2018 college football upset spotting, week two

A quick recap:

• The system expected about 3.6 upsets last week and got 4, or slightly above expectations, which is good.
• It expected two upsets of a Top 25 team beating another Top 25 team; we got three.
• Likewise, it expected two double-digit favorites getting upset; we got one (Maryland over Texas) and very nearly got two more in Appalachian State-Penn State and Utah State-Michigan State. So close!

Here’s this week’s numbers:

Observations:

• It’s a pretty boring week of football. The best game involving a Top Ten team is either Clemson going on the road to play a coach in his second game at a new school or Georgia going on the road and beating Will Muschamp by 21. There’s some spots for interesting games and/or upsets, but they’re further down the charts. However…
• Note how there’s a 50/50 shot a Top Five team loses Saturday. It’s the least likely group to lose a game, but it’s still interesting, and it’s all pinned on one of Clemson or Georgia losing. Don’t bet your life savings on it, but it sure seems like one of those games will go down to the wire.
• To find an upset, look at teams ranked 6-15. The best bets here are pretty obvious: #15 Michigan State on the road at Arizona State and #10 Stanford playing host to #17 USC. (Also noteworthy is Penn State as just a nine-point favorite at Pitt after a poor opening performance.) One of these three teams will lose; it could either be a great day for the Pac-12 or for the Big Ten and no one knows.
• Also, the bottom 40% of the poll is more likely to lose a game than not. That’s pretty surprising, given that all but two of the eight games feature teams as 20 point favorites or higher. The most vulnerable teams here are from the SEC: #18 Mississippi State playing a noon game at Kansas State and #25 Florida hosting Kentucky, who they haven’t lost to in my lifetime. Obviously, I lean much more towards MSU losing than Florida, but both are close to equally possible. Maybe keep half an eye on #16 TCU at SMU, too. Plus…
• We should get one loss from a double-digit favorite. Only three are truly vulnerable this week – #2 Clemson at Texas A&M, #3 Georgia at South Carolina, and #25 Florida vs. Kentucky. For 20+ point favorites and teams facing FCS squads, they might be in trouble on an Everything Goes Wrong day: #6 Oklahoma hosting UCLA, #14 West Virginia hosting FCS stalwart Youngstown State, #16 TCU at SMU, or #21 Michigan hosting Western Michigan.

So, to sum it up: it’s a bad week of football, but you’ll get two upsets, including one shocker. I went 0-3 on my personal picks last week, so here’s a shot at 0-5: #17 USC over #10 Stanford and Kansas State over #18 Mississippi State.

2018 college football upset spotting, week one

Same deal as last year. Notes below:

• The top five looks pretty safe, but at least one and likely two Top 6-15 favorites are looking at an unexpected loss. Somehow, Auburn is a Vegas favorite over Washington, but Washington is the computer favorite. That wouldn’t count as a loss, so we’ve got to find another one. Who’s it going to be? Notre Dame to Michigan? Miami to LSU? Alternately…
• There’s just a 32.6% chance that 80-99% teams finish this week unscathed. The most vulnerable of these is Stanford, playing at home against a 10-win San Diego State team that always plays excellent defense. There’s some other interesting games out there, though – Oklahoma, in two of the five systems, is within one standard deviation (16 points) of a loss. Alabama is so in four of the five, but it’s Alabama. Penn State is within a standard deviation in one metric, as are defending national champions UCF. It feels pretty likely that someone is going to be dealt a stunner of a loss this weekend; my wild guess is Stanford.
• Double-digit favorites are expected to lose two games and have just a 10.7% chance of an undefeated weekend. One of those will be covered by the previous bullet, but the other one will have to come from the 60-79% range. You’ve got three options here: West Virginia losing to Tennessee (the most likely of the group), Boise losing at Troy, or Texas losing on a neutral field to Maryland. The last of these did happen last season, while Troy was within four points of Boise with three minutes left last season. Tennessee is a massive unknown, but computers are skeptical of a West Virginia improvement on defense. I’d rank them in likelihood of happening exactly as presented: Tennessee > Troy > Maryland.
• In coin-flip games, anything can happen. It’s strongly unlikely that all three of Washington, Miami, and Notre Dame win. However, two of those three should. If we count Auburn in this group instead, the odds rise of them losing two of the three. Michigan seems much more well-prepared to knock off their higher-rated opponent than LSU is.
• Basically: expect some shake-ups, though probably not at the top. The top five seems unlikely to change after Week One, barring a stunner or the Washington/Auburn winner moving in.
• Removing the 90-99% and top five games from the group, we’re most likely to have about three upsets. This is a shot in the dark, but give me San Diego State over Stanford, Tennessee over West Virginia, and Michigan over Notre Dame. Why not!