UTEP 17–52 Michigan: AI predicted 7–35 ✓

The Consensus Is Total: Michigan Wins — the Only Question Is Margin

All four models land on the same winner for the same core reason, and the evidence is unusually clean for a Week 3 mismatch: common-opponent data. Michigan beat Oklahoma 17-10; UTEP lost to that same Oklahoma team 51-0. A 58-point divergence against identical opposition is about as strong a signal as non-conference scheduling ever provides, and it's corroborated by every underlying number — Michigan's defense has allowed one touchdown in eight quarters with zero red-zone TDs surrendered, while UTEP managed 198 yards and zero points in Norman. Nothing in the panel disputes the winner.

Where the Experts Disagree — and How I Resolved It

The real disagreement is about Michigan's offensive ceiling, and the models split into a modest spread: GLM-5.3 projects 38 points, Grok 4.6 just 31, with Claude Opus 5 and GPT-5.6 in between at 34. Two factual conflicts needed resolving:

• Rest/turnaround: GLM framed UTEP as coming off a 14-day bye; Claude's correction is more credible — UTEP beat Texas Southern on September 12, meaning a normal seven-day turnaround. This slightly weakens the 'rested underdog game-script' narrative but doesn't change the fundamental shape of the game.
• Weather: Grok forecasts 80-90% chance of rain; GPT-5.6 calls for clear, mild conditions. With conflicting reports, I treat weather as a mild suppressant at most rather than a decisive factor — but note it cuts in the same direction as everything else if it materializes: fewer possessions, more runs, less scoring.