San Jose State vs Eastern Michigan Prediction, Picks & Odds — Sep 4, 2026

Eastern Michigan sent 26 players into the portal and brought back 12, a net loss of 30 stars' worth of talent — the kind of teardown that widens every error bar around a season. It is favored by 5.5 anyway, and not because the rebuild went well. San José State returns one percent of last season's passing production. Continuity alone accounts for 8.9 points of the 17.3-point separation between these teams, more than roster talent and quarterback play combined.

0%

-38

0%

-32

0%

-24

1%

-18

3%

-10

6%

-4

12%

+4

17%

+10

19%

+18

17%

+24

12%

+32

7%

+38

3%

+46

San José State winsEastern Michigan wins

Market: Eastern Michigan −6Projected: Eastern Michigan by 17 (89%)

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

GRIDPEX.COM

The teardown left the passing game alone

What Eastern Michigan kept is the part that scores. Noah Kim has four seasons of record and +0.230 career EPA per play, and 80% of last season's passing production returns around him: Harold Mack was worth +0.85 per play on a 15% target share, Nick Devereaux +0.48, tight end Joshua Long +0.39. Those aren't complementary numbers, they're a functioning passing game, and San José State has no equivalent continuity anywhere on its offense.

Three of the twelve incoming transfers are tackles — Guzzy Kayed, Tyler Brumfield, Tyler McMillan — because Mack Indestad and Andrew Steger both left. So the unit with the least continuity on the roster is the one asked to protect the most experienced quarterback on the field. Against a San José State front that produced 18 sacks last season, 110th nationally, that is a survivable problem. Against a better one it would not be.

Kim projects into a 182-to-329 passing yard band with an 82% chance of a touchdown, and Devereaux is the likeliest man to catch one at 51%. The throws are available against a secondary that graded 108th last season. Staying on schedule is Eastern Michigan's second-biggest edge at 2.59 standard deviations, which is the tell: this offense is built to take first downs, not to hunt chunk plays.

Last week flattered the wrong team

Six days ago at USC, Luke Weaver threw for 234 yards on 32 attempts with two touchdowns and ran in a third. San José State scored on 40% of ten drives at 35.5 yards a possession — better rates than Eastern Michigan produced at home against Sacramento State, where 14 drives yielded 23.6 yards each and ended in a three-and-out on 42.9% of them. Carson Clark's 74 receiving yards came against the best defense either team will see for a while.

The opponents are not comparable, so the raw drive-rate comparison overstates San José State's case. What it does establish is that Kim's 279 yards on 38 attempts bought only 28 points, and that Eastern Michigan's offense grinds rather than strikes. That matters here because the preseason baseline — where the continuity gap lives — still carries the projection, and last week's evidence is the rare case that argues faintly against the favorite rather than for it.

San José State can't run, and that closes the game

The most lopsided matchup on the field is San José State's run game against Eastern Michigan's front, 3.24 standard deviations the wrong way, with short-yardage power close behind at 2.52. Last season's ranks read the opposite — a 64th-ranked rushing offense against a 115th-ranked rushing defense, which is precisely why those ranks can't be quoted as current facts. Both belong to rosters that have since turned over.

San José State returns 36% of its rushing production, and the receivers who forced defenses to respect the perimeter, Danny Scudero and Matthew Coleman, are gone, so light boxes are unlikely. Jabari Bates was genuinely good last year at +0.48 per play on a 39% share; he projects for 28 to 69 yards, and Tama Amisone for 11 to 28. That is not enough ground production to keep a 17-point projection honest, which pushes everything onto Weaver's arm.

Neither defense will take the ball away

Eastern Michigan finished 123rd in sacks with 15 and 121st in tackles for loss, then shipped edge rusher Messiah Blair to Jackson State. Quincy Byas and his three sacks are the returning disruption. San José State was barely better, 18 sacks, 117th in TFL; and the returners are Etu Finau, Ethan Powell and Solomone Langi, none of whom generated much of it, with Jayland McGlothen arriving from Sacramento State to help.

Both defenses intercepted seven passes last season, tied for 88th, so neither is set up to shorten the game with a takeaway. The most productive returning defender in the building is Bryce Llewellyn, whose 102 tackles, two interceptions and four breakups anchor a secondary that still graded 114th against the pass. Barry Manning adds two sacks from the same unit, safety pressure, not a front.

With no real rush on either sideline, this becomes a contest of sustained drives, and sustained drives are exactly where Eastern Michigan's 2.59-standard-deviation schedule edge lives. Kim gets time; Weaver gets time too, but he has to convert his with receivers who were somewhere else last November.

The one thing San José State is equipped to stop

San José State holds a single edge in the matchup grid, nearly a full standard deviation, and it is against explosive passing; the Ben Prosper 81-yard version of Eastern Michigan's offense is the one it can take away. Weaver's band runs 138 to 260 passing yards, with Carson Clark at 31 to 95 and Jerry McClure at 29 to 87 as the volume targets. Deny the shot plays, make Kim earn 12-play drives, and 5.5 points is a live number.

That path isn't an upset forecast, 11% is what it is. It's a description of the only shape this game takes for the visitors: Weaver near the top of his range, a total that finishes under 53.5; the likelier outcome at 64%; and Eastern Michigan's 42.9% three-and-out rate from last week showing up again in September at home.

What each side actually has to do

Eastern Michigan makes San José State one-dimensional on early downs and lets Kim take the underneath first downs; 30-13 is the projection, and the projected margin clears -5.5 with roughly 79% likelihood. San José State needs the run-game edge to be smaller than 3.24 standard deviations, McClure and Clark winning outside immediately, and a slow game. A one-score finish rates about a 1-in-4 chance here.