Week 2 argued with Week 1. Chicago, a week after scoring 59, lost 9–3 at home to Minnesota. The Rams, who scored 7 in Melbourne, beat the Giants 28–6 on Monday night. Carolina, which allowed 59 in Charlotte, won 34–3 in Atlanta. Kansas City's defense, the best in football for Week 1, allowed 30 to the Colts and won anyway, 33–30. Of Week 1's ten best offenses, five were top ten again: Buffalo, Carolina, Dallas, Kansas City and San Francisco. Of its ten best defenses, three: Cincinnati, Las Vegas and Minnesota. Of 2025's playoff teams, Houston and the Chargers are 0–2; Buffalo, Philadelphia, Seattle and San Francisco are 2–0.
Last week's issue held Week 1 up to the ruler this newsletter built in Issue 5 — ten seasons of openers, checked against what came next. One game of offense predicts the rest of the season at 0.334, and last season's offense predicts it at 0.374; Week 1 defense carries at 0.157. For players, usage carries — target share at 0.65, Week 1 carries at 0.72 — while Week 1 receiver points carry at 0.56, and only 28% of a Week 1 top-twelve receiver group stays top twelve.
Then Week 2 happened, and the loud results mostly flipped. The thing that repeated was usage, which is what the ruler said. This issue is the check, including the places last week's verdicts can't be graded. The sample is still two games.
Chicago scored 3. The verdict on the offense survives it, barely.
Last week's verdict was "believe Chicago's offense, don't believe the 59." The second half held. The first half took a hit on Sunday.
Caleb Williams went 15 of 26 for 138 yards, no touchdowns and an interception, and was sacked three times. By expected points per dropback — how many points each snap added to the expected score — he ranked 21st of the 34 quarterbacks with at least fifteen attempts, a week after ranking third of 32. His 29 dropbacks averaged −0.036 expected points each before he left midway through the fourth quarter, and Tyson Bagent threw the last nine passes; NFL.com (September 20 and 21) reported a right hamstring injury and has Williams week to week, not ruled out for Week 3. The charting had two of Chicago's 35 throws interception-worthy, after zero of 29 in Week 1. It rained.
The run game, first in the league by expected points per rush in Week 1, ranked 18th: 31 rushes for 134 yards, D'Andre Swift 16 carries for 45 a week after 18 for 124 and three touchdowns. Ten drives: four punts, two turnovers, one field goal, one missed field goal, one turnover on downs, one end of half. Five of fifteen on third down. Minnesota's defense ranked ninth for the week, one of the three Week 1 top-ten defenses that stayed there.
Chicago's offense ranked first in Week 1 and 24th in Week 2. Over the two games it ranks seventh; across 2025 it ranked eighth. The 59 was an event. So was the 3.
The Rams flipped, and play-action came back with them.
Last week's verdict on "Stafford is done and the Rams were a mirage" was "leaning noise, held honestly," with Week 3 named as the grading date. Monday night moved the evidence a long way in one direction.
Matthew Stafford went 22 of 31 for 327 yards, four touchdowns and one interception, and ranked third of the 34 quarterbacks by expected points per dropback. In Melbourne he was 31st of 32. One of his 31 throws was charted interception-worthy, against three in Week 1. Davante Adams caught 8 of 10 targets for 195 yards and two touchdowns, 35.5 points in Half-PPR, second among receivers for the week.
Last week named one charting change to track: play-action, which fell from 34.6% of dropbacks across 2025 to 17.2% in Week 1. In Week 2 it was 34.4%. That is the cleanest reversal in this issue: it went back to last season's rate, not last week's.
The offense ranked third for the week, 15th over two games, and second across 2025. One game each way, with last season still in the room; Week 3 is still the grading date.
The other sideline can't be graded at all. Jaxson Dart threw five passes — 3 of 5 for 20 yards — before leaving on the opening drive; the Giants say a knee injury has him out for the rest of the regular season (Giants.com, September 23). Jameis Winston went 11 of 27 for 111 yards with an interception, and the Giants' offense ranked 31st for the week, after second. Last week graded "Dart and the Giants are real" as leaning real, on a curve. The quarterback that verdict was about is out, so it comes off the board rather than getting a grade.
Carolina's defense, and what a one-game rank can't tell you.
Carolina's defense ranked last in Week 1 and first in Week 2, after allowing 3 points in Atlanta. Kansas City's ranked first in Week 1 and 26th in Week 2, after allowing 30 to Indianapolis. Week 1 defense carries at 0.157 on the ruler, the weakest signal in Issue 5, so both moves ran the direction it expects.
The trouble with either rank is the opponent. In a week of one game, a defense's rank is its opponent's offense read from the other side: the week's best defense faced the week's worst offense by definition. So look at the opponent in the other week. Atlanta's offense ranked 31st in Week 1 too, against Pittsburgh. Denver's, which Kansas City held to 10 in Week 1, ranked 11th in Week 2 against Jacksonville. Indianapolis's, which scored 30 on Kansas City, ranked 27th in Week 1.
Atlanta's two quarterbacks, Cooper Rush (10 of 17, 86 yards, two interceptions) and Jack Strand (8 of 15, 59 yards, one interception), were the two lowest of the week by expected points per dropback among the 34 with at least fifteen attempts, Rush last. Atlanta turned it over five times — three interceptions and two lost fumbles — and Carolina returned one of the interceptions for a touchdown. Carolina's defense ranked 23rd across 2025.
None of that makes Carolina's Sunday fake. It makes the rank hard to read on its own, as Kansas City's was a week ago. Kansas City's defense ranked 12th across 2025; two weeks in, it has been first once and 26th once. Houston's went from 28th to 12th, and Houston is 0–2 after a 20–6 home loss to Cincinnati with an offense that ranked 27th. Pittsburgh's went from second to 14th.
What Kansas City's Week 2 did repeat is the half last week said to believe. The Chiefs ran 29 times for 152 yards and ranked third by expected points per rush, after second. Kenneth Walker III carried 24 times for 117 yards and caught 6 of 10 targets for 61, 20.8 Half-PPR points, second among running backs; his share of Kansas City's carries was 82.8%, after 60.5% in Week 1 and 43.6% in Seattle across 2025. Patrick Mahomes went 32 of 47 for 382 yards, three touchdowns and no interceptions, ninth of the 34 by expected points per dropback.
The team-level ruler, with the sample size out loud: one week across 32 teams can't prove a ruler; it can only fail to break it. It didn't. Week 1 offense tracked Week 2 offense at 0.17; 2025 offense tracked it at 0.29. Week 1 defense tracked Week 2 defense at −0.13, while 2025 defense tracked it at 0.19. On this one week, 2025 tracked Week 2 more closely than Week 1 did on both sides of the ball — a gap one week can't confirm — and Week 1 defense didn't track it at all.
The points moved. The shares stayed.
Last week: "Coker's 26.5% target share is the number to grade him on next Wednesday." It is Wednesday.
Jalen Coker caught 8 of 9 targets for 66 yards and no touchdowns in Atlanta, 10.6 points in Half-PPR, a week after 8 of 9 for 138 and two touchdowns, 29.8 points and the top receiver score of Week 1. The points went from 29.8 to 10.6. His target share went from 26.5% to 28.1%, against 14.9% across 2025. The box score was a mirage; the role was not. That is the ruler's sentence about receivers, written out in one player.
The same shape shows up across the position. Justin Jefferson caught 3 of 6 for 55 yards, 7.0 points, after 27.2; his share went from 39.1% to 31.6%, against 30.1% across 2025. Isaiah Likely, the top tight end of Week 1 at 23.8 points, caught 5 of 10 for 33 yards and 5.8 points with a different quarterback throwing most of the passes; his share was 27.6%, then 31.2%, against 12.4% in Baltimore last season. Christian Watson, second among receivers in Week 1 at 29.7, caught 4 of 11 for 41 and a touchdown, 12.1 points, on a target share that went from 20.0% to 37.9%. Where the share held or rose and the points fell, the share is the part Issue 5 says to trust.
At running back, Derrick Henry went from 24 carries for 144 yards and three touchdowns to 16 for 68 and one, 34.8 points to 16.2; his carry share went from 64.9% to 72.7%. Walker's went from 60.5% to 82.8%. One Week 1 top-twelve receiver who repeated on points was Jaxon Smith-Njigba, the top receiver of Week 2 with 9 catches on 11 targets for 155 yards and three touchdowns, 38.0 points in Seattle's 31–7 win in Arizona; his target share was 45.8% in Week 1 and 42.3% in Week 2. Amon-Ra St. Brown was fourth at 30.7, on 34.2% after 35.9%.
Of Week 1's top-twelve receivers, three were top twelve again in Week 2: Smith-Njigba, St. Brown and Chris Olave. Two of the twelve, Zay Flowers and Nico Collins, did not play — both hamstrings, per Ravens.com and PFN on September 19. Issue 5's 28% was built on ten seasons; this is one week of a group of twelve.
The whole-population version, with the sample stated: across the 103 receivers and tight ends who drew at least three targets in Week 1 and played in Week 2, Week 1 target share tracked Week 2 target share at 0.37, and Week 1 Half-PPR points tracked Week 2 points at 0.24. Across the 57 running backs with at least three Week 1 carries who played Week 2, carry share tracked at 0.71 and points at 0.55. Issue 5's ten-season figures, which measured the whole rest of the season rather than the next game, were 0.65 for target share, 0.56 for receiver points and 0.72 for Week 1 carries (a count, not a share). One week to the next is a noisier test, so compare the order, not the size: for receivers, share ran ahead of points, as it did over ten seasons; for backs, on this week alone, share ran ahead of points too.
Real or Noise
The verdicts, calibrated, on two games. These grade the story, not the team.
- "Chicago's offense was a one-week fluke": leaning noise. The Week 2 game was real — 24th of 32, Williams 21st of 34 by expected points per dropback. Over two games the offense ranks seventh; it ranked eighth across 2025. The 59 and the 3 are both events.
- "The Rams are back": leaning real for the offense, on one game each way. Stafford third of 34 after 31st of 32, one interception-worthy throw on 31 after three, play-action at 34.4% against 34.6% across 2025. Third for the week, 15th over two games, second last season.
- "Carolina's defense is fixed": can't be graded yet. First for one week, against an Atlanta offense that ranked 31st the week before; last in Week 1; 23rd across 2025.
- "The Chiefs' defense is a problem": noise for now — last week said wait on the defense, and this is why. First, then 26th; 12th across 2025; the Colts' offense that scored 30 ranked 27th in Week 1. The part to believe is still the run game: third by expected points per rush after second, Walker at 82.8% of the carries.
- "The breakout receivers were a mirage": the points, yes; the shares, no. Coker went 29.8 to 10.6 in points and 26.5% to 28.1% in target share. Likely 23.8 to 5.8 and 27.6% to 31.2%. Watson 29.7 to 12.1 and 20.0% to 37.9%. Across 103 receivers and tight ends, share carried at 0.37 and points at 0.24.
The Receipts
Verified numbers worth keeping in your pocket.
- Jalen Coker's target share went from 26.5% to 28.1% while his Half-PPR points went from 29.8 to 10.6. Across the 103 receivers and tight ends with at least three Week 1 targets who played Week 2, target share carried at 0.37 and points at 0.24.
- Kenneth Walker III's share of his team's carries: 43.6% in Seattle across 2025, 60.5% in Week 1, 82.8% in Week 2. Kansas City's run game ranked second and then third by expected points per rush.
- Matthew Stafford went from 31st of 32 to 3rd of 34 by expected points per dropback, and the Rams' play-action rate went from 17.2% of dropbacks to 34.4% — against 34.6% across 2025. One interception-worthy throw on 31, after three in Melbourne.
Where to hear it
Muffed is an AI that reads every play, every box score and every advanced stat from all 32 teams — and tells you the real story of the game. Every number above was checked against the source data before it shipped. That's the house rule, and the receipts are in the footer.
The Week 2 Recaps are up for every team. The four behind this issue: Chicago, Los Angeles, Carolina and Kansas City. The players have their own pages and a weekly episode each — Jalen Coker, Kenneth Walker III, Matthew Stafford, Jaxon Smith-Njigba — and the shares in this issue come off the target share board, now carrying two weeks.
Follow a team, a player, a board, or this issue at muffed.ai and it all arrives in one email on Wednesday. If someone forwarded you this, the Read is free, and it's better when it just shows up.
Next week: three games, and the first numbers that get to be called trends.
Methodology: Week 2 of 2026 is the sixteen games from Thursday, September 17 through Monday, September 21. Team offense and defense are expected points added per play on plays whose type is a pass or a run (scrambles included, penalty no-plays excluded), offense as the possession team and defense as EPA allowed, ranked among all 32 teams; "over two games" is the same measure over Weeks 1 and 2 together. This is the count the site's team boards use; Issue 6's unqualified ranks used plays flagged as passes or rushes, so a few Week 1 and 2025 ranks here differ from last week's by a place or two. Quarterback EPA per dropback is ranked among the 34 quarterbacks with at least fifteen attempts in Week 2 (32 in Week 1, one per team). Run-game value is EPA per rush on the same play-type count, scrambles as runs. Charting figures are FTN charting via nflverse: interception-worthy throws are per pass attempt excluding sacks, and play-action is per dropback. Fantasy points are Half-PPR; "top twelve" is by that week's points. Target and carry shares are the player's share of his team's targets or carries in that game. The stability figures — 0.17, 0.29, −0.13 and 0.19 for teams; 0.37, 0.24, 0.71 and 0.55 for players — are Pearson correlations across the stated teams or players between Week 1 and Week 2 only: one week of evidence, not forecasts. The Issue 5 base rates are 2016–2025 historical averages over the rest of the season that followed each opener, and "2025" everywhere means the 2025 regular season. Injury statuses are cited reporting, dated in the text. Data via nflverse; FTN Data via nflverse.