Sources and caveats

2025 is weaker than the rest. The upstream archive stopped collecting after 2025-08-08, so the 2025 board is roughly three and a half weeks stale relative to draft day — late-August injuries and depth-chart moves are not in it. Treat 2025 drift as noisier than 2021–2024, which are pinned to boards from the week of each draft.

888 of 889 auction picks matched a player in the outside data, and 885 of those had a rank on the board the day the league drafted. The gap is mostly team defenses and end-of-draft dollar fliers who were never on anyone's board.

FantasyPros PPR redraft expert consensus rankings, archived by the DynastyProcess data project (github.com/dynastyprocess/data). A player's board rank is his place among his own position on the overall cheatsheet — one ordered list covering every position, so a running back's rank and a tight end's rank describe the same afternoon.

The last archived board before this league actually drafted, not before Week 1. The league drafts Labor Day weekend, and the board moves in the final fortnight, so draft day is the honest cutoff — it is what the room was looking at while bidding.

FantasyPros renders its ADP tables client-side and overwrites them each season, so there is no free machine-readable historical ADP archive. Expert consensus tracks ADP closely in the top 24 and diverges on hype players, who go earlier in real drafts than the experts rank them. Read these as consensus draft value, not an average pick number.

Every public positional-finish list is full PPR. This league is not, and has not been one thing twice: receptions were worth nothing from 2021 through 2024 and half a point in 2025, passing touchdowns have always been worth six, and the kicker and points-allowed tables were re-cut in 2024. Each season is re-scored from nflverse weekly stats using that season's own Sleeper settings, game by game so the per-game yardage bonuses and points-allowed tiers land where they should.

The board is full-PPR consensus; the finishes are this league's scoring. That is deliberate — drift measures market perception against this league's reality — but it is not an apples-to-apples subtraction, and it matters most at receiver, where full PPR flatters pass-catchers relative to how this league counted them until 2025.

nflverse weekly player stats (github.com/nflverse/nflverse-data), regular season only, no games-played minimum. Positional finish is measured against every player at that position who recorded a stat line that year — the pool size is shown on each card. Team defense has no fantasy rows upstream, so it is assembled from team defensive stats, the opponent's blocked kicks, and the final score.

Derived from nflverse game-level data (nfldata/games.csv), which records the coach for every regular-season game. Where a team changed coaches mid-season, both are listed with game counts. A few names carry the source's own spelling quirks.

Not available in any machine-readable historical dataset. This column was compiled by hand and is the least reliable field here — an amber dot marks an entry that is likely but not certain, a red dot marks a guess. Mid-season changes are noted in the cell. Correct anything that looks wrong; the underlying map is a 160-row CSV in data/raw/.

Averages lean negative by construction: the drafted group is a closed set, while undrafted breakouts enter the finish list and push everyone down. Compare players against each other, not against zero.