Research The one totals angle that cleared our gate
The one totals angle that cleared our gate
We tested dozens of college football angles against eight seasons of closing lines and killed almost all of them in public. One survived: early season unders on high totals. Here is the rule, the evidence, and the live count as the 2026 test grades itself.
Most of what this desk has tested does not work. That is not a complaint, it is the finding: our public graveyard holds more than forty angles that looked real in a spreadsheet and died against closing lines, from wind thresholds to altitude to returning production. We publish the kills because they are what make the survivors worth anything.
This article is about a survivor.
The rule, stated before kickoff
FBS regular season, weeks 1 through 3. Closing total of 47 or higher. The side is the under, measured against the closing no-vig total.
That is the whole rule. No team filters, no weather overlay, no discretion. It was pre-registered before the 2026 season kicked off, which means the rule text, the qualifying threshold, and the grading method were written down and frozen while the result was still unknowable. Whatever happens in weeks 1 through 3 this season, we cannot move the goalposts afterward.
The evidence behind it
Across 2018 through 2025, qualifying games went under 54.75 percent of the time, on a sample of 884 games. Break-even at standard pricing is 52.4 percent.
The mechanism is boring, which is a point in its favor. Early season totals leans on offenses that have not been seen yet: returning skill players are visible in August, rebuilt defenses are not, and the market prices the hype side of that asymmetry. High totals concentrate the effect, because a 47 plus number already assumes two functioning offenses in week 2.
We treat the history as reason to test, not proof. Plenty of dead angles had eight good seasons behind them, and a market that reads the same public studies we do can reprice any of this. Which is why the rule now has to pass a forward gate it cannot see coming.
The live test, graded in public
For 2026, the pre-registered gate is 52.38 percent unders or better on qualifying games across weeks 1 through 3. The count starts with the first week 1 final and every qualifying game is graded automatically against its closing no-vig total, win or lose.
The running count lives on the early totals tracker, which updates as finals land. Nothing on that page is editable after the fact: qualifying games are determined by the closing number, grades come from final scores, and the rule was fixed before the first snap.
If the gate clears, the angle keeps its place in the weekly signal stack and its label upgrades from tested to validated. If it fails, it goes to the graveyard with everything else, and the graveyard entry will cite this article.
During weeks 1 through 3 the angle also feeds the weekly game board as a totals signal on qualifying games, labeled with its sample like every other signal there.
The limits, before anyone asks
Three weeks is a small window: at typical volume the 2026 test will grade something like 100 to 150 qualifying games, enough to fail loudly but not enough to prove anything alone. A passing grade extends the evidence, it does not end the question. And the angle is descriptive history, not a guarantee: 54.75 percent means losing about 45 times in 100, priced in from the start.
That is the deal this desk offers on every number: the rule in advance, the count in public, and the same treatment for a miss as a hit.