Our Troy vs Utah State prediction starts with an odd split. The model's number has Troy winning by 3.6 points, so on the straight-up question it likes the Trojans. The market disagrees just enough to hang Troy at +3, with the moneyline on Troy at +126.
The official tier for this game is PASS. We are not backing Troy +3 on Saturday, September 26, 2026, and it is not a bet we are making. What follows is the reasoning behind the number, because the case is worth reading even when the card stays empty.
Why the model is on Troy +3
The model projects Troy by 3.6 and made no adjustment to that baseline. Its win path for Troy runs through sustained offense, and specifically through Crowder. His participation on September 19 is verified, which corrects an earlier claim that his season was over. Participation is not the same as medical clearance, so the model does not treat him as fully healthy.
Kinchen is in the same bucket. His touchdown shows he played, not that he is unrestricted. We give Troy credit for the players who are on the field and no more.
The strongest evidence of competitiveness is the 17-27 road loss to SEC opponent Missouri. Losing by ten to a stronger opponent away from home says Troy can hang with better teams. That is why a +3 line on a team the model rates as the outright winner draws attention.
How Utah State keeps it close
The risk note on this side is specific. Utah State repeats its Washington defensive performance at home, which keeps Troy's modest offense within reach of a few successful Brosterhous possessions. Utah State lost that game 14-16, so the defense showed up and the offense did just enough to stay in it.
Brosterhous is a two-start quarterback, and Logan gives the home team crowd support. If the defense plays like it did against Washington, Troy does not need to be great. It needs a few clean drives, and the model's read is that the Trojans have a path to those. The same script also gives Utah State a way to stay within a field goal.
The case against Troy
Troy's other results do not show dominance. The Sam Houston win was narrow, and the Alabama State win came against an FCS opponent. Neither one establishes an edge in the trenches, and the model has not reviewed film for either team. That matters when a small number like 3.6 rests on offense that is only modestly rated.
The injury picture adds fog. Hillstead's historical injury report shows neither recovery nor an indefinite absence. Taggart's rehabilitation leaves defensive communication and leadership to replacements. Taylor's eligibility is uncertain, so we cannot credit his safety depth. None of this earned a numerical adjustment, but all of it argues for caution.
Utah State has its own problems. The loss to FCS Idaho State and the shutout at Utah show a fragile offensive floor. That floor is why the Aggies are hard to trust as well.
The number that matters
The number is the gap between 3.6 and 3. The model has Troy ahead by 3.6, and the market gives Troy three points. That is a thin cushion on a game where both offenses have shown they can stall. A projected margin that close to the spread is not a lot of room for the injury questions above to go the wrong way.
The +126 moneyline on Troy is the other side of the same coin. It prices the Trojans as the underdog while the model sees the winner. We note that disagreement and leave it there. It is a fact about how the game is priced, not a bet we are placing.
The verdict
The tier is PASS: an official-card PASS with no official action. We are not backing Troy +3, and we are not asking you to. The model's Troy-by-3.6 read, the Missouri result and the Utah State offensive floor are all useful context for anyone who wants to form their own view. For us, this one stays off the card.
- Event · College Football Week 4 · Saturday, September 26, 2026
- Market · Troy +3 +126 — recorded when the pick was posted
- Model · Heismann, Botcappers' AI College Football model — how it is built and what it does not know
- Record · every pick is timestamped before the event and graded in public, win or lose
- Statistics · the model's own match/game database, current to the posting time above. Figures a reader cannot open here are internal to that database, not published third-party data.

