Harris Poll for Week 9

[Keith7]

I went to the site and saw he put a disclaimer up so i took down what i put up here

if you'd like to see then click this link

http://www.smoaky.com/forum/viewthread.php?tid=25952

i'm not afraid of anybody, just respect what this guy does

[44INAROW]

Originally posted by Keith7
AAA
1. Decatur 234
2. West Orange - Stark 230
3. Cuero 229
4. Gilmer 227
5. Marlin 226
6. Jasper 225
7. Liberty Hill 224
8. Rice Cons. 224
9. Cameron Yoe 223
10. Sinton 222

PLEASE DONT TAKE ME TO JAIL!!!

I am crossing my legs, laughing so hard it hurts..............

[sinton66]

Keith ain't skeered

[JHS_c/o_06']

Its amazing how #'s 4-10 are all 1 point difference from one place to the next....one game could mean the difference between 10th and 4th. Glad to see Jasper is still there...im hoping we'll be back in the top 4 by the time playoffs roll around.

[footballgal]

Maybe this is a stupid question, but if I don't ask I will never know... Why is there a number next to the Team? Is it like the AP votes??


Keith, that was pretty funny! " Don't take me to jail!":D

[JHS_c/o_06']

I think, i could be wrong, but i think thats the "power" ranking.....i think its called the Harris Power Poll or something like that ....could be wrong though.

[sinton66]

Yes, it's a composite number based on their own formula.

[JHS_c/o_06']

I wonder how long it took them to come up with it (the formula) it took me nearly an hour to work out some AP Algebra 2 problems....let alone write a formula.

[sinton66]

They're pretty complex. One of my favorites was Data Poll. Here's how they explained it:

How DataPoll Football Rankings Work

Simple Version
the basic idea of how it works.

1. How predicted lines are determined.

The predicted line is the difference between the power ratings of the two teams.

PL=Power1-Power2
2. How new rankings are computed.

First, all teams' power ratings are set to 100. We cycle through all played games starting with the last and moving back to the first, and then from the first to the last. At each game we take the difference(DL) between the predicted line(PL) and the actual line(AL). The power rating of each team is adjusted by some factor(F) of DL/2.

PL=Power1-Power2
AL=Score1-Score2
DL=PL-AL
0<F<=1(predefined)
D=F*DL/2
newPower1=Power1 - D
newPower2=Power2 + D
We repeat the iterations through all the games until the greatest change in power by any individual team in a particular cycle falls below a certain threshold. At this point we decide furthur iterations will not produce signifigant changes.

Its Not That Simple
modifications to basic method.

1. Preseason ratings

Ratings using the simple version are wildly unreliable if teams have played fewer than 5 or so games. The statistical sample is too small and not all teams can be linked through opponents (and their opponents opponents, etc.). So a base pre-season power(BP) is determined.

My method for determining BP takes into account the team's final power rating of last season(PLast), its average final power rating of last 5 seasons(PHistory), and the number of returning starters(RS). A factor(E) is added for the experience of the returning starters and subtracted for the inexperience of the new players.

BP=(RS/22)(PLast+E) + (1-RS/22)(PHistory-E)
All teams' base powers are then adjusted so the total average is 100.
For every cycle through the games, before we visit the first game, we factor in BP for each team:

0<F<=1(predefined)
newPower = Power + F*(BP-Power)
2. Line factor

The predicted line is adjusted using the line factor(fL). The predicted line is increased for power differences indicating that the weaker team cannot compete and will be blown out. However, the line will be reduced for extremely large power differences so that we do not have many predicted lines over 60 or 70 points.

PL=fL(Power1-Power2)
3. Blowout factor

I didn't want teams to be able to increase their ratings greatly by beating up on the weaker teams, or punished greatly by not beating weak teams by enough. So I created a function (fb(x)) which takes a point line, and reduces the value of large point lines. So DL is now has a new equation:

DL= fb(PL) - fb(AL)
The result of this is that a team's power rating is affected more greatly by points scored to decide the outcome of games and less by points to blowout games. For example, if a team wins by 40, the first point affects a team's power the most(because it is the one that determines who won), and each additional point less so, until the difference between winning by 39 or 40 points is almost negligible.

Sheeesh! If you can figure that out, you're a better mathematician than I.

[JHS_c/o_06']

Sinton...what are you talking about....you know i cant read spanish....why did you post all that.....lol....Anchorman was an awesome movie.

But wow....im speechless...im not gonna even read it...brings back too many bad memories.

Oh..and BTW...it took me 2 years......but i finally reached 100 posts...im so happy.