October 2026 · Preview

Zero Weather, Maximum Confusion

We're a football weather site. So it's a little embarrassing that the game our model disagrees with the sportsbooks about most this week is being played under a roof. North Dakota State at UNLV, Saturday in Las Vegas's Allegiant Stadium: no rain, no wind, no snow, a weather adjustment of exactly 0.0. The sportsbooks set the over/under at 47.5. Our model says 61.5. That 14.0-point gap is the biggest of all 58 games on our board. So what's going on in there?

The short version: the model has no idea who North Dakota State is. The Bison are brand new to the top level of college football this year, and our model handles brand-new teams by lumping them in with every lower-division team in the country. So it thinks the stingiest defense in FBS is a generic small-school defense. History says the model runs way too high in exactly this spot. Our call: trust the sportsbooks on this one, and lean under.

Our model vs. the posted total61.5 vs. 47.5

Weather adjustment0.0

Points NDSU allows per game (fewest in FBS)9.8

Overs when the model was 8+ above the line with a first-year team43.2%

The sleek black, fully enclosed Allegiant Stadium under a clear blue sky in Las Vegas, with a delivery truck driving past in front. — Allegiant Stadium, UNLV's home, roof and all. Forecast inside for Saturday: perfect, same as always. Photo: Tomás Del Coro, CC BY-SA 2.0, via Wikimedia Commons.

Meet the new kids

North Dakota State is playing its first season as an FBS team, in the Mountain West. So far it's going great: 5–0, scoring 33.6 points a game and allowing just 9.8, the fewest of all 138 FBS teams. UNLV is 3–2, scoring 25.0 and allowing 23.6. And here's a fun one: both teams are 1–4 against the over/under this year. Their games keep landing under the line, by 5.5 points on average for NDSU and 6.1 for UNLV.

North Dakota State

Wk	Game	Total	Line	+/−
1	vs. Jacksonville State, 33–7	40	46.5	-6.5
1	vs. Fordham, 38–0	38	54.5	-16.5
2	at Air Force, 38–32	70	48.5	+21.5
3	at Sacramento State, 31–10	41	50.5	-9.5
5	vs. Wyoming, 28–0	28	44.5	-16.5

UNLV

Wk	Game	Total	Line	+/−
1	vs. Memphis, 21–27	48	55.5	-7.5
1	at Hawai’i, 21–6	27	55.5	-28.5
2	at North Texas, 6–44	50	58.5	-8.5
4	at Akron, 38–10	48	52.5	-4.5
5	vs. California, 39–31	70	51.5	+18.5

Fun fact for a weather site: three of NDSU's five games have been indoors. The Bison play their home games in the Fargodome, which makes sense for a team from a place where November can be brutal. Saturday is a dome team visiting a dome team.

So why does our model think this is a shootout?

Our model builds a rating for every team's offense and defense. It starts each team from last season's rating, then updates as this season's games come in. Simple enough. But a team that wasn't in FBS last season doesn't have a rating from last season.

The “everybody else” rating

To keep things simple, our model gives every team without a rating one shared rating: every FCS and lower-division team that plays an FBS opponent all at once, like one giant imaginary team. That imaginary team is pretty bad, because it's built from games like Tennessee State's trip to Georgia this year (a 63–3 loss). So the model thinks North Dakota State plays defense like the average small school visiting a big one.

That's the most generous assumption possible about the team that gives up the fewest points in the country. The model figures UNLV will score a pile against that “average small-school defense,” and its total balloons to 61.5. The sportsbooks have actually watched NDSU play.

Want proof it's the newcomer thing? The second-biggest gap on this week's board is Sacramento State at Bowling Green, at 12.3 points. Sacramento State is the other brand-new FBS team this year.

Inside a domed stadium with blue seats and a green field with a yellow end zone reading BISON, as North Dakota State players warm up before a game. — North Dakota State warms up in the Fargodome before a 2019 game against North Dakota. The Bison are used to football with a roof on. Photo: SS2027, CC BY-SA 4.0, via Wikimedia Commons.

Has this happened before?

Plenty. We re-ran our model on every season from 2010 to 2025, each season predicted using only the seasons before it, and pulled out the games with a team that had no FBS rating from the year before. That's mostly brand-new FBS programs: App State, Georgia Southern, Charlotte, Liberty, James Madison, Jacksonville State, Sam Houston, Kennesaw State, Delaware, Missouri State and more.

Games with a posted total, 2017–2025	Games	Model vs. line, avg.	Went over
Everyone else	7,334	+0.76	49.5%
A first-year team involved	127	+4.73	44.9%
… and the model 8+ above the line	37	+12.39	43.2%

With a newcomer on the field, the model sits way above the line, and the games go under more often than not. When the gap was as big as this week's, the games came in 11.15 points below the model on average. The reason is the same one we found above: the “everybody else” rating expected those new teams to allow 35.9 points and score 21.0. They actually allowed 30.6 and scored 26.3. New FBS teams are a lot better than the average small school. Who knew?

Domes don't help the model either

Since our whole thing is weather, we had to check. In dome games with a posted total, the game went over 44.2% of the time (328 games), a bit under the usual coin flip. And when our model was 8 or more points above the line in a dome, only 6 of 18 went over. Tiny sample, but it points the same way.

One more twist

This week we started testing a tweak that shades a prediction when a team's games have been beating, or missing, the posted total (here's how that started). Both of these teams keep going under, so the tweak shades this game up by 1.4 points, to 62.9. It's betting their unders will cool off. On this one, we're not so sure.

Our call

Trust the people who've actually watched North Dakota State. The sportsbooks' 47.5, with NDSU a 3-point favorite, works out to roughly North Dakota State 25, UNLV 22. That's our best guess too, with a lean under, since both teams keep coming in under their totals. The model's 61.5 is what you get when it has no idea who one of the teams is.

The good news: it's a fixable blind spot. A brand-new FBS team could start from a rating built on its FCS seasons instead of sharing the “everybody else” rating. That's going on our list. (And the usual reminder: betting is for adults, and the house usually wins. We're just keeping score.)

Data: the model's prediction and the DraftKings total as shown on our dashboard Friday afternoon, October 9 (both update until kickoff), and this season's games, lines and team stats through Week 5. History: every game from 2010 to 2025, domes included, from CollegeFootballData.com, with the model trained only on earlier seasons' outdoor games and dome games predicted with no weather, the way the live model works. Lines start in 2017. “First-year team” means a team with no FBS SP+ rating from the season before; a few are programs returning after a gap. The full analysis is in ndsu_unlv_preview_2026.ipynb in the project.