Penn State vs Michigan Prediction — College Football Week 7, Saturday, October 17, 2026

The Penn State vs Michigan prediction for Saturday, October 17, 2026 comes with a lopsided number attached: Michigan is favored by 6.5 points, and the moneyline on the Wolverines to win outright sits at -245.

Run the projection and Michigan clears that spread. But clearing a spread on paper and clearing the price the market is charging for it are two different questions, and this is one of the weeks where the model stays off the field entirely.

Why the model leans Michigan

The model starts Michigan at a 7.9-point favorite, then adjusts: subtract 2.0 points for offensive line continuity risk, add back 0.5 for an extra week of preparation. That lands the projected margin at 6.4 - still a Michigan win, but a number that reflects real questions up front, not a comfortable blowout.

Underwood and Becht are the readable, healthy starting quarterbacks on each side. Earlier-season chatter about a Michigan quarterback competition doesn't carry weight here; five straight Underwood starts settle that question. The model's path to a Michigan win straight up runs through defense first - contain Penn State's offense, then lean on sustained possessions against a Penn State front that's shorthanded.

The injuries doing the real work

Michigan's blocking picture is the swing factor in that -2.0 adjustment. Sprague is out. Babalola is back at practice, but a practice return isn't a game clearance. Separate concerns around Efobi and Frazier threaten pass protection without rising to confirmed absences - enough uncertainty to knock points off the projection, not enough to call it settled.

Penn State isn't clean either. Tatsch and Granville are out, Smith's practice progress hasn't translated into clearance, Howard remains out and limits Becht's targets, and Riker's injury complicates the interior line. Both rosters have holes. The model reads Michigan's protection questions as the heavier variable for this particular matchup, which is where the discount in its number comes from.

The number that matters

Both defenses have trended well. Michigan has allowed 16.7 points per game across its last three (Oklahoma, Iowa, Minnesota). Penn State has scored 18.7 over its own last three (Wisconsin, Northwestern, USC) - two points above what Michigan's defense has been giving up lately.

That's a real number, but it's a thin one on its own. Both teams also have lopsided wins over Group of Five opponents on their resumes, and neither of those games tells us anything meaningful about trench play in a game like this one. The model isn't treating either team's scoring trend as proof the line is wrong in either direction.

Why this is a pass, not a play

This grades out as a PASS. The model projects Michigan to win this game outright, but it is not backing Michigan -6.5, and no bet is going out on this side of the card. That's the full extent of the tier: not a lean, not a smaller-stakes version of a play, simply not a bet we're making today.

Penn State's own path to staying competitive, or even finding a way to win, leans on reproducing the level it played against Wisconsin and getting enough offense through Becht to win a game with fewer possessions than usual - particularly if Michigan's blocking issues push its offense back toward the 14-to-19-point output it has shown at times in conference play this season.

The verdict

Kickoff is Saturday, October 17, 2026, with Michigan sitting at -245 on the moneyline to win outright - a number that lines up with the model's own lean. The spread is the piece we're not touching, and that's where this one stays: a game the model likes straight up, and a bet it isn't making.