Home Field Advantage

[BlondeTuco]

Last October, I stupidly thought it would be a fun task to create my own football rankings, similar to the computer component of the BCS. Quickly it became an obsession, and I spent hours trying to find the math to justify different parts of my equations. The biggest ad hoc problem was measuring the magnitude of home field advantage.

A fix for all the "solutions" currently online required a way to compare opponents. Winning percentage at home is an awful measurement - a certain team who has a high winning-percentage on their blue turf may benefit from a weak schedule. If two teams are of equal strength, we would expect them to split their series 50/50, making them prime targets for my experiment; if Team A wins 60% of their home games against equally-talented Team B, they're presumably netting a win 10% more often because they're playing in front of a home crowd.

Obviously I can go more in-depth if anyone actually cares about the math and logic, but here are how the conferences compare to each other, factoring in 1515 games from the last two seasons:

Big 12: +15.2%
Big Ten: +14.1%
C-USA: +14.1%
SEC: +12.9%
PAC-12: +11.3%
ACC: +10.8%
MWC: +9.1%
WAC: +8.2%
MAC: +8.0%
BEast: +6.9%
Indys: +5.0%
SBelt: +3.3%
(based on conference alignment for the 2012 season)

 

[ColoradoHAWKEYE]

This brings insomnia to a whole new level.

 

[DuffMan]

At first I thought your name was blue taco.

 

[olddude]

This brings insomnia to a whole new level.

No kidding, but I will throw a log on, I enjoy a big fire. How did you figure they were equally talented? What about injuries? There were 757 (give or take) games in each of the last two years that meet all the requirments to come up with a dependable formula?
Why not just go with 3 like most people? If you want to go 3.5 or even 4 (like that matters) do so.
Dont get me wrong, I like a good formula as much as the next guy, but for 1 point, it hardly seems worth it, seems to me if you feel bold, go with 3. If not go with 4. I would just wait for starters to come out and assign "points" to each team. On positions with high injuries, go two deep, that way you can adjust very quickly during the season.
Good luck.

 

[BlondeTuco]

This brings insomnia to a whole new level.

Like you can't even imagine.

I only considered home field advantage for my binary rankings (wins and loses), in a similar manner to how RPI is calculated for college basketball. Basically a win at home counts as .786 wins, and a loss 1.214 loses. Since my aim is to evaluate schedules and not predict future games, I don't consider injuries at all (are the BCS computer rankings any different?).

I considered two schools to be "equal" if they were within .15 ranking points from each other according to my personal ranks (~+/- 15 positions from each other. Very ad hoc, but I needed to start somewhere. 384 match-ups fell within this range, fairly-accurately measuring HFA.

The Big East home teams performed very poorly against comparable opponents (-17.9%), but upset more frequently and are upset less frequently. This forced me to find a way to score every game, not just ones against closely-matched teams. So:

%Win - Expected %Win = HFA
in "equal" games....
%Win - 50% = HFA
60.7% - 50% = 10.7%

against superior opponents....
15.8% - Expected %Win = 10.7%
Expected %Win = 5.1%

and against inferior opponents...
93.6% - Expected %Win = 10.7%
Expected %Win = 82.9%

Therefore every game is scored, with HFA being a weighted average of Actual Win % - Expected Win %. In actuality, the stronger conference's home field strength are most strongly impacted by games against weak push-over opponents, because of the relative ease of their non-conference schedules. The Big Ten for example:
28-8 against similar opponents
9-35 against superior opponents
79-8 against inferior opponents

.778 - .5 = .28
.2 - .05 = .15
.908 - .83 = .08

(.28*36+.15*44+.08*87)/167 = .141

/ridiculous stats rant

 

[FalseMirage]

At first I thought your name was blue taco.

I was wondering why he left the Lee Van Cleef characters name out of his handle.

 

[bcl20]

Obviously I can go more in-depth if anyone actually cares about the math and logic, but here are how the conferences compare to each other, factoring in 1515 games from the last two seasons:

I would like to see more details myself.

 

[DuffMan]

anyone know the difference between a blue taco and a blue waffle

 

[FalseMirage]

anyone know the difference between a blue taco and a blue waffle

it would make sense if a blue waffle was called blue taco

 

[GhostofBTT]

 

[1hawkeye1]