Tweaked BCS polls

[Nutriaitch]

over the last few years, I have been tinkering with the formula a bit for the BCS.

I may have finally come up with one that cuts down on (what I perceive to be) flaws with the current formula

My formula gives all 8 polls (6 computers, coaches, harris) equal weight.
unlike now where each human poll is 1/3, and the computers combine for the other 3rd.

My formula also gives equal weight to the computers as the humans.

Results obviously differ a bit from the actual BCS standings

1. Florida ---------- .9700
2. K-State ---------.9683
3. Bama ---------- .9580
4. N-Dame -------- .9263
5. Oregon --------- .9165
6. Oregon St. ----- .8797
7. LSU ------------ .8727
8. Oklahoma ------ .8256
9. Miss. St --------.7823
10. USC ------------.7815

 

[Jake]

over the last few years, I have been tinkering with the formula a bit for the BCS.

I may have finally come up with one that cuts down on (what I perceive to be) flaws with the current formula

My formula gives all 8 polls (6 computers, coaches, harris) equal weight.
unlike now where each human poll is 1/3, and the computers combine for the other 3rd.

My formula also gives equal weight to the computers as the humans.

Results obviously differ a bit from the actual BCS standings

1. Florida ---------- .9700
2. K-State ---------.9683
3. Bama ---------- .9580
4. N-Dame -------- .9263
5. Oregon --------- .9165
6. Oregon St. ----- .8797
7. LSU ------------ .8727
8. Oklahoma ------ .8256
9. Miss. St --------.7823
10. USC ------------.7815

What you did is what the BCS formula used to be until it dared to produce a result that conflicted with the human polls. They then changed it to give the humans more weight.

I think that was in 2004, right after the AP threw a hissy fit and withdrew from the BCS.

 

[Nutriaitch]

What you did is what the BCS formula used to be until it dared to produce a result that conflicted with the human polls. They then changed it to give the humans more weight.

I think that was in 2004, right after the AP threw a hissy fit and withdrew from the BCS.

well, I don't have the SoS or Quality win component.

would love to add them, but don't have the time or patience to go dig up that info.

2004 was the 1st year with the current formula.
AP was still used instead of the Harris, but it was the same mathematical formula we have now.

 

[MililaniBuckeye]

Any poll that doesn't have Alabama as #1 right now is fucked. They are clearly the best team in the country...

 

[OHSportsFan]

Any poll that doesn't have Alabama as #1 right now is fucked. They are clearly the best team in the country...

Except for Florida's whole strength of schedule thing.

 

[Nutriaitch]

Any poll that doesn't have Alabama as #1 right now is fucked. They are clearly the best team in the country...

I agree that I think they're better than anyone else.

but if we're going off of what's already happened, Florida has the better resume so far. (with games against UGA, FSU, and SEC West Champ still to come)

meat of Bama's schedule starts this week.
Miss St, LSU, then A&M.

 

[Dryden]

What you did is what the BCS formula used to be until it dared to produce a result that conflicted with the human polls. They then changed it to give the humans more weight.

I think that was in 2004, right after the AP threw a hissy fit and withdrew from the BCS.

The BCS was never like what Nutria did here. His suggestion gives every poll equal weight. The original BCS was an index of the Coaches & AP average + three CPU polls (Sagarin, A&H, NY Times) averaged + SoS + # of Losses. The poll components being averaged netted (it was thought) a 1:1 CPU consensus vs human consensus. Nutria's idea is strictly 3:1 weighted towards the computers.

The major BCS tinkering occured after the 2000 regular season when Oklahoma (#1 AP) played Florida St (#3 AP) instead of Miami in the 2000/2001 title game. Florida St (as well as Nebraka, Oregon, and Kansas St, among others) had been accused of running-up scores throughout the 2000 season to influence the margin of victory component in the computers used in BCS rankings. In this case, Miami beat FSU head-to-head during the regular season, but FSU overtook Miami in the final BCS standings largely by virtue of beating Clemson 54-7 in that Novembers "Bowden Bowl." That season, Oklahoma and Miami both had tremendous defenses, but FSU ultimately got the chance to play for a title because the offense averaged 550 yards and 42 ppg.

Even the at-large criteria came under heavy fire for the first time that season after Notre Dame was taken ahead of Virginia Tech by the Fiesta Bowl, and Notre Dame was subsequently demolished by Oregon St.

Ahead of the 2001 season, the BCS replaced the NY Times and Dunkel computer indices with two that did not have margin of victory so heavily weighted, changed the CPU average to remove the high and low score (previously they only adjusted the worst deviation ['98] or tried just removing the lowest ['99 &'00]), and they also added the Quality Win component based on wins against other teams finishing in the BCS Top 15.

Before 2002 even more computers were removed because of MoV, NY Times returned with scores removed, and Jeff Sagarin was forced to make a BCS alternative (ELO_CHESS).

Go back to the end of the '98-'99 season though and there was already a strong sentiment to "fix" the formula when it was recognized scoring margin against weak schedules was going to prove problematic. Despite the changes after the first season of the BCS the formula was still "broken" in Year 3.

 

[lvbuckeye]

Any poll that doesn't have Alabama as #1 right now is fucked. They are clearly the best team in the country...

Except for Florida's whole strength of schedule thing.

and the fact that Oregon rolls up 40-60 points in the first half and plays walk-ons for the entire second half.
329 rushing yards IN THE FIRST HALF at Arizona State is pretty freaking impressive.

 

[JXC]

Except for Florida's whole strength of schedule thing.

SoS is way overrated as far as a sole factor in comparing teams.

Who a team beats doesn't make them a better team.

Alabama has played an easier schedule than Florida. But Alabama is clearly better than Florida, doesn't matter who they played. If Alabama had only played FBS schools, they still would be better than Florida.

I like the fact that there will be a committee who will look at everything starting in 2014. Computers don't have brains.

After each team plays their 12 or 13 games, you take a step back, and look at the whole body of work. You look at how the team performed, how the players performed, how the coaches performed. You see how they looked against bad teams, good teams, on the road, at home. You look at how they did in close games. And then you take all of that information, and come up with who are the four best teams.

We still have plenty of games to go, but in my opinion, Alabama is a clear #1. After that I give Kansas State an edge at #2. Then I like Oregon at #3 and I like Florida at #4.

I think humans do a WAY better job of figuring out who the best teams are. Especially when those humans can be as unbiased as possible. I think the NCAA committee for the basketball tournament does a great job in this.

 

[Jake]

Any poll that doesn't have Alabama as #1 right now is fucked. They are clearly the best team in the country...

Ten years ago, some were saying the Miami Hurricanes were better than some NFL teams. The "eyeball test" isn't always the best measure.

If only there was a way to have the undefeated teams play each other in a single elimination format...

 

[redguard117]

and the fact that Oregon rolls up 40-60 points in the first half and plays walk-ons for the entire second half.
329 rushing yards IN THE FIRST HALF at Arizona State is pretty freaking impressive.

I guess that's not bad . I want an Oregon UF matchup

 

[JXC]

Ten years ago, some were saying the Miami Hurricanes were better than some NFL teams. The "eyeball test" isn't always the best measure.

If only there was a way to have the undefeated teams play each other in a single elimination format...

It's a lot better than computers.

 

[Jake]

SoS is way overrated as far as a sole factor in comparing teams.

Who a team beats doesn't make them a better team.

It plays a part in their performance, and how a team "looks".

Alabama has played an easier schedule than Florida. But Alabama is clearly better than Florida, doesn't matter who they played.

Yes, and the Miami Hurricanes in 2002 were clearly better than Ohio State. Everyone knew it. Many questioned why they even had to play the game.

If Alabama had only played FBS schools, they still would be better than Florida.

I like the fact that there will be a committee who will look at everything starting in 2014. Computers don't have brains.

They also don't have thoughts clouded by emotions, or care what name is on a team's jerseys, or their school's tradition, etc..

After each team plays their 12 or 13 games, you take a step back, and look at the whole body of work. You look at how the team performed, how the players performed, how the coaches performed. You see how they looked against bad teams, good teams, on the road, at home. You look at how they did in close games. And then you take all of that information, and come up with who are the four best teams.

It sounds so simple when you put it that way. And yet, every year there will be disagreement with whichever humans make that choice.

We still have plenty of games to go, but in my opinion, Alabama is a clear #1. After that I give Kansas State an edge at #2. Then I like Oregon at #3 and I like Florida at #4.

I disagree. My feelings are different than yours. Now what?

I think humans do a WAY better job of figuring out who the best teams are. Especially when those humans can be as unbiased as possible.

There is no such thing as an unbiased human. Not even close. As for a WAY better job, I'll let the world in general speak to how great humans are at figuring things out. The majority have no clue.

I think the NCAA committee for the basketball tournament does a great job in this.

And yet, every year people go apeshit over some of their selections and seedings. Every year.

 

[Zander42]

Ten years ago, some were saying the Miami Hurricanes were better than some NFL teams. The "eyeball test" isn't always the best measure.

If only there was a way to have the undefeated teams play each other in a single elimination format...

And three weeks ago the WV Offense was unstoppable.

 

[Jake]

It's a lot better than computers.

Both humans and computers are flawed ways to select playoff teams. But, unlike humans, computers apply their formulas consistently to everyone.