49ers 24, Broncos 14 — Week 4: A Fumbled Punt, Then Kittle

Three to three at halftime. The 49ers had been stopped on fourth and one at the Denver 34 on the first snap of the second quarter, and Denver's Allen blocked a 54-yard Pineiro field goal as the half ran out. Then, tied with nine 52 left in the third, a 49ers punt from their own 11 went 61 yards, Marvin Mims fielded it at the Denver 28, and San Francisco's Neal forced the fumble and recovered it himself. San Francisco scored nine snaps later, scored twice more in the fourth quarter, and beat the Broncos 24 to 14 at home to go four and oh. Neither offense turned the ball over, and neither quarterback was sacked.

The passing game opens on how much of its day came on one snap. 29 pass plays were worth +8.1 expected points added, which measures what a snap was worth given the down, the distance and the field position. Over the first three weeks this passing game ran at about +6 tenths a play, a rate at which those 29 would have been worth about 17.8. Before halftime, 15 pass plays gained 89 yards for +0.4. Then the snap that decided it. 17 to 14, two 25 left, second and eight at the San Francisco 44, right after Denver's first timeout. Brock Purdy threw short left to George Kittle three yards behind the line, and Kittle ran 59 yards after the catch for a 56-yard touchdown. +5.5 expected points, about two-thirds of the day's passing total, and 15 points of win probability, the model's read on each team's chance of winning at that moment. Mike Evans drew a team-high eight targets and caught five for 76, and his 17-yarder in the fourth quarter set up Christian McCaffrey's one-yard touchdown. The charting reads cleaner than the box score. Purdy's throws were catchable on 20 of 29, 69 percent, below the 80 he ran over the first three weeks, but none of the 29 was contested, after 12 percent across those weeks, and none was interception-worthy. The fakes went away. Play-action came on four of 33 dropbacks, 12 percent, last of the 30 charted offenses this week, against 28 percent over the first three weeks. Motion did not move: 51 of 79 charted plays, 65 percent, second of the 30, right at the team's own rate.

The run game opens on its success rate, because the yards flatter it. 33 rush plays, a hundred 42 yards, and 45 percent of them successful, meaning they gained what the situation asked for. By expected points they came to −0.7; at the first three weeks' rate of about +8 hundredths a carry, the same 33 would have been worth about +2 and a half. McCaffrey carried 15 times for 52 yards and the touchdown, and his carries cost about two and a half expected points. Deebo Samuel carried three times for 44, including 22 20 yards, and was worth +3. Denver showed six or fewer defenders in the box on 16 of 29 designed runs, 55 percent, tenth-most of the 30 charted offenses and about where the first three weeks were. Those 29 designed runs gained a hundred 18 yards, 4.1 a carry, against four and a half over the first three weeks. The costliest run was the fourth-and-one sneak at the Denver 34: Purdy up the middle for no gain, −3.5 expected points and 10 points of win probability, San Francisco's worst snap of the day.

The pass defense opens on field position, because Denver almost never got close. The Broncos ran one snap inside the San Francisco 20 all afternoon, and it was the touchdown. 42 Denver pass plays were worth +4.7 expected points; on defense a positive number is the bad direction. At this defense's rate over the first three weeks, +4 hundredths a play, those 42 would have been about +1.8. Before halftime Bo Nix's 24 pass plays came to −3.2; after it, 18 came to +7.9. The defense got off the field on nine of 17 third downs. It also changed its rush. San Francisco sent five or more on 19 of Nix's 44 dropbacks, 43 percent, eighth-highest of the 30 charted defenses this week, after 26 percent over the first three weeks. No sacks came of it. Nix's throws were catchable on 28 of 42, 67 percent, close to the 70 percent opponents managed in the first three weeks, and none of the 42 was contested, after 18 percent.

The run defense held its line but did not improve on it. 22 Denver rush plays, 82 yards, −0.02 expected points, essentially zero; at the first three weeks' rate of about −7 hundredths a carry, about −1.6. On defense, negative is good, so this was a shade worse than its own standard. 20 designed runs gained 72 yards, 3.6 a carry, against 3.7 allowed over the first three weeks, with six or fewer in the box on 10 of them, 50 percent, up from 42. Where it gave ground was Denver's touchdown drive midway through the fourth: five runs for 30 yards from J.K. Dobbins and RJ Harvey, then Jaylen Waddle's 13-yard touchdown and a two-point run that made it 17 to 14.

The turning point is the punt. +6 and a half expected points and nearly 21 points of win probability, the largest swing of the game. It came on the snap after a third and seven from the San Francisco 11, where Purdy's deepest throw of the day, 48 yards in the air for Evans, fell incomplete. From the Denver 28 the drive still needed a fourth-down conversion: Kaelon Black for two on fourth and one at the 19, worth almost six points of win probability. McCaffrey was stopped at the one on second down, and on third and one Purdy threw short left to Samuel for the touchdown. 10 to three.

The situational line. Third downs: six of 13. Red zone: 13 snaps, two touchdowns. 10 drives: three touchdowns, a field goal, the failed fourth down, and three three-and-outs.

The stat lines, scored half-PPR: Purdy, 19.6, on 19 of 30 for 218 yards and two touchdowns, +10 carries for 29. Samuel, 15.5: five catches for 26 and the touchdown, plus the 44 rushing. Kittle, 15 even: four catches for 70 and the touchdown. McCaffrey, 14.5. Evans, 10.1. For Denver, Harvey caught all 10 of his targets for 68, 14.3.

What repeats. Last week's notebook asked whether Purdy's drop to a 59 percent catchable rate against Arizona was one afternoon or the start of something, and what box Denver would show. Both got answers. The catchable rate came back to 69 percent with nothing contested, so Arizona looks more like the afternoon. Denver's box was ordinary, light on just over half the runs, and the run game got four yards a carry out of it. For anyone new: San Francisco has beaten the Rams 27 to seven, Miami 35 to 13, Arizona 36 to 30 and Denver 24 to 14. The line for the notebook now is the play-action. 28 percent across the first three weeks, 12 percent on Sunday, last of the 30. Watch whether the fakes come back, or whether this passing game has decided it does not need them.

That's your 49ers — every game, by the numbers. This is Muffed: we do the same deep dive for all 32 teams. Follow the 49ers show wherever you're listening, so next week's episode lands the moment it's out. No theater. Just the football.

_Data via nflverse. Charting: FTN Data via nflverse._