Analysis 4thand4cast
Pick 'Em Week 10: How the Pool Shook Out
The Slate That Wouldn't Cooperate
Florida opened at 59% to beat Oklahoma and lost 7-23. Utah was a 65% favorite over BYU and won 28-20. Arizona State, favored at 77% against Colorado, got 20-10. On the surface, a slate where favorites mostly held. Underneath: a week that punished confidence and rewarded the margin traders.
GIE finished 4-for-10, capturing 30 points in a slate that offered clean separations. The model's board had the Utah-BYU and Arizona State-Colorado calls correct, along with Clemson's 18-13 road win at Syracuse (35% pregame) and Iowa's 24-10 victory at Northwestern (66% pregame). Four wins in ten is below the model's seasonal pace, a reminder that even structured probability generates variance week to week.
The field, notably, produced no entries this week. A 0-0 slate from human competition left GIE alone atop the board—a thin lead, but a lead.
The Traps That Bit Back
Northwestern's 66% pregame win probability looked safe on paper. Iowa came in as a measured underdog, and the home team held that designation through kickoff. The Hawkeyes won 24-10, a 14-point road win that classified as an upset by the pool's 40%-or-lower threshold. It wasn't a collapse—it was a miscalibration of how Iowa's offense would function in a November road spot, or Northwestern's defense would hold up. GIE read it correctly. The field did not test it.
The real damage came at Florida-Oklahoma. The Gators' 59% pregame edge evaporated in a 16-point loss. That's not variance. That's a fundamental gap between pregame assessment and on-field execution. Oklahoma's 23 points against a Florida team projected to win more often than not suggests either the model underweighted Oklahoma's offensive capability or overweighted Florida's defensive profile coming into the week. Either way, it's the kind of game that costs confidence pools points but teaches them something about recalibration.
The Stunner Nobody Saw Coming
Florida Atlantic, a 31% underdog at home against UTSA, won 60-29. A 31-point margin in a contest the model gave FAU a three-in-ten chance at. That's not an upset within the pool's framework—UTSA was favored—but it's a demonstration of how lopsided a result can be when the underdog breaks through. FAU's 60-point performance wasn't close. This was one of those rare games where the worse team on paper turned in a dominant offensive display and the pregame model missed badly in both direction and magnitude.
Tulsa's 60-10 dismantling of Tulane, by contrast, moved with the pregame line. Tulsa was favored, and the margin—exactly as predicted by a 47% call—affirmed the board rather than challenged it.
What Carries Forward
GIE leads the week and the pool sits reset. The model's 4-for-10 isn't a concern in isolation; the concern is whether pregame calibration drifted this week or whether this was simple noise. Next week will clarify.