More throws under pressure, fewer of them caught. The sacks left. The 40% stayed.

Caleb Williams completed 40% of his throws under pressure in 2025. A year earlier it was 44.1%. The direction is wrong, but the volume moved further than the rate: 205 pressured attempts against 145 as a rookie, therefore the number that changed most is not the completion rate. It is how often he was throwing through pressure at all.

Where the number comes from

That figure reaches us second-hand. Gene Chamberlain printed both rates and both counts on Feb. 15 and attributed them to Pro Football Focus. PFF's free player page does not print 40% or 205; it prints 568 attempts, 637 dropbacks, 3,942 yards, 27 touchdowns, 7 interceptions and a 10.8% pressure-to-sack rate. We could not read the paywalled table, therefore the pole stands on Chamberlain's attribution, and we say so rather than pretend we read it.

Why the volume rose while the sacks fell

Chamberlain's own sentence holds the mechanism: attempts under pressure rose greatly even though the sacks he took decreased dramatically, because of better pass blocking and the threat of a running game. Windy City Gridiron's Johnathan Wood, working from PFF, has him pressured at the 10th-lowest rate and sacked at the 4th-lowest; 11% of his pressures became sacks, 3rd in the league, after 28% as a rookie, which was 2nd-worst. A quarterback who was getting sacked is now getting the ball out, but getting it out under pressure is not the same as completing it, therefore the pressured attempts pile up in the denominator of a rate that used to end in a sack instead.

Three sources print that sack number and they do not agree to the decimal. Wood has 11%. PFF's free page has 10.8%. PFF's 2026 QB Annual has 9.6%, 3rd among qualified starters behind Purdy at 9.0 and Nix at 9.5. Same direction, three windows or three roundings, therefore we print all three and collapse none of them into one figure.

What the rate is made of

Wood's fifth close-up puts a number on the shape: 22% of Williams' pressured passes were thrown away, highest in the NFL and almost double the 12% average, among 32 quarterbacks with 250 or more attempts. Kurt Warner said the same thing on 104.3 The Score in December: a couple of incompletions each game are throwaways after evading sacks. Williams said it himself in July. He throws it away the most to stay in advantageous positions because Coach will go for it on 4th down. A throwaway is an incompletion, but it is not a sack and it is not a turnover, therefore a completion-under-pressure rate can fall while the offense is getting better at surviving pressure. That does not make the rate good. It tells you what the rate is made of.

The same piece has his accuracy 7% below the league average under pressure against 3% below when clean, scrambles on 11% of pressured dropbacks (10th, average 8%), and teams blitzing him at the 4th-highest rate, with pressure on 42% of those blitzed dropbacks, the league average, and 30% when not blitzed against an average of 33%; throwaways ran 10% of blitzed attempts (average 5%) and 6% otherwise (average 4%). Wood reads the blitz rows as a bad idea for defenses on the 2025 tape, but his line about coordinators backing off next year is a projection, therefore it is his forecast and not a row here.

He held it longer

He held the ball longer: 3.2 seconds on average by Next Gen Stats, per Chamberlain, against 2.9 in 2024; only Shedeur Sanders was slower at 3.24. Chamberlain ties it to an offense built on boot passes and play-action throws. Wood's third close-up dests the same style: play-action on 17% of rookie attempts and much higher in 2025, a Ben Johnson trend from Detroit; about two-thirds of attempts over 2.5 seconds; out-of-pocket escapes 2nd-highest, with out-of-pocket ANY/A and rating poor. PFF's page prints 3.25 seconds, a different warehouse and a different decimal, therefore two prints, not one. Holding the ball is a style choice in this offense, but it is also more seconds in which pressure can arrive, therefore the pressure count and the time to throw are the same story from two sides. Nothing on these pages names a scheme family. We do not either.

The rows that ran the other way

Chamberlain's other rows ran the opposite direction: passing EPA 12th among starters at 43.96 by Sumer Sports after 73rd; play-action throws second-most in the league with the grade up 30 spots; third-down rating 97.2, 8th, after 76.0 and 27th. Season completion still fell, 62.5 (351 of 562) to 58.1 (330 of 568), the worst of any playoff starter since Andrew Luck in 2012 at 54.1, and Ben Johnson's 70% target was missed. Full season, EPA against pressure finished 21st among starters; against the blitz, 4th; when trailing, 3rd at 65.82. Those rows sit next to the pressure rate, but they do not erase it, therefore the honest sentence is that the quarterback got much better and still completed four in ten when the pocket broke.

An independent print, a different window

Mike Tanier printed 41.8% under pressure through six games, with 12.0 pressured dropbacks per game, 4.6 yards per attempt and a 2.3 ANY/A that ranked 25th of 37 quarterbacks with 25 or more pressured throws, the Bears 4-2 and time to throw already stretching from 2.92 to 3.17 by NGS. The same day Chamberlain had in-season EPA under pressure at -33.01 by Sumer and 3.3 yards per attempt outside the pocket at about 42% by TruMedia, with Johnson saying the route runners were not as disciplined as we wanted executing scramble rules. A six-game 41.8 is not a full-season 40, but it is the same direction from an independent print, therefore it corroborates the pole without becoming it.

What this is not

One playoff game is not the season. In the wild-card win over Green Bay PFF charted six big-time throws, noted 25 big-time throws on holds over 3 seconds led the NFL in 2025, and had the lowest adjusted completion of the weekend by about 10 points; PFF names Cover 2 on two Loveland throws. That is texture. It is not the regular-season pole, and it does not license a coverage name we did not read.

It is not a preview of the Vikings game. That is a slate row and a different pole; the join is empty and stays empty here. It is not a roster story. The cut deadline is Sunday at 6 p.m. Eastern and nothing official has posted, therefore no name in this piece is a roster claim.

What moves it

The pole reopens when a first-party PFF page prints the rate or the count, when the NGS table for last season reads first-party, or when a second full-season print arrives that is neither Chamberlain nor a tweet. Until then: 40%, on 205, in 2025, after 44.1% on 145.