College Football Picks — September 25, 2026

Today's College Football picks September 25, 2026 board features four matchups where the data points toward favorites in three of them. You have strong model conviction on Harvard, Navy, and Indiana, each carrying edges above 80 percent, while the Army-Temple contest sits near a coin flip. Overall lean stays with the higher-confidence sides that align with superior QB and line play in these spots. College Football picks today reward disciplined selection rather than chasing every number, so focus on where the percentages separate most clearly from the posted lines.

Harvard vs Brown — Harvard ML

Take Harvard at -600. The model assigns this side a 90 percent win probability, the highest on the board, reflecting a clear talent gap that home-field crowd noise at Brown is unlikely to overcome. Brown’s +425 price offers value only if the line moves further, but the data shows Harvard controlling the trenches and quarterback matchup decisively.

Navy vs UAB — Navy ML

Take Navy at -275. The model rates this 81 percent, driven by Navy’s established ground-game identity against a UAB defense that has struggled to slow similar option attacks. The 24-point spread differential in the moneyline implies the market already prices Navy as the stronger unit on the road.

Northwestern vs Indiana — Indiana ML

Take Indiana at -1600. The model gives this 80 percent confidence, underscoring Indiana’s superior offensive line and quarterback stability versus Northwestern’s limited ability to generate consistent pressure. Even at a short price the edge remains intact because the data shows minimal variance in this tier of matchup.

Army vs Temple — Army ML

Take Army at -150. The model sits at 54 percent, the lowest of the slate, indicating the slight lean stems mainly from Army’s rushing attack rather than a dominant overall profile. Temple’s +130 price keeps the game competitive, so this remains the board’s least-conviction selection.

Model edges on this board range from clear separation to near-even, so size bets accordingly and remember that percentages reflect historical tendencies rather than guarantees. Manage your bankroll with the same discipline the data demands and avoid forcing action on lower-confidence games.