Diving into the advanced stats can be a great way to both be more informed about college football and spot some potential betting trends and value before they fully emerge. Better yet, it can make you feel more knowledgeable about the college football landscape as a whole. Let's dive into the sea of advanced stats and see what we can find.
Let's talk about volatility
My SP+ ratings have been doing unbelievably well against the spread this season. They're hitting 58.4% against the spread for the season and 64% the past two weeks. There has to be an inevitable crash coming soon, but it has been a nice run. It appears that you can get a pretty good lay of the (betting) land by simply looking at those ratings.
That said, a single rating -- presented in an adjusted points per game figure (so you can subtract one team's rating from another to find out the projected scoring margin if those teams were to play each other) -- doesn't tell you all you need to know. Among other things, it tells you nothing about a given team's volatility level.
Not that we have a reliable sample just yet, but I thought it would be interesting to take a look at which teams have been pretty consistent when it comes to performance against the spread and which have been all over the map. A good place to start in this regard: good old standard deviations.
I looked at the difference between the spread and the final scoring margin for each game, then looked at each team's standard deviation. The following teams have almost never deviated in their performance against the spread so far this season:
1. Oregon State Beavers (2.8-point standard deviation)
2. Wake Forest Demon Deacons (3.0)
3. Oklahoma State Cowboys (3.7)
4. SMU Mustangs (4.3)
5. Vanderbilt Commodores (4.6)
6. UTEP Miners (5.1)
7. Texas Longhorns (5.2)
8. Fresno State Bulldogs (5.3)
9. Alabama Crimson Tide (5.7)
10. Louisville Cardinals (5.8)
This doesn't necessarily mean these teams have been performing close to the spread. SMU, for example, has routinely been about two touchdowns ahead of the spread this season (the Mustangs have overachieved vs. the spread by 12.5, 9.5, 18.5, 19.5 and 13 points so far), and Vanderbilt has steadily been about 4.5 points behind it. They've been sure things in their own ways.
Then there are the less-than-sure things: teams you might want to avoid. Here are the teams that were spectacularly unreliable and unpredictable in September:
1. Maryland Terrapins (46.8-point standard deviation)
2. Georgia State Panthers (36.0)
3. Nevada Wolf Pack (33.4)
4. Penn State Nittany Lions (33.2)
5. Kansas Jayhawks (29.6)
6. Hawai'i Rainbow Warriors (28.8)
7. UL Monroe Warhawks (26.8)
8. Oregon Ducks (26.7)
9. Boston College Eagles (26.3)
10. Western Michigan Broncos (25.4)
In three games against FBS opponents, Maryland has overachieved by 41 points against the spread and underachieved by 10 and 52.5 points. That is absolutely ridiculous. If the Penn State game is any indication, the Terps might be on their way to becoming more reliable -- in all the wrong ways.
Nevada is another particularly befuddling team. The Wolf Pack have covered in two of four FBS games, but when they didn't, they underachieved by a staggering 47 and 53 points. Their performances against Oregon and Hawaii were maybe the two most disappointing performances of the season.
Gambling relevance: It is indeed early, so maybe it doesn't mean much, but if it were my money, I'd avoid betting on games involving those super-volatile teams for a while, until we figure out exactly what these teams are.
SMU and Tulsa are both overachieving (even if you've noticed only the former)
SMU is one of the coolest stories of the early season. Sonny Dykes' Mustangs are ranked for the first time since they received the NCAA's death penalty in the late 1980s. They are 5-0, and as we saw above, Vegas has not yet caught up to how good they look.
Texas transfer Shane Buechele is completing 69% of his passes, he has a trio of super-efficient receivers (James Proche, Reggie Roberson Jr. and tight end Kylen Granson), senior running back Xavier Jones is a rock, and the defense both forces third-and-longs and gets after the passer. Hopefully you bet the SMU win total in the preseason, and hopefully you bet the over. It was six; now SP+ projects the Mustangs to finish with an average of 9.4 wins.
I'm curious whether the Mustangs' run of overachievement will continue this week, though, as they play an underrated Tulsa Golden Hurricane squad. Philip Montgomery's Golden Hurricane have also overachieved this season, but you're forgiven if you haven't noticed -- they don't have a gaudy 5-0 record to show for it. Instead, though they're up to 76th in SP+ (they started the year 99th), losses to top-25 Michigan State and Oklahoma State have rendered them 2-2. Because of a brutal AAC schedule -- the AAC West is as good as it has ever been, and they draw UCF and Cincinnati from the East -- SP+ still gives them only a 31% chance to reach bowl eligibility.
Gambling relevance: They are indeed solid, though, and while the line for this game has been hovering around SMU -14, SP+ projects the Mustangs as only 12-point favorites. If they manage to once again beat the spread by a couple of touchdowns in this one, it might be a sign that they're ready to make a run at the Group of 5's New Year's Six bowl.

West Virginia is a three-win team in record only
For each FBS game, I create a measure I call postgame win expectancy. The idea is to take the predictive factors that go into my SP+ ratings and say, "Based on this game's key stats, you could have expected to win this game X percent of the time." Wins are great, but not all wins are predictive, and this is one way to figure out whose record is a bit on the inflated or deflated side.
The West Virginia Mountaineers: inflated. I guess I can't tell you to sell high on a team that isn't all that high, but I can tell you the Mountaineers have had a postgame win expectancy under 20% in three of four games, including two of three wins.
It was 17% against James Madison because of a massive success rate disadvantage (JMU 49%, WVU 32%) and some turnovers luck.
It was just 9% in Week 4's 29-24 win over Kansas. WVU made no big plays, again enjoyed some turnovers luck and got outperformed massively when points were on the line (points per scoring opportunity: KU 6.0, WVU 3.3).
I had WVU -4 as one of my Week 4 Best Bets, and I got as lucky as Neal Brown's team did. SP+, which is designed to look at the predictive components of each game and not the result itself, reacted to the game by dropping WVU's rating and raising KU's. The Mountaineers are now 64th in SP+ and projected underdogs in every remaining game.
Gambling relevance: I'm betting that the 3-1 record, combined with the out-of-nowhere easy win over NC State, will distract some from just how shaky this team has been. Don't be fooled. SP+ projects the Mountaineers as at least 10.5-point underdogs in seven of eight remaining games. It has Texas by 14.1 in Week 6, ISU by 12 in Week 7 and Oklahoma by 29.8 in Week 8. The Texas-WVU line has been around UT -11 or -12 so far, but unless the Longhorns go back to their "play as poorly as possible when you're a solid favorite" ways, they have more of an advantage than that.

Colorado a potential overs machine
Colorado Buffaloes quarterback Steven Montez is on pace for nearly 3,500 yards and 30 touchdowns throwing to the foursome of Laviska Shenault Jr., K.D. Nixon, Tony Brown and Dimitri Stanley. The Buffaloes have scored at least 34 points in three of four games and were held under that in only a low-tempo game against Air Force. They are seventh in offensive SP+, and it has powered a 3-1 start.
They have also given up at least 30 points in every game. They rank 123rd in success rate allowed and 113th in defensive SP+. Only one remaining Colorado game -- at Oregon in Week 7 -- is projected to feature fewer than 67 total points.
Gambling relevance: SP+ projects this week's game against Arizona to feature 75 points, and though that doesn't account for the fact that Arizona's Khalil Tate's status is uncertain because of a hamstring injury, there's quite a cushion between 75 points and an over/under that has thus far been around 64. This will be a test of CU's ability to suck anyone into a shootout, but don't underestimate the Buffs in that regard.

Georgia Tech a potential unders machine
The team Colorado shared the 1990 national title with is going about things a little bit differently in its head coach's debut season. Geoff Collins' first Georgia Tech Yellow Jackets squad is currently 120th in offensive SP+ and 41st in defensive SP+. The Yellow Jackets' offensive rating has fallen by 16.4 adjusted points per game this season, and the defense has improved by 11 points. In their three games since losing 52-14 to Clemson, the Jackets' three games have averaged 33.7 points combined.
Gambling relevance: Seven of Tech's eight remaining games are projected to feature fewer than 50 points. This week's game against UNC is the exception, and SP+ has that one at 51.2. It seems this week's over/under is going to be pretty close to 51, but the projected totals are going to fall from here. Keep that in mind.

It's a defining week for Northern Illinois
It appears that SP+ has given up on another team with a new head coach: Thomas Hammock and the Northern Illinois Huskies. Following a loss to a disappointing Vanderbilt squad (which followed blowout losses to Utah and Nebraska), it dropped them to 114th overall, 10 spots lower than this week's opponent, Ball State.
The Huskies are MAC heavyweights and the defending conference champions, and I don't think the lines have caught up to how bad they might actually be. They are around a 4- or 4.5-point favorite against BSU, but SP+ projects them to lose by an average of 1.4.
Gambling relevance: I'm not sure I'll make this one of this week's Best Bets, but we're going to call this a pretty defining week for the Huskies. They have talented and interesting players, from linebackers Antonio Jones-Davis and Jordan Cole to cornerback Jalen McKie to all-or-nothing receiver Spencer Tears to former Cal quarterback Ross Bowers. But if they fall at home to the Cardinals, in advance of road trips to Ohio and Miami (Ohio), this season could snowball in DeKalb, and you might find some value in realizing it's happening before the lines do.