College Football Rank 'em --- Week 8 - Page 10

BobJoeJim · #91 10-22-2007, 05:51 AM

For that matter, I'd also be interested in the same analysis using the #60 rating. How much better than "average" has each of these teams done, given their schedule?

Edit: This would get rid of those pesky 99% numbers that make Kansas' schedule look like the toughest in the nation. Whether this is a good thing is debatable. It's actually probably better with the #5, when comparing top teams. Still, it would be interesting to see the differences.

MyTurn2Raise · #92 10-22-2007, 05:56 AM

Yeah, I hope to do #25 and #50 as well. It's getting near 5AM in my neck of the woods, so the bed is calling to me.

FWIW, Hawai'i has 5 99% and 2 games near 92%

obviously, with 0 loses they are still going to be a nice positive number according to the rankings, but they won't be anywhere close to the other undefeateds.

maxtower · #93 10-22-2007, 05:57 AM

Do your rankings based on where you think teams will be before the bowl games.

I know most of you don't really think ASU is going to be top 10 by then.

Here's mine!
1. OSU
2. LSU
3. Oregon
4. West Virginia
5. Virginia Tech
6. Oklahoma
7. Hawaii
8. Kansas
9. Florida
10. USC
9.

MyTurn2Raise · #94 10-22-2007, 05:58 AM

[ QUOTE ]
[ QUOTE ]
In a results oriented world,
Kansas is the #1 team!!!!!!!!!!

You math nerds might like that. Give some feedback on how to make the assumptions better.

[/ QUOTE ]
Hmm, very interesting. I'm curious to see if it would come out any differently if you used #1 as a basis instead of #5. I'm also surprised to see that this methodology ranks Kansas' schedule as tougher than ASU/OSU/BC, considering that Sagarin's SoS numbers have those three as all having played noticably tougher schedules than Kansas.

I guess the difference is that Kansas has played more COMPLETE patsies, and once the winning chances hit 99% it doesn't matter how much worse the team is. Playing the #120 team isn't really worse than playing the #80 team when you look at it this way, since a "true" #5 team should beat either 99% of the time. Looking at it this way, it doesn't matter that Kansas' weak opponents are collectively weaker than those of the other three, they can ride the KSU game (which ranks as the toughest that any currently undefeated team has played) and come out as having the toughest schedule of any current undefeated because of it.

I'm not sure yet if this is a flaw in your analysis methodology, or a revelatory realization, lol. I'm gonna have to think a little more about it.

[/ QUOTE ]

I have always complained that sagarin and the other stuff I've seen use a linear weighting for SOS when reality doesn't match that.

iggymcfly · #95 10-22-2007, 06:00 AM

I can run the numbers for Hawaii off the top of my head without even looking at the Sagarin ratings or Hawaii's schedule:

99.0
99.0
99.0
99.0
99.0
99.0
99.0

= 6.93 expected wins

Hawaii = +0.07 wins.

If you actually try to compare Hawaii though, you'll have to get off the 99.0 system a little. For instance, the current Sagarin #5 (Oklahoma) would be a 68.5 point favorite over Northern Colorado at home. I think you can pretty comfortably call that 100.0%.

MyTurn2Raise · #96 10-22-2007, 06:00 AM

Here is the link to the colton index
which shows the flaws in rpi and linear SOS weightings

MyTurn2Raise · #97 10-22-2007, 06:03 AM

[ QUOTE ]
I can run the numbers for Hawaii off the top of my head without even looking at the Sagarin ratings or Hawaii's schedule:

99.0
99.0
99.0
99.0
99.0
99.0
99.0

= 6.93 expected wins

Hawaii = +0.07 wins.

If you actually try to compare Hawaii though, you'll have to get off the 99.0 system a little. For instance, the current Sagarin #5 (Oklahoma) would be a 68.5 point favorite over Northern Colorado at home. I think you can pretty comfortably call that 100.0%.

[/ QUOTE ]
considering I bet Hawai'i -59.5 that game and thought it was a tough break to not win. Yeah...100%.
Even changing the 99%s to 100%s does very little to change the overall results. Even changing one team's 99%s to 100%s wouldn't change the final rankings or come close to moving anyone a spot.

BobJoeJim · #98 10-22-2007, 06:06 AM

[ QUOTE ]
Here is the link to the colton index
which shows the flaws in rpi and linear SOS weightings

[/ QUOTE ]
Loving this, thanks for a great link!

iggymcfly · #99 10-22-2007, 06:07 AM

Yeah, really comparing the Top 5's expected wins only works with Top 5 caliber teams. Hawaii's schedule isn't in the ballpark to even be relevant.

I would be interested in seeing what the numbers come out to for percentage of an undefeated season with Hawaii's schedule with a #50 type team though. Looking it over today, I was saying that I thought Duke would be > 50% to be 7-0 against those teams. Hawaii hasn't even played one team in the Sagarin Top 100 all year!

MyTurn2Raise · #**100** 10-22-2007, 06:15 AM

[ QUOTE ]
Yeah, really comparing the Top 5's expected wins only works with Top 5 caliber teams.

[/ QUOTE ]

This is a really good point

perhaps if I had the computational power, one could get a decent ranking system by seeing how all the teams split themselves up at different dividing points--such as #5 team, #10 team, #20 team, #35 team, #50 team and seeing how the near teams fall out from that jumble

oh well...it's only a results oriented ranking anyway

it says nothing about who the better or best team is

I'm really struggling with the notion that it spit out Kansas #1

I realize I'm heavily dependent on sagarin's rating structure. Perhaps, he just has @ K St and @ Colorado rated a bit too highly.

I am somewhat comforted by the process if one compares the chances that the respective teams would be undefeated right now. It clearly shows that LSU only had a 18.93% chance to be without a loss if they were equivalent to the #5 team. Every other team had >37.5%. That makes sense.
Just didn't realize how tough a game @ K St is. At least I feel better about betting K St v Kansas and Colorado v Kansas.