RoadRage

I have what I think are the correct formulae finally.

I've added iteration steps, so that all results are distributed evenly throughout the entire matrix without having to resort to matrix algebra. For example, tonight game (WV v Miami) has an effort on Cincinnati, a lesser effect on Miami-Ohio, lesser still on Iowa, etc, until the effect disappears into the error noise.

2002 results with the system? I thought I'd have them last weekend. Blah. I get it sometime.

http://members.cox.net/wd40

Don't bitch about Texas. They haven't beaten anyone of consequence yet (#103 New Mexico State, #108 Rice, #68 Tulane). They're lucky to be as high as they are. Take it up with your team's scheduler, or your team for losing to the only worthwhile opponent they've faced.

__________________
Join TFP Team SETI
43K workunits complete, 34 members, more of each needed.

The_Dude

just an observation, but arkansas seems to climb up every time you post an update.

maybe they will be #1 next week?

__________________
"Luck is what happens when preparation meets opportunity." - Darrel K Royal

Kurant

lol.

Prolly after they beat Flordia on the 18th they'll be #1.


Miami 8th?

:

gov135

The rankings look great from the MAC conference perspective. Northern Illinois is situated correctly, and Miami of Ohio is probably the best of the rest. They look good all the way through Ohio University.

Why would the WVA/Miami game have a lesser effect on Miami of Ohio then Cincinnati? After all, Miami of Ohio beat WVA?

I been holding my tongue, but I gotta tell you, Roadrage, I hate margin of victory being included in these formulas. USC may beat the scum of the Pac-10, Arizona, by 30 points. Ohio State may beat the scum of the Big-10, Northwestern, by twenty points. It doesn't mean one is more impressive than the other - teams that play and rely on defense and ball-control offense like Ohio State and Virginia Tech tend to be penalized. But Virginia Tech may be the best in the land.

Finally, this is alot of work and it's fun to look at, so thanks. All you people who are whining about where your team is, you need to realize that you can't judge RoadRage's formula until the end of the season - Texas, for example, will move up if they take care of business in the meat of their schedule.

RoadRage

Miami had been 2nd until they just barely beat West Virginia. Miami certainly didn't look like a Top 10 team last night.

The equations:

mu0=(W*+1+.01*CAL)/(W*+L*+2+.02*CAL) where CAL = calendar days left in season (64 as of today), W* = number of wins of all type, and L* = number of losses of all type
OTW=(6+sqrt(mov))*mu(OPP)
W=(9+sqrt(mov))*mu(OPP)
SOW=(12+sqrt(mov))*mu(OPP)
OTL=(6+sqrt(MOL)*(mu(OPP)-1)
L=(9+sqrt(MOL)*(mu(OPP)-1)
SOL=(12+sqrt(MOL)*(mu(OPP)-1)
sigma0=sum of all OTW, W, SOW, OTL, L, & SOL
omega0=12*sigma/(W*+L*)
Iterations:
Generate a best-fit line of omega = slope*mu+(y-intercept)
Calculate the expected omega for each team based on the mu.
delta=(omega(calculated)-omega(expected))/(2*(omega(all teams max)-omega(all teams min)))
mu1=mu0+delta
Lather, rinse, repeat, until delta=0 (at least to three digits).

Now, if you can CORRECTLY explain to a reasonable panel of unbiased judges how that's biased toward Arkansas, I will send you a check for $1000. Actually, with the lessened award for an OT win, I should be being accused of bias AGAINST Arkansas.

If you don't like margin-of-victory, set it up and run it without it. For all I know, it may be so similar that MOV isn't needed. I think it's a needed component to ranking, and using the square-root of it provides a good deflator for runaway games. (Third- and fourth- roots are under consideration if something gets to looking too out-of-whack.)

Miami dropped from 76.34 before the game. Arkansas increased from 76.30 to 76.36. Miami-Ohio went from 45.95 to 45.96. Cincinnati dropped from 18.58 to 18.57. Somehow, it's just the way the math works out. (Miami-Ohio doesn't play West Virginia this year. Cincinnati did.)

For those interested, Colley (one of the BCS ranksters) allows you to "play god" and change games to see what-if and their effects. http://www.colleyrankings.com/ Sometimes the effects are ones you wouldn't have imagined, even on teams that haven't played either of the teams you manipulate.

__________________
Join TFP Team SETI
43K workunits complete, 34 members, more of each needed.