagree, the disagreement is the thing worth reading — the published chart is the only order the free sites can show, because they are it.

2025 regular-season opportunity (18 weeks)

first non-fallback render after 2026-09-24

why this ordering is not this season

There is no 2026 measured usage. The nflverse stats_player release has no 2026 weekly asset yet (HTTP 404 for https://github.com/nflverse/nflverse-data/releases/download/stats_player/stats_player_week_2026.parquet); the first regular-season file is expected ~2026-09-14. This ordering is 2025 data against the 2026 published depth chart, which is a real comparison and is NOT a claim about this season.

what this order rests on8 statements
  • opportunity = carries + targets, from nflverse stats_player_week. It is NOT a snap count: that file carries 150 columns and none of them is one. A player who ran routes and saw no target and a player who never left the sideline are indistinguishable in this number.
  • The ordering is NOT 2026 data. There is no 2026 measured usage. The nflverse stats_player release has no 2026 weekly asset yet (HTTP 404 for https://github.com/nflverse/nflverse-data/releases/download/stats_player/stats_player_week_2026.parquet); the first regular-season file is expected ~2026-09-14. This ordering is 2025 data against the 2026 published depth chart, which is a real comparison and is NOT a claim about this season. The first non-fallback render is after 2026-09-24, which is seventeen days after the draft.
  • The published depth chart is NOMINAL. This snapshot is preseason and has never been read against a played game; no row here says who was actually on the field.
  • Position appears twice and the columns are not interchangeable. `chartPosition` is where the team lists him and `scoredPosition` is what he is scored as; 2 of the 909 skill rows disagree (Jackson Meeks, chart TE / crosswalk WR; Velus Jones Jr., chart RB / crosswalk WR).
  • A null opportunity means UNOBSERVED, never zero. 365 of 909 players carry one. A player absent from the DynastyProcess crosswalk is not absent from the NFL: 115 of 909 skill rows (12.7%) are absent from it at deep ranks, and no key choice touches that — the miss is a property of the crosswalk, not of the key. This counts rows with NO crosswalk entry, which is not the same population as rows with no gsis id: one player IS in the crosswalk and carries no id, so that second count is one higher. The sentence above is the first of the two.
  • `posRank` is PER-POSITION, not per-slot. A `posRank <= 2` filter on WR excludes the WR3.
  • 137 of the 544 players carrying a 2025 opportunity earned most of it for a team other than the one they are listed under here — QB 20, RB 32, WR 60, TE 25. 32 of the 32 teams carry at least one, so there is no clean page and no "nothing to see here" state to reach for. THIS IS NOT A DATA GAP; IT IS THE READING. A number printed under a team that did not earn it, with nothing on the row saying so, is the defect. `priorTeam` on every player is the correction, and it is never defaulted to the team the row is listed under.
  • 19 players on this chart appeared for more than one team in 2025, so the single number beside the name is a sum of two rooms rather than a count from one. 9 of them have a primary team EQUAL to the team they are listed under, so a marker that fires only when the team changed says nothing at all about them. The largest: Jakobi Meyers is listed under JAX showing 115, of which 66 was earned in that room — the printed number is 74% larger than what he did here. The split is therefore carried for all 19 of them on each team's `multiTeam`, not only for the ones a change-marker catches, and an arm at zero opportunity is carried too (4 of them have one): he was on that sheet for those weeks and saw nothing, which is an observation and not a gap. The rule that picks the primary team has three clauses — most opportunity, then most weeks, then the team code ascending — and on this artifact only the first is ever reached: no player here has two arms level on opportunity, so nothing on this file depends on the other two.
what moved, 2026-08-18 to 2026-08-258 changes

what moved, 2026-08-18 to 2026-08-25

7 days of the published 2026 chart, both endpoints present. Checked again against 2026-08-26.

The filter is a rank cut — QB, RB and TE to 2, WR to 3 — and it is the whole of it: it takes 90.9 changes a week down to 24.0, and counting one event for one event takes that to 16.7 a week (skill positions (pos_abb QB/RB/WR/TE), in-season Tuesday-to-Tuesday, 2025 asset, 16 spans from 2025-09-09 to 2025-12-30, measured 2026-08-28). Depth-1 confirmation removes none of it — it is a label, and if it filtered it would move only 4.7% of the rows. Some reverts survive both: a change that holds for a day and flips on the second is marked confirmed and reverted anyway, and a depth that caught it would suppress real reversals to do it.

7 — a name that was not in a top slot is in one

  • ATL
    • TE2Austin HooperthenCharlie Woerner
    still in place the next day
  • HOU
    • WR2Jayden HigginsthenXavier Hutchinson
    • WR3Xavier HutchinsonthenJaylin Noel
    still in place the next day
  • IND
    • QB2Anthony Richardson Sr.thenRiley Leonard
    still in place the next day
  • LAC
    • TE1Oronde GadsdenthenCharlie Kolar
    still in place the next day
  • LV
    • WR3Malik BensonthenJack Bech
    still in place the next day
  • NYG
    • RB2Tyrone Tracy Jr.thenNajee Harris
    still in place the next day
  • SF
    • RB2Jordan JamesthenKaelon Black
    still in place the next day
1 — the same names, in a different order across the league
  • IND
    • WR2Keenan AllenthenJosh Downs
    • WR3Josh DownsthenKeenan Allen
    still in place the next day
what this feed does not establish
  • WHAT A DEPTH-CHART CHANGE MEANS IS NOT MEASURED HERE. This feed reports that a team moved a name between published slots. Whether that helps or hurts anyone is a judgement nothing in this artifact made, and there is no arrow and no colour on the screen that supplies one.
  • A CONFIRMED CHANGE CAN STILL REVERT, AND THE RATE IS UNESTABLISHED. Measured over the 2025 asset, the share of cut-rank changes whose previous player returns to the same slot within seven snapshots reads 16.1% over the whole file and 21.3% over the in-season window alone, against the roadmap's recorded 17.1% and 13-07-PLAN.md's correction of 13.7%. Four readings, no recorded definition for the first two. The phenomenon is confirmed; the magnitude is not.
  • A SUMMER FEED IS NOT AN IN-SEASON FEED. Measured on the 2026 asset at this cut: 3.7 events a week in April to June and 5.6 in July and August, against 16.7 in season. This artifact carries 8 events over 7 days ending 2026-08-25. WHICH BASELINE APPLIES IS NOT DERIVED ANYWHERE: the depth-chart asset carries no season column, no week column and no date beyond `dt`, so nothing here can say whether a given Tuesday was a regular-season one, and a hardcoded kickoff date would be a guess wearing a field name.
  • `pos_rank` IS PER-POSITION, NOT PER-SLOT. A rank-2 cut on WR excludes the WR3, who is the slot receiver and is startable in a 12-team league. That is why the cut is mixed, and the mixed volume was measured before it was chosen rather than picked as a middle.
  • THE SPAN IS TUESDAY TO TUESDAY AND IT IS A NET DIFF. A slot that changed on Wednesday and changed back on Friday appears here as nothing at all, and a slot that changed after the later Tuesday is not here yet. The two `dt` endpoints on this artifact are the only statement of how old it is; there is no clock anywhere in the code that wrote it.
  • ANALYST CHARTS ARE NOT DIFFED AND MUST NOT BE. FantasyPros-style depth charts are ECR-derived, so diffing them measures an analyst changing his mind rather than a team changing its plan. This reads the team-published chart nflverse republishes.

backfield — unestablished

Every card below carries two measured shares. The classification is not measured, and this screen does not supply one.

No handcuff band is established. Deriving one from the 2025 distribution was attempted and refused: across the 32 teams the RB1 share of the room's 2025 opportunity runs 38.9% to 95.7% with a largest interior gap of 7.6 pp, and 32 points drawn uniformly on that range produce a gap that large 29.3% of the time (100,000 trials). The distribution is not bimodal, so there is no trough to put a threshold in, and a round number would be a guess wearing a measurement's clothes. The shares below are measured; the verdict is not. NOTE the population: they are the 2026 room's 2025 opportunity, so a back who changed teams carries his old volume into his new room and this describes no backfield that existed in 2025.

271 of 544 measured rows across the league sit at a rank the team did not publish.