UTEP at Michigan. Pick: UTEP +35.5. Pick: Under 48.5.

Spread.Take UTEP +35.5. EL makes Michigan -30.3, a 5.2-point difference, but its failed ATS validation means that is context rather than an established edge. My stronger argument is Michigan’s limited offensive production: its Oklahoma win featured 263 yards and a decisive touchdown drive starting at the opponent’s six. Covering requires sustained scoring that Michigan has yet to demonstrate. The market already recognizes this: the dossier’s opening prices of -36.5 to -37.5 have shortened, reducing the opportunity. Getting above 35 still matters. The strongest Michigan argument is UTEP’s 51-0 loss at Oklahoma; another near-shutout plus short fields could overwhelm this cushion. Colson’s rebound against Texas Southern supports basic offensive competence, but deserves a substantial opponent discount. Hansen and Forbes remain questionable on Covers, without independently confirmed absences, so I assign no specific injury adjustment. My judgment favors a comfortable Michigan win with insufficient offensive separation; uncertainty about UTEP’s scoring floor keeps conviction modest.

Total.Take under 48.5. Michigan’s clock-controlling approach against Oklahoma and UTEP’s shutout against its only FBS opponent support the under independently of EL’s 46.94 projection. The over case is credible: Michigan faces a much weaker defense, UTEP just demonstrated explosive scoring, and turnovers or late reserve possessions can inflate a mismatch’s total. New personnel and coaching also weaken last-year-based estimates. Nevertheless, I trust Michigan’s defensive advantage more than an offensive breakout large enough to carry this total. The move from 49.5 to 48.5 has already absorbed some under value. Thunderstorm risk provides only secondary support; kickoff wind was not independently verified, and EL already includes a wind deduction. I use the corroborated 48.5 rather than an inaccessible Oddschecker snippet showing 50.

Where our number looks wrong.UTEP’s rise from -15.034 preseason to -9.975 after a 51-point FBS loss and an FCS blowout looks overly generous; opponent-level distortion could make -30.3 too short. The -1.1% returning-production input also needs a data audit. Conversely, stretching the raw Michigan margin from 25.7 to 30.3 does not establish that its offense can sustain a blowout. The totals documentation contradicts itself: it says weather is absent, yet the supplied calculation already subtracts 2.656 points for wind. Do not apply that adjustment twice.

Spread.Michigan has scored 30 points in two games and is being asked to win by 36. Kyle Whittingham's first Michigan team beat Western Michigan 13-12 on a 47-yard Hail Mary after a second was put back on the clock, then beat Oklahoma 17-10. Bryce Underwood was the lowest-graded Michigan offensive player on PFF against WMU at 35.8, with an interception and a fumble that gave WMU six of its twelve points. This is a Whittingham team: elite defense, run-heavy, clock-draining, no interest in a 42-point fourth quarter. I just do not believe this offense gets to 38, and Saturday's forecast makes it harder.

Total.Michigan's two games have totaled 25 and 27 points. Whittingham's defense has allowed 12 and 10, and the offense cannot get out of its own way. UTEP was shut out by Oklahoma, the same Oklahoma that managed 10 points in this stadium two weeks later. The realistic script is Michigan grinding to the mid-20s or low 30s, UTEP scoring once or twice, and a final in the 30s. The NWS forecast for Ann Arbor on Saturday is a chance of showers through 2pm and showers and thunderstorms after 2pm with a 3:30pm ET kickoff, which points at a wet ball, a shortened passing game and possibly a delay. The market has already dropped the total a point at four books. I would rather be on the under at 48.5 than chase the last point.

Where our number looks wrong.Our number is -30.3 and calls this a pass at a 5-point gap, which understates how much I want the points. The rating cannot see the thing that matters most here: Michigan changed head coaches, changed both coordinators, and the resulting offense has scored 13 and 17. The in-season refit also still carries a preseason prior built on a 54% blue-chip ratio and 69% returning production, which describes a talent level Michigan is not currently converting into points. So the model is right-ish on the side and badly wrong on the reasoning, and I would be more confident at -37 than I am at -35.5 after the market already moved two points my way.

Spread.Michigan is 2-0 but has scored only 13 and 17 points in its first two games (a 1-point win over Western Michigan, a 7-point win over a ranked Oklahoma team), genuinely stingy defense (22 points allowed combined) but an offense that has not shown it can blow a team out. Laying 35.5 with a team averaging 15 ppg is a lot to ask, even against a UTEP team that was shut out 51-0 by Oklahoma. UTEP did respond by beating an FCS team 51-10, which tells us little. Michigan QB Bryce Underwood has been effective on the ground (87 yards, a TD vs Oklahoma) which helps control the clock more than it explodes a score. Add wet, breezy conditions (dossier: 73F, 13.5 mph wind, 47% precip) and I don't see the explosive, multi-score-margin outcome the number implies. I'll take the points with UTEP, though this is a soft lean given Michigan's clearly superior overall talent and defense.

Total.My lean under is built on three things: forecasted rain/wind at kickoff (73F, 13.5 mph wind, 47% precipitation per the dossier) which tends to suppress passing efficiency and scoring pace; Michigan's defense has been legitimately stout (22 points allowed combined against Western Michigan and a ranked Oklahoma team); and Michigan's own offense has not shown explosive scoring even in wins. If Michigan builds a big lead on a UTEP team that was shut out 51-0 by Oklahoma, I'd expect a run-heavy, clock-control second half rather than a track meet, which caps the total. The counter-case (Michigan's offense finally breaks out against a bad defense, pushing this well over) is real but I don't have fresh evidence supporting it yet.

Where our number looks wrong.The model is in its own pass zone here, so it isn't claiming an edge, but I'd flag that its rating is still mostly preseason prior plus two results and has no way to read that Michigan's per-game scoring output (15 ppg) has been modest all season regardless of opponent quality, it only sees the wins, not the offensive profile. If Michigan's true offensive ceiling is higher than shown so far, the model (and I) would be underrating the home side; that's the main way this pick is wrong.