Ball State at Liberty. Pick: Ball State +14. Pick: Under 49.5.

Spread.EL's Liberty -2.3 is 12.2 points away from this market, but that is not a demonstrated betting edge. My judgment is closer to Liberty -10: favor the home team while questioning its ability to separate by three scores. Liberty needed a fourth-quarter quarterback change to escape Gardner-Webb; Purdie's passing struggles make laying more than two touchdowns uncomfortable. Luster's efficient performance gives Ball State a credible route to sustaining drives, although an FCS opponent limits the evidence. The strongest Liberty argument is that Henderson could improve the offense immediately, while Ball State remains largely untested between the extremes of Ohio State and Stony Brook. Captured spreads show no broad move toward the underdog, so the market is not endorsing EL's assessment. Nevertheless, the hook above 14 offers useful protection against a favorite whose offensive execution remains unsettled. Unconfirmed availability keeps this a clear lean rather than a strong position.

Total.Under 49.5 is a modest lean after much of the apparent value has disappeared. EL's 49.77 provides no support. The under case rests on Liberty's inconsistent quarterback execution and misleading recent scoring: Ball State, Stony Brook was 27-3 with six minutes remaining before a late touchdown exchange produced 58 points. Liberty also ran 45 times Saturday, supporting a possession-consuming approach, although I could not verify a season-level tempo change. The over deserves respect: Luster showed passing improvement, Henderson may unlock Liberty's offense, and both teams demonstrated explosive-play potential. Captured totals already fell sharply, with substantially different opening numbers across books; that is not proof of informed injury betting. Forecast uncertainty and unconfirmed injuries receive no additional scoring deduction. At the current number, the under advantage is small.

Where our number looks wrong.EL appears to overrate Ball State's improvement: its rating jumped approximately 7.9 points after a 53-point loss and an FCS victory. The preseason ratings plus 2.5-point home field imply Liberty approximately -12.3 before calibration, much closer to the market. Two extreme opponent matchups do not establish near-equal team strength. Liberty's quarterback change also makes a margin-only rating incomplete. The totals estimate offers essentially no edge and cannot distinguish sustainable scoring from late explosive touchdowns.

Spread.Two weeks of football say these teams are close and the market is still pricing the August ratings. Liberty needed a 54-yard touchdown with 2:34 left to beat winless FCS Gardner-Webb 24-23, its first win since 11/1/25, and did it after Chadwell benched starter Deshawn Purdie, who was 6-of-17 for 122 yards with minus-18 rushing. Gardner-Webb had 23 first downs and 304 yards against that defense. Liberty's leading back Justin Marshall missed the game with a leg injury. Meanwhile Ball State put 482 yards and 41 points on Stony Brook with Keldric Luster completing 27 of 32. The market's -14.5 is almost exactly the preseason SP+ spread (Liberty minus-9, Ball State minus-23 plus home field), and that prior has not been updated at all: the spread has not moved a half point since Sunday while the total fell nine. I do not need our model's full 12 points to like this. Fair is Liberty by 7 or 8.

Total.So 50 is the honest market number, and our ridge agrees almost exactly at 49.77 - no model edge either way. I lean under on personnel. Liberty is the team that gets the most snaps as the favorite and it is the worst offense in the game: 13 points at James Madison, 24 against a winless FCS team, a benched starting quarterback, a four-man camp competition and its leading back out with a leg injury. Chadwell's offense ran 257 yards on the ground, which burns clock. Ball State's 41 came against FCS. I make it roughly Liberty 25, Ball State 20.

Where our number looks wrong.Our minus-2.3 is too kind to Ball State, and the dossier tells us why in its own words: 'against an FCS opponent the number is least reliable.' The in-season refit moved Ball State from a minus-18.443 preseason prior to minus-10.557 - nearly eight points - on the strength of a 41-17 win over an FCS team and a 56-3 loss the rating apparently forgave because Ohio State is elite. That is a lot of credit for a roster with 5.7 percent returning production that has not beaten an FBS team. The rating also under-rates Liberty's 20-13 loss at James Madison, comfortably the best opponent either side has played close. Fair here is roughly Liberty by 7 to 8, which means I think our model is about five points too low and the market is about seven points too high. Taking the points is still clearly right at 14.5, but anyone reading a strong 12-point gap as a full-size play is reading a number the FCS result inflated.

Spread.This is our largest gap of the slate (-12, 'Pick'), with our number seeing Liberty as only a marginal ~2-point home favorite against a market that has held at -14 to -14.5 (steady since opening, per five books captured 2026-09-14). Every outside source I found disagrees with our number and agrees with the market: aggregated computer-model previews give Liberty [figure removed], and a 2026-09-13 blog.youwager.lv preview backs Liberty -13.5 on roster and home-field grounds. That said, the on-field evidence for Liberty's offense is thin: QB Deshawn Purdie (Wake Forest transfer) went just 6-of-17 (35%) for 122 yards against Gardner-Webb, and Liberty's first two games have combined for only 33 and 47 total points, both well under this week's number even before Ball State's defense factors in. Ball State's Keldric Luster was sharp (27-of-32, 3 TDs) against Stony Brook, though that was a lighter FCS test. Given Purdie's inefficiency and how low-scoring Liberty's games have run, I'll take Ball State and the points, with real uncertainty given how one-sided the outside win-probability models are.

Total.The market total has fallen hard this week: outside preview sites (dimers.com, blog.youwager.lv) were still working off numbers in the 54.5-58.5 range as of 2026-09-13, while our own captured lines show the real market has settled at 49.5-50 by 2026-09-14, a swing of roughly 5-8 points. That move lines up with the underlying results: Liberty's two games have totaled only 33 and 47 combined points, and QB Deshawn Purdie has been an inefficient passer (35% completion vs Gardner-Webb). Our own totals model (49.77) sits almost exactly on the current market (49.7), essentially no edge by the model's own math, but the model also doesn't know about this week's real total move or Purdie's accuracy issue. Given the market has already corrected sharply toward the under and the underlying scoring data supports it, I'll side with Under.

Where our number looks wrong.Our rating still treats Ball State as roughly a full field-goal underdog on a neutral field, which the market's firm -14.5 and the outside [figure removed] both reject. The model has no way to see that Liberty's offense, not Ball State's defense, looks like the bigger question mark right now given Purdie's 35% completion clip and the team's low combined scoring totals, exactly the kind of skill-position signal a score-margin-only rating structurally can't use this early in a season.