So thanks to my FO compadre, Brian Fremeau, and his FEI projections, I figured out a way to create at least a starting point for my own projections. I basically applied his projected increases and decreases to my own numbers, then added a few more variables. One variable wasn't uncovered until last week, which means that all of those projections I was doing over the summer have now changed slightly. Basically I discovered that Phil Steele's whole "Turnovers = Turnaround" thing was even more prescient than previously imagined.
For those out of the Steele loop, his theory is that the teams with particularly good or bad turnover margins in a given year will, all else being equal, see a regression to the mean the next year. Well, if you remember, a while back I created a Turnover Points Margin figure. Basically, since I have point values for every yard line, I can value turnovers by looking at where the team was when they turned the ball over, and where the other team ended up after the turnover. That's much more accurate than just putting a +1 or -1 in the turnover column, so in theory it should give us an even better view of the year-to-year impact of turnovers.
The results were staggering. Teams that had a turnover points margin of, say, +5.0/game in 2007 fell an average of almost exactly 5.0 points the next. Same with +/- 2.0 points or +/- 8.0 points. Whatever the margin, the regression to mean was amazingly close. The only teams that didn't necessarily apply were the teams in the ranked about #1-20 and #101-120. The top-ranked teams tended to stay about three points above water (meaning if they were +8.0/game one year, they only fell to about +3.0/game the next instead of 0.0/game), while the bottom-ranked teams tended to stay about three points below water. It was an amazing thing to find.
So what does that mean? Well...using 2008 turnover points margins, we can apply a certain per-game point deduction or addition to every team's projected S&P+ ranking, and you've got a (hopefully) more accurate look at projected rankings.
Tomorrow we're going to look at projected results, records, bowls, etc. For now, let's just look at the 2009 projected Top 40 using my projected S&P+ rankings.
- Ohio State
- Penn State
- Oklahoma State
- Boise State
- Oregon State
- Texas Tech
- Ole Miss
- Florida State
- Boston College
- South Florida
- Georgia Tech
- Virginia Tech
- Southern Miss
- Michigan State
Other Big 12 teams: Baylor (#62), Kansas State (#64), Colorado (#66), Texas A&M (#72), Iowa State (#87).
Other Mizzou opponents: Illinois (#42), Bowling Green (#86).
Thoughts after the jump.
- The S&P+ numbers really did the Big 12 some favors last year because of the strength of schedule adjustment. They faced great offenses, therefore their defenses didn't take too hard a hit. Missouri finished 10th and Kansas 20th, so that's how you have Missouri slipping slightly and KU moving up slightly, and that's how you have seven Big 12 schools in the top 22. I actually think the conference will slip more than that this year, but I tried to minimize the manual adjustments here. This is basically what the numbers say, not what I say.
- How is Mizzou 14th? First of all, most teams at the top of the S&P+ list last year benefited tremendously from turnovers. Missouri did not. Oklahoma was +9.4/game, Florida +7.8, TCU +6.9, Utah +6.5, Alabama +5.7, Ohio State +4.6, Penn State +4.0, USC +3.8. Missouri was +0.3, so when the turnover adjustment was made, everybody else slipped while Missouri pretty much stayed in place.
The other thing that helped Missouri was that one of the adjustments was made using returning starters figures. It's true that Missouri doesn't return a ton of starters, and their numbers were hurt because of it, but there's no "magnitude of starters lost" penalty, if that makes sense. Instead of "Mizzou lost an All-American WR and TE," it's just "Mizzou lost a starting WR and TE," that sort of thing. I actually penalized them a smidge more because of the magnitude of the loss, but there they are, still at #14. That's higher than I personally would rank them, but hey...not complaining at the thought of it...
- There's also not a "starting a true freshman" penalty, at least not yet, so that's how USC is at #2. I think with the naming of Matt Barkley as a starter, USC will probably finish 10-2 or so and out of the Top 2-3, but it's not like they'll finish too far from the projected #2 ranking, no matter who is QB.
- I think Georgia could be in for a better-than-expected season. It will basically depend on five games: Oklahoma State (projected: lose by 0.7), Arkansas (win by 1.0), LSU (win by 1.9), Tennessee (win by 5.3) and Florida (lose by 10.0). If they get through those games at 3-2, then ten wins should be in the cards. If they beat OSU, however, then suddenly the Former World's Largest Outdoor Cocktail Party (what are they calling it now?) goes from huge to gigantic.
- Oh yeah, and Arkansas. The numbers like them as a sleeper candidate, but their schedule is tough enough that 8-4 or 9-3 might be their ceiling. The Georgia, Alabama, and Ole Miss games are up in the air, but otherwise they're set with mostly likely wins (ATM, Auburn, Eastern Michigan, South Carolina, Troy, Mississippi State) and likely losses (Florida, LSU). But if they go 2-1 against Georgia, 'Bama, and Ole Miss, they'll have at least a chance at the SEC West title...though they'd probably need to go 6-2 to have a great shot.
- We'll get into predictions full-force tomorrow, but right now there appear to be five projected undefeated teams: Florida, Texas, Penn State, TCU, and Boise State. I doubt Boise goes undefeated because I think Oregon will beat them on Saturday, but TCU's got a chance. Basically it boils down to three games: Virginia, Clemson and BYU...with Clemson probably being the single toughest. Really, the other thing TCU has to fight right now is...well, illness. They've had some players go down with swine flu, and if that spreads, all bets are clearly off.
How is TCU that high to begin with? Because they finished 4th in last year's S&P+ rankings on the heels of a) two really good losses (at OU, at Utah) and b) by far the nation's best defense. The D will probably slip a smidge this year, but they were so far ahead of the pack on that one, that they'll only slip so far.
- Where's Notre Dame, you ask? 53rd. Sorry, Lou Holtz and mcboomofdoom. I think they'll probably be a bit better than that, and they'll certainly have a chance at double-digit wins because of their schedule. Right now, they face TEN games projected within nine points. Only USC (19.8-point loss) and Washington State (13.9-point win) are gimmes. If they're a touchdown better than expected, they could be looking at 11-1. A touchdown worse? 3-9.
- Right now, 0.08 points separate Oklahoma and Texas. Yikes.
Alright, that's enough spoilers. Full projections tomorrow.