San Francisco 49ers 2026: The Coverage Report — 69 graded cutdown calls, and the field came in below a coin-flip baseline

Across the whole database, the San Francisco 49ers carry 749 tracked takes, 279 of them graded. That is the eighth-largest take count of the 32 clubs, against a league median of 639. The graded pile ranks ninth, level with Baltimore.

So this is not a coverage-gap story. The Bills (1,050 takes) and Rams (936) draw more, and Arizona (466), Atlanta (351) and Indianapolis (388) draw much less, but San Francisco sits comfortably in the top quarter — roughly where a club that went 12-5 and made the playoffs in 2025, after 6-11 in 2024, ought to sit.

The average graded take about the 49ers scores 62.1 accuracy with a skill delta of +31.11 — above the naive baselines our methodology grades against. That headline number hides a split, and the cutdown is where the split shows up.

The cutdown: 133 calls, 69 graded, and a negative skill score

Roster Spot Calls are the club's largest graded category: 133 takes, 69 graded, 66.7 average accuracy — and an average skill delta of −20.29. That is the one negative-skill category of any size on the 49ers ledger. Two-thirds of the individual claims were right, but the field collectively did worse than the dumb rule our grader uses as its yardstick ("he was on somebody's 53 last season, so he makes one again").

Timing explains the volume: 126 of the 133 came in August 2026, 64 of them graded so far. Another 64 roster-spot calls across the file are still pending, so this record can move.

Where the naive rule beat the pundits

The misses cluster on one move: writing off a veteran who had a 53-man spot last year. Of the 23 graded misses, most are exactly that shape.

Tom Jensen's pre-cutdown 53-man projection on August 21 listed a long release-and-practice-squad group. Graded, it is the roughest 49ers record on the site: 16 graded roster calls, 18.8 accuracy, −112.5 skill. Eli Apple, Siran Neal, Luke Gifford, Malik Turner, Brayden Willis, Luke Farrell, Robert Jones, Quinton Bell, Ashtyn Davis and Khalil Herbert were all logged as cuts or practice-squad landings; all graded 0 with −200 skill.

Khalil Herbert drew the same call from a second source. Carter Bahns, in a CBS Sports list of veterans at risk, wrote that Herbert "looks to be one of multiple roster-limit casualties, as San Francisco can only carry so much insurance for Christian McCaffrey." Graded 0, skill −200.

Jose Sanchez called Jacob Cowing "definitely a cut candidate after camp" on August 15. Graded 0.

Where the pundits beat the rule

The +200 rows are the reverse: players the baseline expected to be gone who stuck.

Brian Peacock's position-by-position 53-man projection supplied most of them — Cowing, Gracen Halton ("he's for sure gonna make the roster"), De'Zhaun Stribling, Romello Height, Carver Willis ("he's locked in, he's a fourth round rookie"), Kaelon Black, and Kurtis Rourke, who Peacock argued "you cannot cut… because he is your future backup." Each graded 100 with +200 skill. Back on May 5 he had also put Ephesians Prysock on the roster; that one is another +200.

Jensen, whose overall roster grade is the worst here, produced the single best contrarian read in the file. He kept Cowing on the 53 for a specific reason: "He would have been a boomerang cut, but then the punt return touchdown against the Chargers. Supply and demand for punt returners is so dire Cowing would be at risk of being picked up." That is +200, on the same player Sanchez had gone the other way on. Jensen also kept James Thompson Jr. — "cutting him would be a risk" — for another +200, while Parker Hurley's practice-squad projection put Thompson on the squad and graded 0.

Hurley's best is his August 15 line that Nick Bosa, Mykel Williams and Romello Height are "locks". Bosa scored 100 at zero skill — the baseline agreed. Height scored 100 at +200, because it did not.

Peacock is the only pundit with a substantial 49ers roster sample: 44 graded roster calls at 88.6 accuracy, +22.73 skill. Jensen's 16 and Hurley's 4 are thin, and Hurley's four in particular are too few to rank anyone on. One caveat we should state plainly: two of Peacock's graded rows (Jordan James, James Thompson Jr.) are logged with a claim that reads opposite to the quoted evidence, and both graded as misses. Take his 88.6 as, if anything, conservative.

The rest of the ledger

Nineteen prediction types carry at least one graded 49ers take. The substantial ones:

Two things stand out. First, the field was emphatically right that this team would play in January — 23 graded playoff calls at 91.3 accuracy and +138 skill is the strongest category on the page. Second, it was emphatically wrong about *how*. Of 11 graded division-winner calls, one is a hit — Ryan Reynolds on the 2023 team. Nine tracked pundits picked San Francisco to win the 2025 NFC West and all nine graded 0, among them Brady Quinn ("I've got them winning the division at twelve and five"), Aaron Schatz with the FTN projections, Jason McIntyre, John Middlekauff, Chris Brockman, Jonas Knox (twice) and Chosen Anderson. Skill on all nine is 0, because the naive pick — last year's division winner, the Rams — was also wrong. Seattle finished 14-3.

The win-total field was directionally shy. Thirteen 2025 calls landed on San Francisco, averaging 10.5 against 12 actual wins, with a range from 7 to 13. Quinn's 12 was exact.

Several categories are logged but barely graded: 110 depth-chart takes with 10 grades, 95 sleeper calls with 9, 55 player-season-stat calls with 4, 36 fantasy-bust calls with 1. Those are not records yet, and we will not treat them as any.

Two pundits clear a five-take floor on the club overall: Peacock (56 graded, 80.7 accuracy, +19.54) and Jensen (16 graded, 18.8, −112.5). Everyone else — Kyle Crabbs at 13 graded, Rich Eisen at 12, Brockman at 11 — is in the range where one call swings the number.

The takeaway

The 49ers get covered at a top-eight rate, and that coverage graded well on the big directional question and badly on the small operational one. The cutdown file says it cleanly: 66.7 accuracy, −20.29 skill. Predicting that a proven veteran gets released is the bet that keeps failing here, and the pundits who beat the field — Peacock across a real sample, Jensen on the one player he thought hardest about — did it by keeping people, not cutting them.