BCS has achieved consensus on 4-team seeded playoff. - Page 6 - TigerDroppings.com

Posted byMessage
H-Town Tiger
Toledo Fan
Member since Nov 2003
42560 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

thats a pretty significant adjustment man; it is also gives evidence that using the committee is beneficial


I agree, but I think most of the time they would just rubber stamp the poll results anyway.

To Baloo's point it maybe better, since it offer's a failsafe, we can take Oregon over Stanford in 2011, Utah over Texas and Cal in 2004. So in that sense, it would be an improvement.







Back to top
  Replies (0)
molsusports
LSU Fan
Member since Jul 2004
26794 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:


Polls aren't objective. They are made with subjective biases. Why don't we just take teams as they are ranked in the RPI? The BCS poll is incredibly subjective, as 2/3 of it is an opinion poll. It is no accurate gauge of quality.


I agree with that for the larger part

If they are really going to do this committee thing then I'd need to hear what their criteria are before I passed more judgement.

If it is going to be about picking the "best team" well then that strikes me as just another opinion poll where we listen to people rationalize their favorite teams deserving of a mulligan. If they were to place an emphasis on respecting the importance of the regular season results then this could be a great thing for college football. Because the biggest risk facing cfb with playoffs is the risk of making cfb just another sport






Back to top
  Replies (0)
RollTide1987
Alabama Fan
Aurora, CO
Member since Nov 2009
24770 posts
 Online 

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

I'm pointing out that things LSU did which should have been positive, actually HURT their chances of winning the title. Which is frankly absurd.



Life can be unfair sometimes.







Back to top
  Replies (0)
TulaneUVA
Virginia Fan
Member since Jun 2005
20550 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

Polls aren't objective. They are made with subjective biases. Why don't we just take teams as they are ranked in the RPI? The BCS poll is incredibly subjective, as 2/3 of it is an opinion poll. It is no accurate gauge of quality.


Wasn't talking about polls. Was talking about the computer formulas.

ETA: I think there need to be like 10 formulas for determining the strength of each team. Have a bunch of sport experts, writers, statisticians submit their formulas...have some committee pick the best ones. Use the average results of them. The computers may want to incorporate poll results, but that's up the formulaic author.

If it were me, I'd balance the formula to include the following:
25% OOC result and relative strength of OOC opponent per own formula
25% Another embedded formula that incorporates offensive and defensive statistics (redzone efficiency, Turnover ratios, etc)
25% Coaches poll
25% Look at 5 previous years results/strength (this part of the formula basically says "a team that was really good 2 years ago is probably still pretty good this year...or at least has the essential cogs to be talented such as a coach, overall talent level, etc")





This post was edited on 6/21 at 10:22 am


Back to top
molsusports
LSU Fan
Member since Jul 2004
26794 posts

re: BCS has achieved consensus on 4-team seeded playoff.


But people judge the goodness of computer formulas on the basis of whether or not their results match the opinion polls though. That is why the BCS formulas have been subjected to repeated adjustment from year to year.






Back to top
TulaneUVA
Virginia Fan
Member since Jun 2005
20550 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

But people judge the goodness of computer formulas on the basis of whether or not their results match the opinion polls though. That is why the BCS formulas have been subjected to repeated adjustment from year to year.


That's why you need sports data junkies and statisticians in the room.

And even if someone judges the quality of the poll on the basis of whether or not their results match the poll opinions in a given year, THAT'S FINE. It may match one year, but if you keep it consistent and don't change it...it may show a different result in subseqeuent years.



This post was edited on 6/21 at 10:26 am


Back to top
Baloo
LSU Fan
Formerly MDGeaux
Member since Sep 2003
43901 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

That's why you need sports data junkies and statisticians in the room.

And they will arrive on their flying unicorns throwing pixie dust on everyone to make everyone agreeable and happy.






Back to top
Rohan2Reed
Grambling Fan
T'uela me la pela
Member since Nov 2003
71173 posts
 Online 

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

commissioners prefer a selection committee that picks "best four,"


Great. ESPN is already getting their lobbyists geared up to play a key role in influencing this "committee."






Back to top
  Replies (0)
TulaneUVA
Virginia Fan
Member since Jun 2005
20550 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

And they will arrive on their flying unicorns throwing pixie dust on everyone to make everyone agreeable and happy.


I know it sounds very ivory tower, but its the only way to keep corruption (media spin, bias, potential bribery) and human ignorance out of it. The current BCS formula partially acknowledges that fact.



This post was edited on 6/21 at 10:38 am


Back to top
  Replies (0)
molsusports
LSU Fan
Member since Jul 2004
26794 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:


25% Look at 5 previous years results/strength (this part of the formula basically says "a team that was really good 2 years ago is probably still pretty good this year...or at least has the essential cogs to be talented such as a coach, overall talent level, etc")


That's horrible - recently we've seen new coaches arrive at programs and transform them into national championship caliber teams within 3-4 years. The difference from year to year can be enormous - in the last 10-15 odd years we've seen several examples of teams that were mediocre the year before win a national championship the following year (auburn 09-10, lsu 02-03, tOSU 01-02, OU 99-00)

I'd rather let the Hilton and Kardashian sisters pick teams on the basis of "hotness" than that... because if it is a joke I'd rather it be understood to be a joke than pretend legitimacy







Back to top
molsusports
LSU Fan
Member since Jul 2004
26794 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:



That's why you need sports data junkies and statisticians in the room.

And even if someone judges the quality of the poll on the basis of whether or not their results match the poll opinions in a given year, THAT'S FINE. It may match one year, but if you keep it consistent and don't change it...it may show a different result in subseqeuent years.



There have been tons of statisticians, math geeks, and other informed/interested parties working on this for years.

The problem is the one you touch on in the second part of your comment - the formula that looks good one year doesn't work as well the following year. There really isn't going to be a set and dry "objective" answer from the computers or formulas because the different teams in competition are playing hugely different types of schedules.






Back to top
TulaneUVA
Virginia Fan
Member since Jun 2005
20550 posts

re: BCS has achieved consensus on 4-team seeded playoff.


Worst case scenario, it's 25% of one formula. If there were 10 formulas averaged, the resulting impact is nothing.

Now onto the basis, you're right. There are definitely cases of swings from one year to another. In all of those cases you cited (which are only a few), the prior teams were always in strong conferences and were top 40 or so finishers. Not exactly top 5 but still strong. This part of the formula prevents huge swings such as Tulane coming in at 12-0 or Wake Forest winning the ACC. You don't just have a sub-90's ranking program one year and turn into a national title condender the next.






Back to top
TulaneUVA
Virginia Fan
Member since Jun 2005
20550 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

the formula that looks good one year doesn't work as well the following year.


Then it wasn't a very good formula to begin with was it?

I don't think it should be that hard. Simplify it further then!

33% offensive/defensive stats
33% RPI
34% Coaches Poll
Done. How can you deny that this isn't objective? It should work every time! The cream of the crop WILL RISE. Fact.






Back to top
molsusports
LSU Fan
Member since Jul 2004
26794 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:


Worst case scenario, it's 25% of one formula. If there were 10 formulas averaged, the resulting impact is nothing.

Now onto the basis, you're right. There are definitely cases of swings from one year to another.


don't include it as part of your formula if it doesn't make sense

quote:

In all of those cases you cited (which are only a few), the prior teams were always in strong conferences and were top 40 or so finishers.


1) RE: the strong conferences thing? We don't really consider teams that are not from strong conferences (even if they are Boise, TCU, or Utah teams that have been good for a while)

2) If 8 win LSU and Auburn teams were ranked in the top 40 then it was inflation on the basis of conference reputation - not because either one of them deserved that. The 7 win OU and tOSU teams were just well below average teams if you judge them for what they did

Every year is different - teams will sometimes just put it all together and become special. In other cases they will fall apart when they were great the year before






Back to top
molsusports
LSU Fan
Member since Jul 2004
26794 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

the formula that looks good one year doesn't work as well the following year.



Then it wasn't a very good formula to begin with was it?


I strongly disagree. You are presuming a formula can be predictive when you are judging a bunch of different teams playing very different schedules under very different circumstances. Some teams like a 98 Tenn or 02 tOSU are solid workhorse teams that will win close games on teh basis of defense (in good conferences). Other teams like a 99 FSU, 01 Miami, or 03 USC will put up big offensive numbers against a softer schedule.

It is really not a failing of the formulas or their programmers to solve this problem IMO. It is just bad logic on the part of people who assume a formula should be able to do this.






Back to top
SuperSoakher
New Orleans Saints Fan
Member since Jun 2012
2033 posts

re: BCS has achieved consensus on 4-team seeded playoff.


I didn't read through all these pages but when does this take effect, this upcoming season??





Back to top
H-Town Tiger
Toledo Fan
Member since Nov 2003
42560 posts

re: BCS has achieved consensus on 4-team seeded playoff.


I think its after the 2013 season





Back to top
  Replies (0)
TulaneUVA
Virginia Fan
Member since Jun 2005
20550 posts

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

I strongly disagree. You are presuming a formula can be predictive when you are judging a bunch of different teams playing very different schedules under very different circumstances. Some teams like a 98 Tenn or 02 tOSU are solid workhorse teams that will win close games on teh basis of defense (in good conferences). Other teams like a 99 FSU, 01 Miami, or 03 USC will put up big offensive numbers against a softer schedule.


Where I am being predictive? Use the current YTD stats, strength of schedule, and coaches poll ranking to evaluate current standing?

RE: Style of play. The offensive/defensive statistics that are incorporated would account for that.

quote:

It is really not a failing of the formulas or their programmers to solve this problem IMO. It is just bad logic on the part of people who assume a formula should be able to do this.


I think you vastly underestimate the power of algorithms.

My point is this: Because formulas are the only way to keep it honest, we MUST get it correct.



This post was edited on 6/21 at 11:09 am


Back to top
  Replies (0)
Archie Bengal Bunker
LSU Fan
Member since Jun 2008
15258 posts

re: BCS has achieved consensus on 4-team seeded playoff.


This change is coming about largely because of "the eyeball test.". A four person committee is nothing but an EYEBALL TEST! They will have to choose between 6-9 viable teams, and it will come down to who those four experts think look the best.





Back to top
  Replies (0)
RandySavage
Member since May 2012
8128 posts
 Online 

re: BCS has achieved consensus on 4-team seeded playoff.


quote:

The conference champion requirement is idiotic.


As opposed to a selection committee? Give me the Conference Champ everytime.






Back to top


Back to top