Week 1 gave us 43 FBS-vs-FBS games, a perfectly split ATS card before the push and no repeat of Week 0’s all-under sweep.
The board found useful large underdogs and absorbed three enormous favorite misses. The exact retained totals card finished 2–1; a separate frozen full-slate check was also profitable. Those records belong beside each other, not blended together.
43FBS-vs-FBS games
11–11–1Published ATS
2–1Exact totals card
15–12Frozen totals check
The opening read
Week 1 punished certainty.
The published ATS card finished 11–11–1. Rutgers lost outright as a 29.5-point favorite, Tulsa beat a team favored by 13.5, and Michigan won by one while laying 27.5. Those were not all the same kind of miss, but each argues for restraint when a preseason projection looks precise.
The 43 games finished with 24 overs and 19 unders against the saved totals. The average final was 51.72 points against a 52.34 market average. Week 0’s 5–0 under start was worth investigating; Week 1 supplied the larger reminder that direction alone is not a betting case.
The smaller one. The distinction matters more than the nicer number.
The exact retained public card contained three bets across the final four games: Washington State–Washington under 51.5, Louisville–Ole Miss under 55 and Wisconsin–Notre Dame over 46.5. It finished 2–1, good for 0.9 units when risking one unit at −110.
For the full review, all 43 games were also rescored from the frozen through-2025 totals model, using the saved consensus and no Week 1 outcomes or weather. That check went 15–12 and improved average error by 0.13 points per game versus the saved market.
The 15–12 result is a useful diagnostic reconstruction. It is not the historical live card. The ledger below labels every reconstructed opinion so the two records cannot be mistaken for each other.
The same Week 0 format, reserved for the games that changed the argument or clarified the record.
01
SHI Stadium · Massachusetts 37, Rutgers 21
Massachusetts at Rutgers
37–21
The largest market shock arrived before the weekend had properly started.
Rutgers entered as a 29.5-point favorite and lost 37–21. The result traveled 45.5 points across the saved spread, the widest market miss on this 43-game board. All Takes passed the side, which avoided a loss but did not anticipate the upset.
This is the kind of opener that belongs in the next rating update without being allowed to become the entire rating update. Massachusetts earned the result; one game still cannot tell us how much of the gap was a true level change and how much was opening-night volatility.
What travels to Week 2
A useful pass, a major rating signal and the clearest warning against treating preseason certainty as game evidence.
Tulsa turned the most confident favorite on the ATS card into an outright loser.
All Takes laid 13.5 points with Oklahoma State. Tulsa won 24–10. That made the ticket 27.5 points wrong against the spread and produced the largest loss among the 23 published Week 1 ATS recommendations.
The miss was also broad. The frozen totals reconstruction leaned over 57.5 in a game that reached only 34. When both the expected margin and scoring environment fail by that much, the review has to treat it as football information rather than late-game noise.
What travels to Week 2
This game gets a full re-check in the Week 2 ratings; there is no cover-margin excuse to soften it.
The marquee underdog case never reached competitive football.
Clemson +10 looked like one of the board’s sturdier underdog positions. LSU won 51–10. The 31-point ATS miss was the largest by cover margin, and the Tigers’ 41-point win cleared both the market expectation and the model’s confidence in Clemson.
The full-slate totals check passed this game, so the review does not manufacture a second opinion after the score. The side was the call, the side was wrong, and the scale of the miss matters more than explaining it away.
What travels to Week 2
A decisive side loss that should change the prior more than an ordinary missed cover.
Michigan Stadium · Western Michigan 12, Michigan 13
Western Michigan at Michigan
12–13
Michigan won the game and still missed the number by nearly four touchdowns.
Michigan closed −27.5 and escaped Western Michigan 13–12. The All-Takes side missed the cover by 26.5 points. The frozen totals check also leaned over 49.5, turning this into the rare game where both projected separation and projected scoring were plainly too high.
A one-point win should not be read as a one-point team rating, but it is powerful evidence against the precision of the opening projection. Week 2 should absorb the result while keeping the opponent and the single-game sample in view.
What travels to Week 2
The win column hides the board’s clearest favorite-rating problem outside Oklahoma State.
Boise State supplied the cleanest large-underdog win on the card.
All Takes took Boise State +25. Oregon won 34–27, leaving 18 points of room on the ticket. The Broncos did more than survive a backdoor window: they stayed inside the number for the full result and made the model’s skepticism of the inflated spread useful.
The total offers the counterweight. The frozen reconstruction preferred under 52.5 and the teams scored 61. A strong side does not make every opinion on the same game right.
What travels to Week 2
Keep the side win intact and the total loss separate; the underdog price was the valuable read.
Notre Dame closed the late card with agreement across the two live lanes.
Notre Dame −20.5 covered in a 41–13 win. The exact retained totals card also played over 46.5, which cleared by 7.5 points. It was one of only three totals bets preserved as an auditable live Week 1 record, and one of two that won.
The frozen full-slate reconstruction reached the same over side through its probability layer. That agreement is useful context, but the public result remains the simpler fact: one side win and one totals win at the displayed numbers.
What travels to Week 2
A clean late result for both published All-Takes calls, with no grading ambiguity.
All Takes took Florida State +3. SMU won 27–24, landing exactly on the saved spread. A push is neither a rescued win nor a disguised loss; it returns the stake and belongs in the record as its own result.
The retained totals card passed, and the frozen full-slate check did too. The 51-point final offered no reason to revise those decisions after the fact.
What travels to Week 2
The honest Week 1 ATS line is 11–11–1. Dropping the push would make the record less accurate, not cleaner.
ATS grades use the published recommendation and saved price. The retained totals lane is intentionally blank outside the exact late card. SP+ is graded only on the projected winner, never as an unmade spread bet.
Week 1 finals
Thursday, September 3
4 games
01
Massachusetts at Rutgers
37–21
Massachusetts won by 16; Massachusetts covered. 58 total points, 3.5 above the saved 54.5 total.
Week 1 supplied real rating information without supplying enough evidence for a structural model rewrite.
01 / Team strength
Make the biggest misses answer football questions.
Rutgers, Oklahoma State, Clemson and Michigan each missed in a different context. The next ratings should absorb those results team by team instead of inventing one universal Week 1 penalty.
02 / Totals
The full-slate check earned continuity, not promotion.
The frozen totals diagnostic went 15–12 and edged the saved market on average error. That supports keeping the model stable. It does not turn reconstructed bets into a forward record.
03 / Big spreads
Separate rotation risk from rating failure.
Ohio State and Texas covered enormous numbers; Michigan did not come close. Large favorites are not one bet type. The opponent, projected scoring range and likelihood of a full-game effort still matter.
04 / Recordkeeping
Keep passes and pushes visible.
A pass avoided the Rutgers upset, and Florida State +3 pushed exactly. Neither belongs in the win column. Preserving both makes the board more useful when the next slate tests the same decisions.
Recordkeeping note
The published board stays frozen.
ATS results are settled against the saved All-Takes recommendation and its recorded line. Late movement and postgame explanations never change the grade.
The exact totals record includes only the three preserved public bets. The full-slate lane is separately reconstructed from the frozen through-2025 model and labeled on every card.
Final scores come from the repository’s retained CFBD results pull. Game links open the corresponding ESPN recap and box score.