I love reading Sunday Morning QB when I've got about eight hours to kill reading sundry football material (it's engrossing, I tell you, but unfortunately I have a day job). Today, SMQ has a bone to pick with the result of Kyle's home-field advantage poll. He's run the numbers, and interestingly since 1998 it looks like Mississippi State has the best home vs road winning percentage differential among SEC teams (in conference games), ergo they technically should be considered to have the best home field advantage in the conference.

I think SMQ's real gripe is with the modern connotation of the phrase "home field advantage" which, as he admittedly notes in his post, does not equate to "toughest place to play." While in days of yore, HFA very well could have been taken to mean how much better a team plays at home than on the road, as long as I've known it the heart of the reference has really been that of "How difficult it is to play at stadium X."

My issue with his calculation: a pure home-road differential clearly exaggerates the poorest teams, who invariably play poorly on the road. For a perenially winning team like Tennessee (well, till 2005), whose winning percentage on the road is a whopping .714 over this span, the MAXIMUM home field advantage they could have exhibited per SMQ's calculations (i.e. had they won every single home game over that stretch) is 0.286, and in fact Florida's max would be 0.289, which both happen to be lower than the differential actually delivered by Arkansas or Mississippi State over that period. No question that given that the latter two are just so awful on the road (0.333 and 0.152 winning percentages, respectively), it's impressive that they managed to accomplish anything at home, but the bar is just that much lower than for a team like Florida, Georgia, Tennessee, or LSU. The simple fact that the very best teams are so constrained in this calculation by their excellent play on the road completely dilutes the statistical rigor of this method and thus renders it invalid, in my humble opinion.

I'm not sure what a better way to go about calculating this would be - I'll give it some thought - but certainly this is not the correct way (or if it's the correct way, it's not the correct calculation).

(One more note, it may be a bit of a stretch to consider the SEC Championship game - held annually in Atlanta - a Road/Neutral game for the Georgia Bulldogs, given that Athens is an hour away.)

One thought offhand: one can always adjust a few things here and there. Though it could be arguable, I'm going to adjust the SEC Championship games that Georgia has played in to make them home rather than road/neutral. Further, I don't think anyone (even LSU) plays up his team's Home Field Advantage during rough stretches in said team's history (ever hear about the Swamp pre-Spurrier?) - so note that in 98 and 99 (the beginning two years of SMQ's analysis), LSU was 4-7 and 3-8. Since we're working with percentages here rather than absolutes, and since I don't have the time to go aggregate the data for the 11 other schools, I'm going to drop LSU's SEC record from those two seasons (2-6 home, 1-7 road).

Now, as mentioned above if a team like Tennessee has an astounding 0.714 winning percentage on the road during this span, and their maximum improvement is already lower than what was exhibited by MSU and Arkansas, what's the best way to adjust for their real performance at home? One method would be to calculate the maximum improvement possible (in UT's case, 0.286), and divide into that the actual improvement exhibited by the team. For instance, if your maximum improvement is 0.250 (if you had a 0.750 road winning percentage), and your actual improvement is 0.050 (for an 0.800 home winning percentage), your "improvement" if you will is 0.050/0.250 = 20%. Making these adjustments only, the rankings fall out as follows:

1. MSU 45%
   
2. Ark 44%
   
3. UGA 41%
   
4. LSU 38%
   
5. UF 26%
   
6. Miss 21%
   
7. Bama 16%
   
8. SC 10%
   
9. UK 8%
   
10. Aub 2%
    
11. Tenn 2%
    
12. Vand (-4%)

Taking it further, the Home winning percentage required by Florida or Tennessee under this methodology in order to merely have an "equal" HFA to that of Mississippi State winds up being 0.840! Methinks this asks a lot of said teams, how about you? What's more, since memory doesn't serve me and I don't have the time to dig up my old stats book to figure out how to manually calculate a margin of error, I'd say given the small sample set of data (n= ~30-35) and that a t-distribution doesn't really approach the true normal distribution until around n=120, I wouldn't say it's unreasonable to assert that the margin of error here would be at least = +/-3%. Given that LSU falls within 6% of MSU, clearly the probability is nontrivial that using the data set above, the LSU Tigers do indeed have the biggest Home Field Advantage in the SEC.

Again, no question that Mississippi State has been remarkably better at home than on the road; that said, in my opinion this is far from the conventional (and intended) definition of the term "Home Field Advantage."