UConn at Southern Miss. Lean: Southern Miss +3. Lean: Under 54.5.

Spread.Edge Labs makes Southern Miss -1.9 versus a market of UConn -3 to -3.5, but the disagreement rests on suspect inputs: the dossier treats Southern Miss as retaining 90.2% of its production without a coaching change, whereas the team’s current preview says Blake Anderson is a first-year coach with 74 newcomers and only two returning starters. Correcting that removes the model’s apparent home-team value. UConn’s 223-yard performance against Maryland was poor, but Kalieb Osborne remains a credible dual-threat starter; backup Jake Merklinger’s questionable designation should not move the spread materially. Southern Miss quarterback Ethan Hampton has produced only 6.5 yards per attempt with three interceptions. Returning WR Davis Dalton and potentially RB Robert Briggs improve the home offense, while extreme heat strengthens home field, so I would not lay UConn -3.5. At the available -3, however, UConn’s better transfer-built roster and more established coaching operation justify a small position. My corrected line is approximately UConn -3.5.

Total.I lean under 54.5 despite the model’s 58.42 projection. The over case has merit: the forecast has negligible wind, UConn owns explosive-play capability, and Southern Miss may regain Dalton and Briggs. However, the totals model leans heavily on 2025 offenses that no longer resemble these transfer-rebuilt teams. UConn ran only 52 offensive plays and gained 223 yards against Maryland. Hampton has averaged 6.5 yards per pass with three interceptions, while Southern Miss scored only eight at Auburn. Both Week 1 explosions came against FCS opponents and should be heavily discounted. Heat could tire defenses, and turnovers could manufacture short fields, preventing a stronger position, but I price the total near 52.5.

Where our number looks wrong.The Southern Miss inputs appear materially wrong: 90.2% returning production and no first-year coach conflict with the current report of 74 newcomers, two returning starters and first-year coach Blake Anderson. The model also overreacts to two final margins against radically different opponents and cannot distinguish starter Kalieb Osborne from injured backup Jake Merklinger. Its Southern Miss -1.9 projection is not credible; I make UConn roughly -3.5.

Spread.Our in-season rating has these teams nearly even (SoMiss -6.93 vs UConn -6.46), so with 2.5 HFA our number lands near pick'em and reads value on the home dog getting 3; the market at UConn -3 splits the difference between preseason priors (UConn ~ -12 neutral) and the compressed in-season picture. The fundamental edge for UConn is real (its passing game vs a Southern Miss run defense that surrendered 223.5 rush yds/gm), but that number is inflated by drawing Auburn's SEC line, and UConn's own offense managed just 14 on Maryland. UConn is a portal-rebuilt roster (2.3% returning production, first-year HC Jason Candle) traveling to a hostile Hattiesburg night game in 89-degree heat. Home dog + tough environment + a cohesion question on the road nudges me to the points. Low conviction, essentially a lean.

Total.Both teams lit up FCS bodies then got shut down by Power-4 defenses (UConn 14 on Maryland, SoMiss 8 at Auburn), suggesting real offense against a comparable opponent is capped. UConn's edge is run-based against a soft front, which tends to shorten the game and drain possessions, and UConn's own run defense (94.5 rush yds/gm allowed) plus Southern Miss's questionable WR/RB can limit the home side's scoring. Heat is a factor but wind is a non-issue (2.6 mph). Slight lean under, low conviction.

Where our number looks wrong.The in-season refit compressing UConn and Southern Miss to near-even likely overrates Southern Miss: it reads final margins only, so it rewards the Golden Eagles' 49-3 rout of FCS Alcorn while ignoring that UConn's preseason SP+ prior (+5.1 vs SoMiss -7.1) reflects a genuine talent gap, and it is blind to UConn's clear passing-vs-run-defense matchup edge. The rating also does not see that neither team's QB room is what the priors were built on (Braxton gone at SoMiss, full portal rebuild at UConn).

Spread.But line movement independently supports the Southern Miss side: 4 of 5 tracked books shaved the UConn number from -3.5 at open (9/13) to -3 by 9/16, with only DraftKings moving the other way. Both teams are 1-1, having blown out FCS opponents in Week 1 before getting outclassed by real competition in Week 2 (Southern Miss lost 8-43 at Auburn; UConn lost 14-38 vs Maryland), so the gaudy per-game offensive numbers in the dossier are largely FCS-inflated. The UConn QB injury note is close to a non-issue since it concerns third-stringer Jake Merklinger, not starter Kalieb Osborne. Leaning the home dog, at home, run-heavy, against a UConn team on its first true road trip.

Total.Both offenses were held well down against the one quality opponent each has faced (Southern Miss scored 8 at Auburn; UConn scored 14 vs Maryland), which reads as a better proxy for this FBS-vs-FBS matchup than the FCS-inflated Week 1 scoring the per-game averages carry. Hot kickoff conditions (89F, negligible wind) and a Southern Miss offense built around a 183.5 rush-yard-per-game clock-control identity add a mild further lean toward a lower-possession, lower-scoring game than the listed number.

Where our number looks wrong.The in-season refit pulled Southern Miss's rating up almost to UConn's level largely off two results: a 46-point margin over FCS Alcorn State and a 35-point loss at Auburn, while the preseason prior had UConn nearly 4 points better than Southern Miss. That FCS-margin inflation, combined with the model reading none of our own injury/availability data, means the 5-point model-vs-market gap looks more like a rating-mechanics artifact of an early-season small sample than a real signal - consistent with the model's own documented 46-50% ATS hit rate across every gap bucket in the holdout. We're leaning Southern Miss here on the line-movement signal, not on trusting the model's number.