I came up with this method for projecting game scores during last year's bowl season. It worked reasonably well, so I'm giving it another shot during the season. I have a feeling it's not going to be terribly accurate this early since some teams have played only three I-A teams so far, but it should get better as the season goes on.
Anyway, I'm going to use Ole Miss as an example of how it works, but I have done this for all the SEC teams. I basically find what each team has done relative to its opposition.
Ole Miss is gaining an average of 26 points a game and is allowing 12.3 a game against I-A competition. To find out the cumulative opposition stats for points for and allowed, I just combined what they all do together:
|Opponent||Points For||Points All.||Games|
The averages are what I'm looking for here (and all I-AA results have been removed for all teams, non-SEC opponents included). Now, I can compare what effect Ole Miss has had on an absolute and a percentage basis to its opponents:
The Rebels score 2.42 more points a game than their opponents allow on average (a 10.25% increase) and they allow 5.25 fewer points per game than their opponents score on average (a 29.86% decrease). Once this process has been performed for all teams, I can project the final score of SEC matchups on either the absolute basis or the percentage basis. Intuitively, I expected the percentage basis to be more accurate but in practice, the absolute basis was far more accurate during last bowl season. I'm including both here for the sake of being thorough, but I put more stock in the absolute numbers.
ALABAMA at OLE MISS
Since I used Ole Miss as the example team, I am going to start with this matchup. I have a feeling this projection won't be overly accurate since Ole Miss has had just three I-A opponents so far, but here goes.
To make this table easier to read, I put the numbers that pertain to Ole Miss in bold. What this table does is actually predict two final scores for the game, one based on the offenses and one based on the defenses, and then it averages the two together for the final projection.
Ole Miss averages 26 points a game, Alabama holds its opponents to 11.85 points under their average, so the first score for Ole Miss comes out to 14.15. Alabama's defense is allowing 14.4 a game, Ole Miss scores 2.42 above its opponents' average, so the second score for Ole Miss comes out to 16.82. The final score prediction for Ole Miss is 15.48. The same process goes for Alabama, who gets a 28.87 prediction.
Now, imagine this table but with the percentage effects instead of the absolute effects. That's how I get the percentage prediction.
ABSOLUTE PROJECTION: Alabama 29 - Ole Miss 16.
PERCENTAGE PROJECTION: Alabama 21 - Ole Miss 17.
FLORIDA at LSU
This is another one where a small sample size causes me concern. Florida, like Ole Miss, has played just three I-A opponents so far. In addition, I'm expecting to see John Brantley play for the Gators, which makes the offense a lot different from what it has been so far. Ultimately, I expect the UF defense to be the difference in the game, and that comes through in the projections.
ABSOLUTE PROJECTION: Florida 32 - LSU 9.
PERCENTAGE PROJECTION: Florida 22 - LSU 11.
AUBURN at ARKANSAS
This one has all the hallmarks of a shootout. Arkansas has very little defense to speak of, and Auburn's defense hasn't been quite as crushing as we might have expected based on last year. If the projections are accurate at all, we'll see some pinball-like numbers.
ABSOLUTE PROJECTION: Auburn 46 - Arkansas 32.
PERCENTAGE PROJECTION: Auburn 44 - Arkansas 34.
GEORGIA at TENNESSEE
At first, I was surprised to see Tennessee favored over Georgia. The more I think about it though, the more it makes sense. The Vols will be at home, for one, and the Tennessee defense has been one of the few consistently great units on any team in the conference. If I had to put money on something in this game, I think putting it on Monte Kiffin and Eric Berry would be the safest bet. The projections seem to indicate that not only will UT's defense will be a big difference in this game, but also that the offense will have a good time as well.
ABSOLUTE PROJECTION: Tennessee 31 - Georgia 21.
PERCENTAGE PROJECTION: Tennessee 27 - Georgia 25.
KENTUCKY at SOUTH CAROLINA
I think we've touched on this once this week somewhere along the line, but Kentucky has not defeated South Carolina since the first digit of the year rolled over from a 1 to a 2. Managing to take down the Gamecocks would be an encouraging sign for the Wildcats after two weeks of being rolled by the conference heavyweights. For whatever it's worth, the projections don't like their chances.
ABSOLUTE PROJECTION: South Carolina 25 - Kentucky 18.
PERCENTAGE PROJECTION: South Carolina 22 - Kentucky 21.