Central Michigan vs New Mexico Prediction, Picks & Odds — Sep 5, 2026

New Mexico is an 11.5-point home favorite over Central Michigan, and our projection wants 16.5. The five-point gap reads modest until the stadium comes out of it. The decomposition closes exactly — in-season form 0.0, last season's carry +5.12, roster talent +4.39, returning production +0.44, transfer portal +2.17, quarterback -0.14, skill transfers +0.02, home field +4.50, summing to 16.5 with a residual of 0.0. Strip the 4.50 and our neutral-field number is New Mexico by 12.0. Charge the market's -11.5 the same 4.50 and it is 7.0 — a disagreement worth 71 percent of the market's number, not the 43 percent the raw five points suggest. That step assumes the books price this home field where we do; if they price it lower, the neutral gap narrows back toward the raw one. There is no book dispersion to hide behind: Bovada and DraftKings are both at -11.5, both at 47.5.

Two terms supply 9.51 of that 12.0: last season's carry at 5.12 and roster talent at 4.39. Returning production, the portal, quarterback and skill transfers add the remaining 2.49, nearly all of it portal accounting. Central Michigan brought in 12 and lost 17, 19 incoming star-points against 50 outgoing, net -31. New Mexico went 14 in and 14 out, net +7. Our own read on that class calls it turnover, not continuity, and asks for wider error bars; the portal term spends 2.17 points on it anyway.

The talent bar runs against the talent numbers

The third-largest term credits New Mexico with 4.39 points of roster talent — home -0.96 against away -5.34. The published talent figures in the same file run the other way. Central Michigan grades 481.2, 121st nationally; New Mexico grades 378.8, 131st. That is 102 talent points and ten rank places in the underdog's favor. The component is described as “recruiting pedigree plus an independent preseason roster grade,” so the grade is overriding the pedigree by more than four points, in the opposite direction. Neither recruiting class argues otherwise: New Mexico's 23 commits rank 65th, Central Michigan's 21 rank 80th, neither with a blue-chip.

What the grade appears to track is the power rating, which is the problem, because the rating is already in the number. New Mexico rates +0.7 and Central Michigan -9.6, a stated gap of 10.23; carry (5.12) and talent (4.39) sum to 9.51. If the grade is derived from last year's rating, New Mexico is paid twice for the same season. The cheap check sits in the league scatter: Hawai'i carries a talent score of exactly 481.2, identical to Central Michigan's, and rates +2.5 — 12.1 points clear on the same pedigree.

The biggest term is one season, with nothing from 2026 against it

Last season's body of work is the single largest term at 5.12 points, larger than home field. Nothing from 2026 sits beside it: form weight 0.0, prior weight 1.0, effective games 0. New Mexico's 9-4 (27.1 points scored, 22.3 allowed) carries the full prior weight, and Central Michigan's 7-6 (22.5 and 23.5) does the same.

Our own analog work argues against taking that 9-4 at face value. Among 155 FBS teams whose win total rose by four or more (jump seasons 2014-2024), the following year averaged 8.1 wins against 9.4, with 57 percent declining; the floor mostly held, 74 percent still winning seven or more. Our season-level model concedes as much, projecting New Mexico for 6.1 wins over a 10-game schedule that includes a trip to Oklahoma. The game-level carry term does not.

The tightest analog has produced almost none of this

Filter to what this game is — a Week 0 or 1 opener, non-neutral site, both teams outside the SEC, Big Ten, ACC, Big 12 and Pac-12, home team laying 8 to 15 — and 2014-2025 yields 18 games. The favorite won ten of them, 55.6 percent. The dog won outright 44.4 percent. Favorites covered 38.9 percent. The average margin was 3.2 points on an average total of 49.5, and the favorite won by 20 or more exactly once, 5.6 percent of the cell.

Loosen the filter and it regresses toward normal without reversing: at 6 to 17 points (n=25) the favorite won 60.0 percent by an average of 5.8; across any home favorite in that spot (n=63), 63.5 percent by 8.8. The counterweight is in the same query: Week 1 alone, all of FBS, home favorite of 8-15 is n=60 at 73.3 percent and 11.0; all weeks, same bucket, n=1,282 at 77.9 percent and 11.7. If opening week were the mechanism, those would sag too. An 18-game cell is not cleanly separable from a 60-game one, so treat the estimates as directional. The analog also controls for the price, not the mismatch: New Mexico ranks 66th in power rating, Central Michigan 121st.

It is our shape, more than its level, that stands out. Games matched on both spread and total (n=981, 2013-2025) finished within one score 39.1 percent of the time. All FBS favorites of 8-15 (n=2,455): 36.9 percent. Week 1 favorites of 8-15 (n=148): 32.4 percent. Our 40,000-run simulation makes this a one-score game 25.0 percent of the time, a quarter to a third below every historical cut here. The mass it takes out of the middle sits in the tails: 43.6 percent of runs finish 20 or more apart, though that figure counts either side winning and does not line up against the cohorts' favorite-by-21 rates.

0%

-38

0%

-32

0%

-24

1%

-18

4%

-10

7%

-4

12%

+4

17%

+10

18%

+18

16%

+24

12%

+32

7%

+38

3%

+46

Central Michigan winsNew Mexico wins

Market: New Mexico −12Projected: New Mexico by 16 (86%)

The range of outcomes for this game. Each bar is how often the final margin lands in that range — the more it leans toward New Mexico, the more likely New Mexico wins.

GRIDPEX.COM

Both Sides' listed edges are 2025 measurements

The four unit crossings split two apiece. New Mexico takes both on the ground — its 0.15 rushing offense, 48th of 136, against a Central Michigan rush defense at 0.04, 31st, and Central Michigan's 0.07 rushing offense, 90th, against a New Mexico rush defense at the same 0.04 and 31st. Central Michigan's two are in the air, the bigger being its passing offense — 0.40 PPA, 22nd — against a New Mexico pass defense that allowed 0.26, 73rd. That is the underdog's keep-it-close script, and it belongs to a unit returning 37 percent of its production. Quarterback Joey Labas is gone at a 32 percent share, Langston Lewis at 0.08, D.C. Temple at 0.05; Tyson Davis (0.125 usage, 0.693 EPA) left for Virginia. The replacement is an unresolved three-way race — Angel Flores a soft leader at 0.23 credibility, Jadyn Glasser at 0.50, Marcus Beamon at 0.04. Returning: Tommy Mcintosh (0.163 usage, 0.416 EPA), Collin Payne (0.876 EPA) and four of five projected line starters.

When Central Michigan has the ball— its offense vs New Mexico’s defense

Passing: Central Michigan O #22 vs New Mexico D #73Central Michigan · slight edge

Rushing: Central Michigan O #90 vs New Mexico D #31New Mexico · slight edge

◄ New Mexico defense winsCentral Michigan offense wins ►

When New Mexico has the ball— its offense vs Central Michigan’s defense

Passing: New Mexico O #63 vs Central Michigan D #42Central Michigan · slight edge

Rushing: New Mexico O #48 vs Central Michigan D #31New Mexico · slight edge

◄ Central Michigan defense winsNew Mexico offense wins ►

Read it by possession: each side shows what happens when that team has the ball — its offense against the other team's defense, unit by unit (national rank in parentheses). Last season's grades, shrunk toward average by how much of each unit's production actually returns. Central Michigan returns 37% of last season's passing production and 100% of its rushing production. New Mexico returns 59% of last season's passing production and 71% of its rushing production. Central Michigan's projected line: 4 of 5 held over, 1 portal; New Mexico's projected line: 5 of 5 held over.

GRIDPEX.COM

The stale-rank problem is worse on the other side. New Mexico's defense ranked 20th nationally in sacks, 33 in 12 games. Twenty-two of those belong to departed players — Keyshawn James-Newby (8.0), Brett Karhu (7.0), De'Jon Benton (5.5), Noah Avinger (1.0), Jayden Wilson (0.5) — against 5.0 for the five listed returners, with pass-defense continuity at 0.32. Central Michigan is in the same spot, less extremely: 14 of its 27 sacks departed, headed by Michael Heldman's 10.5, continuity 0.24. Both lists are top-five extracts, so read them as lower bounds.

Model and market agree on the total to within a point: 48.5 against 47.5 posted, modal score 31-17. The disagreement is therefore about how the pool gets divided — 32.5 to 16.0 — not about the scoring environment. Opener cohorts skew under harder than either number implies: 72.2 percent in the 18-game cut, 68.0 in the widened 25, 68.3 across all 60 FBS Week 1 games with a home favorite of 8-15. The all-weeks version of that bucket is 52.3 percent (n=1,282); the 981 lookalikes, 48.2. Our simulation puts 48.4 percent of outcomes below 48 points — a threshold half a point above the posted number, so it is not a read on 47.5. We publish no lean on the total.

Start from the base rate. FBS favorites laying 8 to 15 have covered 50.1 percent of the time across 2,455 games since 2013 while winning 78.6 percent outright, average margin 12.1. We price New Mexico to cover -11.5 at 62.6 percent and to win at 86.0 against a market-implied 80.3. That is twelve and a half percentage points above the historical cover rate for this price, built on a prior with zero 2026 games in it, a talent term pointing away from the published talent, and unit ranks attached to rosters that have turned over. Our own betting note says the same: this early the market has better information than a projection built on last season's rosters, and the gap is a disagreement to track rather than an edge to bet.