LSU (2-0, No. 7) and Ole Miss (2-0, No. 8) play the closest game on a 71-game Saturday board — a 3-point line, the tightest number attached to any ranked matchup today. Both teams arrive unbeaten, and for once the market's respect for the underdog looks earned rather than padding.

Except the underlying numbers don't agree it's that close. LSU's SP+ rating (24.1) runs nearly nine points ahead of Ole Miss's (15.2), and the gap is almost entirely on defense — LSU grades as one of the stingier units on the board (10.9 SP+ defense) against an Ole Miss defense that grades more than a touchdown worse (20.8). The two offenses, by contrast, are close to even (34.8 to 35.4). If anything, Ole Miss transfer Trinidad Chambliss is having the better individual season of the two starting quarterbacks — 0.494 expected points added per play across 74 dropbacks, ahead of LSU transfer Sam Leavitt's 0.320 across 88.

None of that is a projection — it's a description of what each team has actually done through two games, and it says the market is pricing this as a coin-flip-adjacent game between two teams that, category by category, haven't been equally good. A number that opens at 3 and can't get away from it is worth watching as the day goes on.

Georgia (2-0, No. 2) is a 24.5-point favorite at Arkansas (1-1), the kind of number the scoreboard will likely confirm by halftime. The more interesting number belongs to Gunner Stockton, whose 33 dropbacks this season have produced 38.2 expected points added — a 1.16-per-play average that would be an outlier even in a full season's workload, let alone a small one.

It isn't just the quarterback. Georgia's offense as a whole is averaging 0.90 expected points a play with a 70.7% success rate, both marks that grade out among the country's most efficient units two weeks in. Arkansas's defense, by contrast, is allowing a positive 0.24 expected points a play and letting opponents stay on schedule 42% of the time — a defense getting outclassed in almost every phase, not just the score.

Arkansas's own quarterback, KJ Jackson, has actually played a full workload (86 dropbacks) and hasn't been the problem — a modest but respectable 0.23 expected points a play. The gap here is almost entirely about which sideline's defense can stay on the field, and Georgia's has allowed only -0.20 expected points a play of its own.

Notre Dame (2-0, No. 3) hosts Michigan State (2-0, unranked) as a 29.5-point favorite, and the defense is the reason the number is that big. The Irish are allowing -0.29 expected points a play — a negative number means the average snap against them actively loses value — with a 24.1% havoc rate that ranks among the more disruptive units on the board.

Michigan State's offense is nearly the mirror image of what Notre Dame has already done to two opponents: 0.12 expected points a play and a 45.3% success rate, both below what an average offense produces. Put a defense that stingy against an offense that stalled-out, and 51.5 total points looks like a lot to expect.

The number makes more sense once garbage time enters it. A 29.5-point favorite typically empties its bench well before the fourth quarter, and backup-on-backup snaps are usually where a total like this actually gets paid — not the Notre Dame starting defense holding serve against Michigan State's. C.J. Carr has been efficient in his own right for the Irish (0.61 expected points a play across 51 dropbacks), which is part of why the running clock should arrive early.