2026
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Outlined = current league week
Sign in Sign up
Methodology

How we count

Every page shows the box score first. Beside it, a few numbers of our own carry the swan . Each one is below: what it is, what it is built from, and how much evidence it needs before it appears. A number that does not clear its floor is not shown at all.

2026 fit
usage fit v1 2026, trained on 2023–25 · a receiver's game varies by about 22 receiving yards around it
2025 fit
usage fit v1 2025, trained on 2022–24 · a receiver's game varies by about 22 receiving yards around it

Usage Yards

What a player's usage was worth in one game he played. A straight-line fit, per position group, of each stat on the work he got: targets, air yards and red-zone targets for receiving yards and catches; carries, red-zone carries and carries inside the 5 for rushing yards; all six for scrimmage yards; and for a quarterback's passing yards, his dropbacks, the air yards he threw and his attempts inside the 20.

Built from
Each play of the game, counted per player, and his box score. The fit is trained on the three seasons before the one it scores.
Floor
A game already played, in a stat the fit covers. A quarterback's carries the completion rate over the average throw (CPOE) beside it.
Shown
Always beside the yards he gained, never alone and never for a game not yet played.

Role

The change in a player's job, in yards: his Usage Yards a game over his last four games against the season's other games (last season's, while fewer than four of this season's remain). Positive: more work, of the kind that has been worth yards. Of our numbers it is the one that carries from one game to the next (below).

Built from
Usage Yards on his position's main stat: receiving yards for a receiver or tight end, scrimmage yards for a back.
Floor
4 games in the window and 3 outside it. The bar fills only when the change is at least 2 standard errors from zero.
Shown
First on a page when it clears that bar: the home page's list of bigger and smaller roles, a player's, a team's and a game's lead.

Beyond Usage

The yards he gained minus his Usage Yards, averaged over a stated span: his last four games (L4), a season, or the games in a split. Positive: he did more with the work than it was usually worth. It is the gap between the box score and the job, so read it as a discount on the box score, never as a ranking of players: about a third of a gap has carried into the next game, and a season's gap has carried into the next season at r = 0.27 (what holds).

Built from
Usage Yards and the box score, on his position's main stat: receiving yards for a receiver or tight end, scrimmage yards for a back.
Floor
Printed from 3 games. The bar under it fills only when the average is at least 2 standard errors from zero; tap the number to see the error.

Usage TDs

What a player's touches were worth in touchdowns: each carry at the league's touchdown rate from where it started (1, 2, 3–5, 6–10, 11–20 or 21+ yards out), each target at the rate for that spot and how deep it was thrown (behind the line, 0–9, 10–19, 20+), summed over his season. The gap is his touchdowns minus it.

Built from
Every play of the three seasons before the one scored; a cell with fewer than 200 plays uses its yard line's rate.
Floor
8 games in the season. The bar fills when the gap is at least 2 of the sum's own spreads. A season figure only: one touchdown is one event, so a single game never gets one.
Holds
About a tenth of a season's gap has carried into the next season: the touchdowns beyond it are the part of a line least likely to repeat.

Swans

The outliers the site is named after: the results nobody's recent role allowed for. Three kinds.

Weekly
One game at least 2 spreads from what his last four games' usage was worth going in (the same fit, on the usage he had been getting), either way, on the position's main stat. Printed split in two: role, this game's Usage Yards less that reference (a bigger or smaller job than usual), and beyond, his yards less this game's Usage Yards (what he did with it). Final stat lines only.
Season
A regular season's Beyond Usage over 3 games or more, at least 2 standard errors from zero.
Playoff
The weekly bar applied to a postseason's games.
Built from
Final stat lines and Usage Yards on the position's main stat: receiving yards for a receiver or tight end, scrimmage yards for a back, passing yards for a quarterback.

A Swan marks a game not to read as typical; the next game has kept about a third of the role part and about a tenth of the beyond part. The reference is read only after the game, for that game.

Funnel

Where a defense gives up the most: of backs' rushing yards, backs' receiving yards, receivers' yards and tight ends' yards a game, the one it ranks worst at in the league.

Built from
The defense's profile after each week: every final game it played, season to date and the last four.
Floor
Top 5 in the league (1 = allows the most) in either window, over at least 3 games.

Situations

A player's yards beyond what his recent usage was worth, in one situation his next game is in. Seven kinds, each with a football reason: without a teammate who is out, with this quarterback, under his own play-caller, against the defensive play-caller or head coach he faces, against a man-heavy or two-high defense, against a blitz-heavy one, and against light boxes (each style: the top quarter of the league that season).

Built from
His games over the last three seasons, the coaching staffs, the injury report and each play's charting of coverage, rushers and box.
Floor
The split's own minimum games (3 to 5), at least 2.5 standard errors from zero, and at least 3 of his games in the window outside it. Measured Tuesday, for the week's game only; a game already played has none.

Second week

A change in a player's role that happened again: the same kind of move, on the same player, two weeks running.

Built from
What changed, week over week.
Floor
Two weeks in a row.

First on it

The report that broke a story, when a club site or a city outlet had it before any national one, with how long the national coverage took. Later reports of the same story fold under it.

Built from
Headlines grouped by the player and the kind of news within 48 hours.
Floor
A national outlet reported it later.

What holds

A number is worth reading only if it carries from one stretch of games to the next. We check each one three ways, over every running back, receiver and tight end final line we have fitted, on the position's main stat, and print the result here as it stands.

StatOdd weeks vs evenSeason to seasonLast four vs next game
Usage Yards 0.921146 player-seasons 0.83721 pairs 0.6012971 player-games
Role — — 0.269016 player-games
Yards a game 0.861146 player-seasons 0.80721 pairs 0.6012971 player-games
Beyond Usage 0.241146 player-seasons 0.27721 pairs 0.1212971 player-games
QB Usage Yards (passing) 0.54121 player-seasons 0.4575 pairs 0.211491 player-games
QB passing yards a game 0.63121 player-seasons 0.5175 pairs 0.271491 player-games
QB Beyond Usage 0.53121 player-seasons 0.5475 pairs 0.161491 player-games
TD gap (TDs less Usage TDs) 0.071023 player-seasons 0.08648 pairs —
Swans — — 0.401027 Swans
Funnel (yards allowed) 0.19512 defense-seasons 0.18384 defense-seasons 0.097064 defense-games

Season to season, a player's Usage Yards a game have carried over at r = 0.83 and his Beyond Usage at 0.27. Over his last four games, Usage Yards have tracked the next game (r = 0.60) as well as his yards have (r = 0.60); about a third of a gap has carried into the next game, and about half of a change in role has shown up in the next game. That is why Role leads a page and Beyond Usage sits beside the box score. A Swan marks a game not to read as typical: the next game has kept about a third of the role part and about a tenth of the beyond part.

Efficiency stats

The standard advanced stats on player and game pages, checked the same way. Shares of the work and rates per route carry over; per-play results such as EPA per target, yards per carry and drop rate mostly do not, so read those over a season, not a month.

StatOdd weeks vs evenSeason to season
Yds/route WR/TE 0.63 630 0.59 518
EPA/tgt WR/TE 0.12 957 0.15 550
Success WR/TE 0.17 957 0.15 550
YAC/rec WR/TE 0.18 951 0.34 550
Tgt/route WR/TE 0.76 630 0.64 518
aDOT WR/TE 0.53 957 0.59 550
Air share WR/TE 0.82 959 0.73 550
20+ catch WR/TE 0.19 951 0.18 550
Separation WR/TE 0.25 631 0.43 451
YAC over avg WR/TE 0.20 629 0.32 451
Catchable WR/TE 0.14 957 0.15 550
Drop rate WR/TE 0.05 955 0.09 550
Contested WR/TE 0.10 957 0.27 550
Yds/carry RB 0.23 323 0.20 181
Yds over avg/carry RB 0.13 251 0.20 169
EPA/carry RB 0.13 323 0.11 181
Success RB 0.11 323 0.19 181
12+ run RB 0.17 323 0.28 181
Carry share RB 0.91 323 0.58 181
Inside-5 share RB 0.54 309 0.39 179
Light box RB 0.31 303 0.22 162
Stacked box RB 0.34 303 0.19 162
EPA/dropback QB 0.43 209 0.46 114
CPOE QB 0.33 209 0.37 114
Pressured QB 0.53 189 0.31 96
Time to throw QB 0.63 189 0.58 96
aDOT QB 0.40 209 0.37 114
Cmp% QB 0.34 209 0.42 114
Sack rate QB 0.28 209 0.44 114
Scramble rate QB 0.64 209 0.74 114
Team PROE QB 0.56 209 0.49 114
Neutral pass QB 0.33 209 0.33 114

r is the correlation between the two stretches, from 0 (none) to 1 (the same number). Seasons 2022–26, measured Wed Sep 30 8:05 AM EDT; rerun each Tuesday.

Data and licences

Stats, play-by-play, participation, schedules, rosters, injuries and depth charts: nflverse, CC BY 4.0. Next Gen Stats (rushing yards over the average back on the same carries, separation, YAC over the average catch): the NFL's tracking numbers as nflverse publishes them, CC BY 4.0. Charting: FTN Data via nflverse, CC BY-SA 4.0, shown and never exported. Headlines link to their publishers. Not affiliated with the NFL or any team.

Data and context only, never advice. By using Black Swan Stats you agree to the Terms and Privacy Policy. Not affiliated with the NFL or any team.

Read them