First, oh great and wonderful college football fans, let’s be clear. There were some great games yesterday. Teams that should have taken care of business did across all conferences and there were some close games between really good teams resulting in heartbreak for the defeated. But, that’s not what I want to talk about tonight. We will explore some odd losses and other closer than they should have been games and why that matters. It also seems like the right time to remind longer term readers and show new readers how the logic of the GCR rankings work – what they consider and what they don’t. But we will start with the 14 games the GCR picked as upsets (meaning the GCR picked a different winner than the Vegas line).
…..
Overall, the upset record was 6 correct and 8 incorrect for a paltry, embarrassing, and horrible .429 pct. I will be talking to the programmer (ok that’s me) to get those numbers up. I missed Virginia Tech (West Virginia), UMASS (E Michigan), Bryant (Akron), Arkansas St (Washington), Charlotte (Georgia St), Southern Miss (Troy), Virginia (UNC), and Jackson St (ULM). As we will see in G’s Expectation later this week, the prediction model is better for the FBS than the FCS right now (by over 120 percentage points) so Bryant and Jackson St probably wouldn’t be picked if those games took place later in the year. UMASS might fall into that category as well if they perform like previous years. But, I picked them and I have to live with it. I did pick Michigan St to take down Miami FL, Kansas St over Nevada, San Diego St to upset Utah, Utah St to ground Air Force, Oklahoma St to give the blue meanies to Boise St, and BYU to drop Arizona St. I’ll keep track of how the GCR performs and expect the ratio to get better.
…..
So, how does the GCR work? First, I only track 4 things: what sub-league a team is in, who a team plays, where that game is played (home, away, neutral), and points scored. The schedule and location is important because a road win has a bit of a bonus and a home loss hurts a little more than a road loss. Road games (and to a lesser extent neutral games) add a bonus to the strength of schedule as well. Playing up a sub-league (FCS playing Group of 5 or Power 5, or Group of 5 playing Power 5) generates a bonus which is subtracted (essentially) from the team playing down (includes FCS playing Division II or lower). Strength of Schedule is essentially the adjusted number of wins your opponents average. But the SOS is not impacted by future games until they are played. So SOS changes each week even if a team is idle. All games that have been played are constantly adjusting because the formulae driving the GCR see every game that has been played as being played “now”. It doesn’t know what a team was ranked in week 2, it looks at the week 2 game as happening at the same time of all other games. I hope that makes sense. The performance score is based on decision (win/loss) for each team (the winning team gets a larger base adjusted by opponent league and score differential). The score differential is based on a sliding scale – winning by 10 gives a bigger bonus than winning by 3, but winning by 30 isn’t substantially more valuable than winning by 20 – in fact, once the differential hits 21, the increase is almost nil. The average performance score for a team doesn’t change if a team is idle. The total score is a mix of the two scores with performance slightly outweighing SOS. The total score is what the ranking is based on. I don’t keep track of turnovers, 3rd down conversions, injuries, coaching changes, etc. First, I don’t have access to the data and don’t have time to calculate it all, but more importantly, a team could have 5 turnovers and fail every 3rd down conversion and win 14-10 – points are the only stat that really matters.
…..
Predictions are based on a number of factors including sub-league, home field advantage (based on hundreds of games over the last few years), and performance score. Obviously, I expect the GCR to get better as the season progresses and it has more data (that said, I was 82-25 last week which isn’t horrible). Probably around week 6, I’ll start talking about predicted records – I already am tracking them, but there aren’t yet enough games for nuances. From that, I can predict the division/conference winners based on upcoming schedules.
…..
This week had a plethora of unexpected or overly close or just weird situations.
- Both Virginia Tech (47) and Miami FL (72) lost games while AP ranked – the 4th and 5th ACC teams to do that already this season and it leaves just Boston College (22) and Wake Forest (20) as the only undefeated teams in the conference – and they are in the same division so a maximum of 1 by year’s end
- Staying in the ACC, Pittsburgh (69) lost to W Michigan (36) of the MAC a conference with a combined 13-23 record and Clemson (65) held on for a 14-8 win over Georgia Tech (165) – the game was in Death Valley east
- The Big 10 had a near scare as Ohio St (38) was facing a pesky winless Tulsa (147) team and was leading by a score of 20-14 with 1:00 to go in the 3rd. They prevailed 41-20 but their defense is very suspect after giving up 31 to Minnesota (68) and 35 in a loss to Oregon (3)
- The Big 10 has 6 undefeated teams but other than Iowa (4) they are all in the East – Maryland (12), Michigan (16), Michigan St (11), Penn St (6), and Rutgers (7) – max of 2 going into championship week
- The Big 12 pretty much took care of business except for Oklahoma (19) who couldn’t put Nebraska (114) away and won 23-16. Keep in mind Nebraska lost to Illinois (117) by just 8 and they followed the win with losses to UTSA (25), Virginia (70), and Maryland (12)
- Joining Oklahoma in perfection in the Big 12 is Baylor (15), Kansas St (5), Oklahoma St (23), TCU (32), and Texas Tech (18) – they all play each other so only one can make it to the end.
- The Pac 12 is just struggling. Just this week, Colorado (141) was shutout at home by Minnesota, Utah (122) lost to Group of 5 San Diego St (24), Arizona (200) lost to FCS N Arizona (204), Arizona St (76) fell to Group of 5 BYU (17), and UCLA (57) fell to Group of 5 Fresno St (37).
- That means that on September 19, there was just a single perfect team in the Pac 12 – Oregon (3).
- In the SEC, Mississippi St (40) lost to Group of 5 Memphis (26) and Kentucky (13) barely beat FCS Chattanooga (166) 28-23. Florida (42) was a missed extra point from forcing OT vs Alabama (2)
- The SEC joins the Big 10 and the Big 12 with 6 perfect teams – Georgia (1), Kentucky (13), Alabama (2), Arkansas (14), Mississippi (8), Texas A&M (10)
- Notre Dame (9) is the only Power 5 independent.
- As of right now, 22 Power 5 teams have a goose egg in the lost column with 18 (82%) of them in the Big 10, Big 12, and the SEC.
…..
What will next week bring? Who knows? But it will be worth watching. Tune in later this week for more predictions and maybe some ranting about some topic (suggestions are always welcome). Thanks for reading. If you don’t like it, let me know. If you do, please share with others and encourage comments. Until next time, G
