Last week's issue ended with a number worth repeating: on charted dropbacks in 2025, quarterbacks produced +0.257 expected points per play from a clean pocket and −0.464 under pressure. One condition, and the best thing in football becomes the worst.
This week is about that hidden layer — the sport under the sport. Charters logged what actually happened on all 19,811 dropbacks last season: who rushed, who got home, what coverage was called, what the offense dressed the play as. Almost none of it makes a broadcast graphic. All of it decides games.
Start with Drake Maye. His 2025 was the best throwing season in football — that was last week's story. Here's this week's: from a clean pocket, Maye averaged +0.549 expected points per dropback, video-game stuff. Under pressure, he averaged −0.243 — worse than the league does on an ordinary play. Same season, same arm, same receivers. The difference between the league's best offense and a broken one was whether one rusher arrived. That happened on about three of every ten dropbacks league-wide, and it is as close as football gets to a master switch.
So just send more rushers, right? About that.
Sending extra rushers — the blitz — is sold as the aggressive path to pressure, and it does buy some: rushing five got pressure on 38.3% of dropbacks, against 26.2% rushing the standard four. Six or more got there 45.3% of the time.
Now the part the highlight package skips: quarterbacks facing five rushers actually produced more per play (+0.064) than they did against four (+0.033). Send six and you're roughly back where you started (+0.027). And the supposedly safe opposite — rushing three and dropping everyone into coverage — gave up more than any rush count in football, +0.121 per dropback. Across all 32 teams, there was essentially no relationship between how often a defense blitzed and how well it defended the pass.
What blitzing actually buys is variance — more sacks (9.1% of dropbacks at six-plus rushers, versus 6.0% at four), and more disasters when the ball comes out anyway. The proof it's about scheme, not aggression: Minnesota blitzed more than anyone in football, on 46% of dropbacks, and allowed the second-fewest expected points per dropback in the league. Cincinnati blitzed the least, 17.7%, and finished 28th of 32. Same dial, turned to opposite ends, and the results flipped the other way. The blitz is a knife — it matters who's holding it.
Man or zone: the coin every defense flips
Every coverage call is a version of one choice: chase receivers across the field (man), or defend space and read the quarterback (zone). The league played zone 69% of the time in 2025, and the split explains more about individual players than almost anything in the box score.
Against man coverage, completions crater — 49.4%, versus 59.8% against zone. But here's the boom-bust: offenses averaged more per play against man (+0.067) than against zone (+0.031). Man coverage takes away the easy stuff — the hitch, one of football's most routine completions, actually lost ground against man — and in exchange it risks the catastrophic play, because there's no one behind you when it breaks.
Which means "who beats man coverage" is a real scouting report, and 2025's answers were blunt. Jordan Love was the best quarterback in football against man, +0.402 per dropback, first of the 22 qualifiers. And Patrick Mahomes — the best player of his generation — had the largest zone-over-man split of anyone: −0.081 per dropback and 50.3% completions against man, +0.237 (third-best in the league) and 63.1% against zone. In 2025, the book on Mahomes was simple: get in his receivers' shirts.
The honest asterisk, and it's a big one: that book was written against a diminished season — Mahomes played 14 games before the ACL tear, with the lowest accuracy-over-expectation of his career. He's been back on the field at Chiefs camp this month. Whether the book survives a healthy Mahomes is one of 2026's genuinely open questions, and we'll be tracking it weekly.
The most efficient pass in football came dressed as a run
Offenses have their own version of the hidden layer: what a play looks like before the snap versus what it is.
In 2025, the single most efficient pass in football, by personnel grouping, came out of "13 personnel" — one back, three tight ends, the most run-shaped formation a team can line up in without a fullback convention. Defenses saw it and loaded the box, an average of nearly seven defenders. Teams ran from it 59% of the time, keeping the disguise honest. And when they passed instead: +0.138 expected points per pass play — more than double what the league's standard three-receiver sets produced (+0.059).
It was also just 5.1% of the league's snaps, which is the concession and the point at once: the edge exists because it's rare, and rare because most rosters don't carry three tight ends who can both block and catch. But the principle scales, and it's the cleanest one-line summary of modern offense we know: the defense defends what it sees, and the best offenses lie.
The run game is where you cash it
If the run-look pass is the payoff, the run game is the setup — and the data is clear-eyed about it: running lost ground on average against every box count in 2025. You don't run because running gains value; you run to control what the defense shows you.
Which makes the extremes instructive. Baltimore saw a light box — six defenders or fewer — on just 33.4% of their runs, the lowest share in football. Defenses respected the run enough to sell out against it, and the Ravens ran through it anyway: one of only six teams in the league to average positive value per rush. That's earning the hard way. Las Vegas got the opposite deal — light boxes on 66.1% of their runs, among the five softest looks in the league — and still finished dead last at −0.306 per rush, nearly double the deficit of the next-worst team. Football's version of missing a layup, for an entire season.
Real or Noise — scheme edition
The verdicts, calibrated. These grade the story, not the team.
- "Blitz more, win more": leaning noise. More rushers bought more pressure but not better defense (+0.064 allowed at five rushers vs +0.033 at four), and league-wide the blitz-rate-to-results relationship was roughly zero. What's real: blitzing buys sacks and chaos, and a great scheme can weaponize it — Minnesota did. Volume isn't the skill; design is.
- "The book on Mahomes is out — play man": leaning real, scoped to 2025. The gap was the largest of the 22 qualifying quarterbacks, and it's built on 185 charted man dropbacks, not three bad Sundays. The scope matters: it was measured through an injury-shortened, career-worst-accuracy season. Real last year; unproven against whatever comes back this year.
- "Screens are the safe call": leaning noise. The screen was one of the least valuable common throws in football in 2025 — negative against zone (−0.045) and against man (−0.092). Safe from interceptions, sure. Also reliably safe from gaining anything.
The Receipts
Verified numbers worth keeping in your pocket.
- George Pickens was the most dangerous man-coverage target in football: +1.005 expected points per man-coverage target — first in the league — with 607 yards on 52 such targets and a 69.2% catch rate. When defenses played Dallas straight-up man, they were volunteering.
- *Drake Maye ranked second in the NFL against man coverage (+0.359 per dropback) and second against zone (+0.267).* The best throwing season of 2025 had no coverage answer — only a pressure one.
- The post was the most valuable route in football — and coverage didn't matter: +0.603 expected points per targeted play against man, +0.603 against zone. Identical, to the third decimal. Some things you just have to cover.
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 always in the footer.
The audio version of any team's season is one click away: Kansas City, Green Bay, New England, Dallas, Minnesota. During the season, Muffed builds a weekly show for your favorite team or your fantasy roster — muffed.ai.
And if someone forwarded you this: the Read is free, every Wednesday, and it's better when it just shows up. Subscribe and it will.
Next week: The luck audit — which 2025 results won't survive 2026. A decade of data has strong opinions about close games, turnovers, and one team in particular.
Methodology: pressure, blitz, coverage, personnel, box-count and route figures: FTN Data via nflverse, 2025 regular season, computed on 19,811 coverage-charted dropbacks (~99.9% of regular-season dropbacks). Expected points added (EPA) — the measuring stick throughout — values every play by how much it moved the scoring math. Team and player context from nflverse play-by-play, 2025 regular season. QB coverage splits: 22 quarterbacks qualified on charted dropbacks. Data via nflverse.