John Middlekauff Wins Both Weekly Awards on Three Calls — One of Them Carried It | MissedTakes Weekly
John Middlekauff takes both Pundit of the Week awards for the week of 2026-W37 — accuracy and MissedTakes Score — off three graded calls across three pieces of work. He hit all three, for a perfect 100 accuracy, and posted a MissedTakes Score of 66.67, meaning his average call beat the naive baseline by a wide margin.
Three calls is a lean, not a record. Two of the three were the sort of pick a coin-flip-avoiding robot makes anyway. The third is the one that won the week.
The call that did the work
On the Week 1 game in Melbourne, Middlekauff took the underdog 49ers over the Rams, arguing San Francisco's long acclimation trip and Kyle Shanahan's edge over Sean McVay made them a live dog. His words: "But I think the 49ers won the game. I do. ... So 49ers to win the game. Yes."
San Francisco won 27-7. Because the naive pick in that spot was the Rams, the call scored 100 for accuracy and a skill figure of +200 — the maximum credit our scoring gives for beating the default. That single call is essentially his entire weekly score.
His other two were both Seattle over New England in the opener — one from his season-preview run ("like them to win by three and a half, like them to win outright"), one from the game preview, where he called for a 10-point Seattle win. Seattle won 13-10. Right on the winner, worth nothing on the skill side: Seattle was the naive pick, so both graded 100 accuracy and zero skill. Correct, but not information.
That's the honest shape of the week: one genuinely contrarian call, two freebies, and an award.
What his fuller record looks like
Middlekauff has 106 graded calls with us across 31 pieces of work — enough to have a real read on him, not a flash sample.
| Measure | Figure |
|---|---|
| Overall accuracy | 60.49 |
| MissedTakes Score | 10.47 (above a naive baseline, not by much) |
| Graded calls | 106 |
Where he's actually good is the draft. His mock-draft calls grade out at 76.12 accuracy with a score of 50 over 19 graded picks — including Fernando Mendoza at No. 1 overall, which he called in January, again in March and again in April, and which came in exactly right. Game picks are his second-best room: 73.81 accuracy, 45.24 score, on 10 graded.
Where he's bad: Super Bowl picks (20 accuracy over eight graded) and trade-and-signing calls, where 10 graded calls come out at 50 accuracy and a score of −66.67 — worse than simply assuming every veteran stays put, which is exactly what the baseline assumes. In February he flagged Travis Kelce, Chris Jones, Trent McDuffie, Kyler Murray and DJ Moore as movement candidates. All five stayed where they were.
The consensus he beat
The 49ers-Rams opener is the cleanest example this week of a field leaning one way and getting run over. Counting every call we graded off that game, 20 had the Rams winning and 10 had the 49ers. Two-to-one for Los Angeles; final score 27-7 the other way.
The 49ers side included Tyler Sullivan's list of Week 1 upset candidates and Eva Geitheim's straight-up picks, each worth the same +200 as Middlekauff's. The Rams side was full of well-known names, and it cost them nothing in skill terms — picking the favorite and losing grades zero accuracy but zero skill loss, because the baseline made the same mistake.
The week in aggregate
We graded 96 calls over the last seven days. Game picks and roster calls did most of the volume as Week 1 settled a huge backlog of preseason talk.
The roster-spot calls produced both ends of the week. On the good end, Andrew Moore went three-for-three on Colts roster questions — Anthony Richardson Sr. making the 53 and winning QB2, Spencer Shrader over Blake Grupe at kicker, and UDFA corner Kenneth Harris having an outside shot — for a perfect 100 and +200 skill. Christian Booher hit the same numbers on Ben Stille making the Lions.
On the bad end, five roster calls scored zero and lost the maximum against the baseline, all of them practice-squad and inactive-list projections: Charean Williams on the Cowboys elevating Israel Abanikanda, Ralph Ventre on Buffalo elevating Greg Dortch to return punts, Patricia Traina's projected Giants inactive list, Randy Gurzi's 17-man Cowboys practice squad, and Matt Matera on Tampa Bay cutting Sean Tucker.
That's the genre in miniature. Across everything we've graded, roster-spot calls average 87.4 accuracy — most of them are safe — which means the ones that miss tend to miss all the way to zero. Transaction-week guesswork is cheap to publish and expensive to grade.
For the overall standings, Lance Zierlein still leads at 39.1 with 72.15 accuracy over 38 graded calls; the full board is on the leaderboard, and the scoring is explained in the methodology.