How much should one game change your mind? San Francisco's four turnovers, Kansas City's 3 points, Tampa Bay's 43, Houston's overtime loss and Philadelphia's 2-5, run through a ruler built for Week 7.

Week 7 ran from Thursday, October 22, when Chicago beat New England 30-24, to Monday night, when Dallas won 38-21 in Philadelphia. Buffalo, Jacksonville, the Chargers and Washington had byes, so it was 14 games. Seven were decided by eight points or fewer, and the season's share of one-score games is now 64 of 107, 60%, still the highest through seven weeks in any season since 2016.

The week's headlines came from the other seven. San Francisco, 6-0 coming in, lost 29-20 in Atlanta. Kansas City, 5-0, scored 3 points in Seattle. Tampa Bay, 1-5 after six straight one-score games, won 43-13 in Carolina. Minnesota beat Indianapolis 16-10 and is the league's only unbeaten team.

Each of those is one game, and the Monday question is how much one game should count. So this issue builds a ruler for exactly that, from the same ten seasons as the last four issues, 2016-2025: every team's game in a given week, set against the games before it and against how the team played afterwards.

The measure is net expected points added (EPA) per play: how many points each offensive snap added to the expected score, minus how many each defensive snap gave up. For every team and every week from Week 4 to Week 12, the latest game's net EPA was weighed against each earlier game's for how well it predicted the rest of the season.

Here is the ruler.

• The latest game's weight, against one earlier game: 1.2. Recency counts, a little. After seven games, the newest one is about a sixth of the evidence.
• The games before Week 7, as a guide to the rest of the season's win rate: 0.39. This and the next two numbers are correlations; 1.00 would be a perfect guide, zero no guide at all.
• The Week 7 game alone: 0.29.
• All of them together: 0.43.

A single game is a real clue about a team. It is also a small one. Adding Week 7 to what came before moved the read from 0.39 to 0.43, not from nothing to something.

Results say the same. From 2016 to 2025, 44 teams ranked in the top five by net EPA played in Week 7. The 16 that lost won 66% of their remaining games; the 28 that won, 63%. The loss still counted in the standings: 12 of the 16 reached the playoffs, against 24 of the 28. At the other end, 12 teams ranked 17th or worse won their Week 7 game by 21 points or more. They won 57% of the rest, against 42% for all 145 teams in that bracket. A rout by a middling team has said more than a loss by a good one.

San Francisco's loss was four turnovers. Turnovers are the part of a game that does not come back.

San Francisco lost in Atlanta because it gave the ball away four times and took it away none, and three of the four came in Atlanta territory. Up 14-9 in the second quarter, on second-and-1 at the Atlanta 28, Kaelon Black fumbled and the Falcons recovered. Atlanta kicked a field goal before the half, then went ahead 19-14 in the third quarter on Michael Penix Jr.'s 5-yard throw to Kyle Pitts.

The next two hurt more. On second-and-14 from the Atlanta 25, Brock Purdy threw short right and Jessie Bates intercepted at the 10. Penix hit Drake London for a 21-yard touchdown on the first play of the fourth quarter. Down 26-20 with 2:42 left, on third-and-9 from his own 48, Purdy was intercepted again and the return reached the San Francisco 19; the field goal made it 29-20. George Kittle's fumble at the Atlanta 20 with 57 seconds left ended it.

Purdy went 27 of 40 for 288 yards, a touchdown and two interceptions; Penix went 25 of 34 for 213 and three touchdowns. Turnovers and all, San Francisco's net EPA for the day was -0.04 per play, 18th of the 28 teams that played. A bad day, not a collapse. Over seven games it is still first in net EPA at +0.22, second on offense behind Buffalo and ninth on defense.

The ruler is clearest on exactly this kind of loss. A team's turnover margin through seven weeks carries to the rest of its season at 0.17, its giveaways at 0.19. Its net EPA carries at 0.52. Across the 256 games from 2016 to 2025 in which a team turned the ball over four or more times, it averaged 1.31 turnovers in its next game, which is the league average exactly. San Francisco had given the ball away six times in its first six games.

The standings take the loss; the forecast barely notices. Teams that were 6-1 after Week 7 reached the playoffs 12 times in 13. San Francisco has a bye in Week 8.

Kansas City scored 3 points in Seattle. The last top-five offense held that low went 9-1 afterwards.

Kansas City lost 20-3 on Sunday night, its first loss and its fewest points of the season. It had eight possessions and scored on one, Harrison Butker's 50-yard field goal; he missed from 58 and 45. Patrick Mahomes went 19 of 33 for 165 yards with no touchdowns and an interception, and the interception, at the Seattle 34, came with Kansas City down 3-0 in the second quarter. The offense ranked 24th of 28 for the week at -0.11 EPA per play.

Seattle's defense is most of the explanation. It ranks second in EPA allowed per play over seven weeks and fifth for this one. Sam Darnold went 20 of 28 for 257 yards, Jaxon Smith-Njigba caught 9 passes for 150 and Emanuel Wilson ran 24 times for 100 yards, including a 21-yard touchdown with 3:52 left.

In ten seasons, one offense ranked in the top five was held to 3 points or fewer between Week 4 and Week 14. It was Kansas City, also in Week 7, a 27-3 loss in Tennessee in 2021. That offense had ranked fourth; it went 9-1 the rest of the way with the league's second-ranked offense. One team is an anecdote, not a rate. The rate is that offense carries from seven games to the rest of a season at 0.58, the most dependable thing a team does, and defense at 0.34.

Kansas City still ranks fifth on offense at +0.12, sixth on defense and third in net EPA. The 3 points are a sixth of the evidence. It goes to Denver next.

Tampa Bay's 43 counts for more than Houston's overtime loss. Both are 2-5, and only one ranks in the top half.

Issue 11 called Tampa Bay the 1-5 team that had lost the coin flips. On Sunday it did not need one. Tampa Bay ran 47 times for 261 yards in Carolina, Bucky Irving 24 of them for 106 and two touchdowns, and Baker Mayfield needed only 11 passes, completing 7 for 64 yards and a touchdown. The defense sacked Bryce Young five times and intercepted him twice. With Tampa Bay up 10-0 in the second quarter, on third-and-14, a sack knocked the ball loose and Calijah Kancey returned it 32 yards for a touchdown.

Tampa Bay's net EPA was +0.48 per play, first of 28 for the week, with the week's best defense at -0.27 allowed. Its season rank rose from 21st to 14th: third on defense, 28th on offense.

The record still trails the team. Tampa Bay has outscored opponents by 11 and its points say 3.8 wins; it has 2, the furthest behind its points of any team in the league at 1.8 wins. Six teams in ten seasons had been further behind after Week 7. Of the 12 teams at least 1.5 wins behind, 4 reached the playoffs. And of the 32 teams that were 2-5 after Week 7, 4 made it, but 2 of the 4 that ranked in the top half by net EPA did: the 2018 Colts, who went 8-1 the rest of the way, and the 2022 Jaguars, 7-3.

Houston went the other way. Issue 10 and Issue 11 called Houston's play "leaning real"; on Sunday the Texans lost 27-21 in overtime to the Giants. The run defense gave up 188 yards and four rushing touchdowns on 34 runs, +0.30 EPA per play allowed after -0.09 coming in. On defense a big positive number is bad. Cam Skattebo ran 17 times for 124 yards.

C.J. Stroud forced overtime. Down 21-14 with three minutes left, Houston drove from the New York 37 to the 3, and on third-and-3 with 57 seconds left Stroud threw a 3-yard touchdown to Xavier Hutchinson. In overtime, Devin Singletary ran 30 yards for the winner. Stroud went 17 of 28 for 153 yards and two touchdowns; J.J. McCarthy went 20 of 28 for 252 for New York, now 4-2-1.

Houston is 2-5, outscored by 5, and 20th in net EPA after ranking 14th a week ago. Of the 28 teams that were 2-5 and ranked 17th or worse after Week 7, 2 reached the playoffs, and as a group they won 36% of their remaining games. That is a different place from last season's Texans, who were 2-4 with the ninth-ranked net EPA. Houston has a bye in Week 8.

Cleveland, Denver and Minnesota keep winning ahead of their points. The wins are banked; the teams are not.

Of the five teams Issue 11 found a full win ahead of their points that played this week, three won: Cleveland, Denver and Minnesota. San Francisco and Kansas City lost.

Cleveland won 27-24 in Tennessee, its fifth one-score win in six tries. Tennessee tied it at 24 with 1:01 left: Cam Ward converted fourth-and-8 with a 26-yard throw to Elic Ayomanor, and Tony Pollard scored from the 1 for his third touchdown. Cleveland took the kickoff to its own 31 with 53 seconds left. Two short completions later, on first-and-10 from the Cleveland 45 with 28 seconds left, Deshaun Watson found Harold Fannin Jr. for 36 yards to the Tennessee 19. Tennessee jumped offside on the field goal try, and Andre Szmyt made it from 32 on the final play.

Watson went 25 of 36 for 308 yards with an interception and ran 9 yards for a touchdown; Quinshon Judkins ran for 96 and two scores, and KC Concepcion caught 8 for 129. Cleveland's offense had its best week of the season at +0.15 per play, 12th of 28.

Cleveland is 5-2 and has been outscored by 13. Its points say 3.1 wins; it has 5, 1.9 ahead. Only four teams in ten seasons were further ahead after Week 7: the 2019 Saints, the 2021 Packers, the 2020 Browns and the 2022 Giants. All four made the playoffs, and they went 7-2, 7-3, 6-3 and 3-6 afterwards. Teams at 5-2 reached the playoffs 38 times in 45, and 3 of the 4 that had been outscored made it. Cleveland ranks 21st in net EPA and 26th on offense. Issue 11's read stands: believe the standings, not the team.

Denver won 30-16 in Arizona, its first win by more than eight points. With the score 6-6 in the second quarter, on second-and-7, Bo Nix ran 62 yards for a touchdown; in the third he threw 52 yards to RJ Harvey for another. Nix went 15 of 29 for 222 yards and ran 7 times for 73. Denver's net EPA was fifth of 28 for the week, and it has climbed from 20th to 16th. It is 5-2, plus 3 in point differential, 1.4 wins ahead of its points.

Minnesota is 6-0 and 4-0 in one-score games, with the league's best defense and its 29th-ranked offense. Indianapolis led 10-6 to start the fourth quarter. On third-and-8 at the Indianapolis 40, Kyler Murray scrambled 35 yards to the 5, and on the next play threw the go-ahead touchdown to T.J. Hockenson. Will Reichard's 52-yard field goal made it 16-10 with 1:44 left, and Harrison Smith intercepted Daniel Jones 29 seconds later. Murray went 14 of 24 for 150 yards and was sacked three times.

All seven teams that were unbeaten after Week 7 from 2016 to 2025 reached the playoffs. None of them had an offense ranked 20th or worse. The nearest match is the 2016 Vikings: 5-1 after Week 7 with the first-ranked defense, the 25th-ranked offense and the second-ranked net EPA. They went 3-7 the rest of the way and missed the playoffs, and they are the only team in ten seasons with five wins, one loss or fewer and an offense ranked 24th or worse at this point. Minnesota's record is a real head start. Its offense is the half of the team that usually holds, and that is the worry. It visits Detroit next.

Philadelphia is last in net EPA at 2-5. No team in that spot after Week 7 has made the playoffs since 2016.

Dallas won 38-21 on Monday night, and Philadelphia had the lead in the third quarter. Tanner McKee's 4-yard touchdown to Makai Lemon put the Eagles ahead 21-20 with 5:49 left in the quarter. Dallas answered, and Dak Prescott threw 16 yards to CeeDee Lamb for the lead with 1:36 left. Philadelphia did not score again. DeMarvion Overshown intercepted McKee at the Dallas 30 midway through the fourth, and Javonte Williams finished it with a 2-yard run.

Prescott went 15 of 27 for 212 yards and three touchdowns, two to Lamb and one to George Pickens, without an interception. Williams ran 25 times for 155 yards. Dallas's offense was second of 28 for the week at +0.32 per play.

Philadelphia has lost five straight and sits 32nd in net EPA at -0.23 per play: 30th on offense and 28th on defense. It ranked 30th in net EPA after five weeks as well, so this is not a two-week slide. From 2016 to 2025, ten teams were last in net EPA after Week 7. None reached the playoffs, and together they won 31% of their remaining games. Widen it to every team ranked 30th to 32nd with two wins or fewer, and it is 0 of 27.

Dallas is 3-4 with the sixth-ranked offense and the 32nd-ranked defense, +0.14 EPA allowed per play. Ten defenses ranked last after Week 7 in the last ten seasons; 2 of their teams made the playoffs, the 2016 Lions and the 2017 Patriots, and the ten won 50% of their remaining games. Offense is the part that carries, and Dallas has that part.

Real or Noise

The verdicts, calibrated on seven games. These grade the story, not the team.

• "San Francisco has been found out": noise. Four giveaways and no takeaways, and turnovers carry at 0.19. Its net EPA for the day was -0.04, and it is still first over seven weeks. Teams at 6-1 made the playoffs 12 times in 13.
• "Kansas City's offense is in trouble": noise. One game against the second-ranked defense. Offense carries at 0.58, and the last top-five offense held to 3 points went 9-1 afterwards.
• "Tampa Bay has turned it around": leaning real. Teams ranked 17th or worse that won by 21 or more in Week 7 won 57% of their remaining games, and Tampa Bay is 1.8 wins behind its points. At 2-5 the playoffs are still a long way off.
• "Houston is back in it": noise. Issue 11 said leaning real. Houston lost to the Giants, fell to 20th in net EPA, and 2-5 teams ranked 17th or worse made the playoffs 2 times in 28.
• "Minnesota is the best team in football": the record, real; the team, noise. All seven teams unbeaten after Week 7 made the playoffs. Minnesota ranks sixth in net EPA with the 29th-ranked offense, and the closest match, the 2016 Vikings, missed.

The Receipts

Verified numbers worth keeping in your pocket.

1. A team's latest game has counted about 1.2 times as much as any one earlier game. After seven games, that makes it roughly a sixth of the evidence. Weeks 4-12, 2016-2025.
2. Teams that turned the ball over four or more times in a game averaged 1.31 turnovers in the next one, the league average exactly. 256 games, 2016-2025.
3. No team last in net EPA after Week 7 has made the playoffs in ten seasons: 0 of 10. Philadelphia is last.

Read more

• San Francisco's loss in Atlanta, Brock Purdy, George Kittle and the Falcons' side
• Kansas City in Seattle, Patrick Mahomes, Seattle's defense and Jaxon Smith-Njigba
• Tampa Bay's 43-13 win, Bucky Irving and Baker Mayfield
• Houston's overtime loss, C.J. Stroud, the Giants' win and Cam Skattebo
• Cleveland in Tennessee, Deshaun Watson and Harold Fannin Jr.
• Denver in Arizona and Bo Nix; Minnesota against Indianapolis and Kyler Murray
• Dallas on Monday night, Dak Prescott, CeeDee Lamb, Javonte Williams and Philadelphia's side
• The boards behind this issue: team EPA and quarterback EPA
• Last week's issue: Which records are running ahead of their points?

Follow the players you care about on Muffed, across all your leagues, and every Wednesday their weeks and the Read land in one email.

Next week: Kansas City visits Denver, Minnesota goes to Detroit, Cleveland visits Pittsburgh, Atlanta visits Tampa Bay, Philadelphia visits Washington, Arizona visits Dallas and Buffalo hosts Baltimore; San Francisco, Houston, the Giants and New Orleans have byes.

Methodology: Week 7 of 2026 is the 14 games from Thursday, October 22 through Monday, October 26; Buffalo, Jacksonville, the Chargers and Washington had byes, and they, Cincinnati, Detroit, Miami, Minnesota, Kansas City and Carolina have played six games. Team offense and defense are EPA per play on regular-season 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; net EPA is offense minus defense. Ranks are among all 32 teams over Weeks 1-7 together unless a single week is named, in which case they are among the 28 teams that played. Records, scores and point differentials are from nflverse's schedule file; a one-score game is decided by eight points or fewer. Expected wins are the Pythagorean expectation, points scored to the 2.37 power over the sum of points scored and allowed each to the 2.37 power, times games played; wins ahead of or behind the points are actual wins, ties as half, minus that. The latest-game weight regresses each team's average net EPA per game over the rest of its season on two terms, the sum of its per-game net EPA before a given week and its net EPA in that week's game, for every team that played in Weeks 4 to 12 of 2016-2025 with at least four games left (2,600 team-weeks); 1.2 is the second coefficient over the first. The Week 7 correlations are Pearson correlations with rest-of-season win rate for the 284 teams that played in Week 7, using pooled per-play net EPA before Week 7, in the Week 7 game, and through Week 7. Turnovers are interceptions plus lost fumbles, credited to the team that lost the ball; stickiness correlates each team's per-game rate through Week 7 with its rate from Week 8 on (320 team-seasons), and net EPA, offense and defense are correlated the same way. Group records and playoff rates use each team's record and ranks through Week 7, or before Week 7 where stated, its record from Week 8 on, pooled, and whether it played a postseason game that season. Player lines are nflverse's weekly player stats; plays are retold from nflverse play-by-play. No charting figures are used this week. Data via nflverse.