Jump to content
SiouxSports.com Forum

Recommended Posts

Posted

Manipulated, as in skewed?

If you manipulate the numbers for the formula, you manipulate the numbers for every team. They all get the skew evenly, which negates the effect of skewing.

However (i think this might be what you meant), if you manipulate the numbers for simply one team, then this is cheating. One way to combat this would be to make the forumla public so it could be checked by ESPN and whoever else to make sure there wasn't a watergate happening.

The problem with the computer polls is that they each take into account different things. For example, one computer might reward an overtime win, while another might punish you. So a lot of coaches don't even know what they should be accomplishing to improve their team, other than winning. But some computer polls seem to even reward losing since some schools with more losses are ahead of teams like LSU and USC in the computer polls.

I think they revamped this, but a lot of the systems used victory margin. This stat could easily be used as a skew since some teams are more likely to run up the score than others. All the computer programmer has to do if he likes a team that is known to run up the score is set his poll to reward running up the score.

IMO, the computer poll system is as flawed as the human poll system since they are both controlled by humans. Each computer poll is different and what makes one man's computer poll better than another's computer poll? How many people would get to create their own polls?

  • Replies 71
  • Created
  • Last Reply

Top Posters In This Topic

Posted

Both of you seem to be hinting that there would be more than one computer which each had it's own, different formula.

Also it seems like you guys think that everyone and their brother would have their own computer formula that would be taken into account.

Sure, whoever wanted to could make up a computer formula and skew it in whatever way he/she wanted. So what?

***There would be only ONE computer with ONE forumla (that would be made public) that would be used for ranking/seeding. The criteria that would be used for this forumla would have to be agreed upon by more than one person and i would suggest that they use the least biased criteria that they possibly could (e.g. things like number of wins/losses which don't matter what the score is).***

Posted

they still should go to a playoff system, something like we have here in D2. while D2 selection and seeding is a little to be desired yet, (see bentley being picked over IUP), at least we can fairly say, that when the dust settles on saturday, that we will have an undisputed national champion. lord knows what kind of mess its going to be in D1a. i was born and raise within shouting distance of the Big House and i was quite excited to hear that they where going to the Rose bowl. know that now i can sit back at home new year's morning, eating my Momma's cookin', watching the rose bowl parade and then go watch michigan wack somebody. but now, i'm kinda torn between cheering for michigan to win, or to see USC win and totally fudge the BCS up. the arguement is that fans dont travel to playoffs. bull schitt!!

Posted
Scores of the games would have to count on some level!

My point is that one computer program or multiple computer programs are still going to have problems. I don't see how any program can account for different team styles and coaching philosophies.

In most or all cases the team that shows little or no class and pounds a crap team will benefit. While a team that chooses not to do that may suffer, even if they are the best team in the country.

Even if the system is not perfect, a 16 team playoff would fix a lot of the problems. Regardless of how they picked teams for the playoff.

You're intitled to your opinion.

That is why i think the one, end all, be all formula must be decided upon by many different people. I don't think it can ever be perfect for everyone, but it will be as close as we'll get to unbiased perfection. And making it public will allow everyone to calculate it themselves as soon as the scores come in. This will lead to great debates of "if team X wins this game..."but if team y loses here..", and so on.

Posted
The only thing that matters are things like the number of wins and loses by yourself and the wins/loses of your opponents and opponent's opponents.

OK .... then how to you weight them?

WinPct is 50%

OppWinPct is 35%

OppOppWinPct is 15%

or how about

WinPct is 33.33%

OppWinPct is 33.33%

OppOppWinPct is 33.33%

or how about

WinPct is 75%

OppWinPct is 15%

OppOppWinPct is 10%

or how about

WinPct is 35%

OppWinPct is 50%

OppOppWinPct is 15%

What the above is called is the Ratings Percentage Index, or RPI.

Ask college basketball (and college hockey) about trying to find the right ratios (between WinPct, OppWinPct, OppOppWinPct) to find the best teams.

Everyone has their own opinion, their own personal bias, as to what the right numbers ratios are.

Posted

Anybody recall how many times the college hockey RPI formula has been tweeked in the last five years? My memory says at least two (new ratios, and then this "adder" stuff from last season).

Trying to get a majority on something like this would make herding cats seem easy.

Posted

I explained the RPI. The bias there comes in how to weight the various percentages applied.

Good teams in weak conferences want a big factor on WinPct and a small factor on OppWinPct.

Bad teams in strong conferences want a small factor on WinPct and a big factor on OppWinPct.

So, which is correct?

The best answer is the Bradley-Terry methodology (called KRACH in college hockey), but because of the few games played in college football the methodology does not work.

The answer is clear: Playoff. However, playoffs favor teams that don't lose late in the year .... like Oklahoma did.

Posted

The answer is a play-off.

However, i fail to see how losing late in the year should hurt anyone.

No formula will work perfectly for everyone. If there was a 16 team play-off then even if oklahoma lost to Kstate 100-0, they would still make the play-off field.

Posted
I explained the RPI. The bias there comes in how to weight the various percentages applied.

Good teams in weak conferences want a big factor on WinPct and a small factor on OppWinPct.

Bad teams in strong conferences want a small factor on WinPct and a big factor on OppWinPct.

So, which is correct?

The definition of which is correct tends to be "what makes it looks like polls" and "what seems consistent with known H2H results".

So many of us despise Massey (the complex one, not the BCS one) because it frequently fails test #2 because of its overemphasis among scoring differential, among other things. The Massey used in BCS does not include all those whacky things, and is actually quite similar in mission and result to Bradley-Terry.

Test #1 is actually not as stupid as it seems -- if you can come up with a simple formula that answers the same question as polls, then you can apply that formula objectively going forward. When it deviates from polls, then you can ask the interesting question of why? For example, polls tend to reward past success (i.e. Gopher Hockey) more than past success actually influences the current season. If you work in the other direction and continuously "correct" the formula, then you're right back at college hockey's RPI or Massey.

As to which is actually correct -- depends on the schedule. When hockey used to have a few strong conferences, emphasizing Win% to 35% produced more palatable results. Once a couple significantly weaker conferences with little interlock were introduced into the mix, it became obvious that SOS needed more weight, leading to the current 25-50-25 (the same as basketball).

Football just doesn't play enough interlocking games to correctly identify the top two teams with any certainty. It seems even Sioux and Bison fans can agree, bring on a DI football tournament.

Posted
The best we can hope for is a 4 team playoff after the bowl season, which means extending the season into mid January. After a few years of this, maybe the bowls will lose their luster as a primary goal.

I think this is a realistic possibility after the current BCS contract runs out. Take the four major bowl winners and extend the season. This would not disrupt the pomp and circumstances that surround the current bowls, such as the Rose. Television revenues could be maximized by running double-headers on both New Year's Eve and New Year's Day. The Sugar Bowl used to be on New Year's Eve anyway. The reason I think this is realistic is the fact that there is money to be made everywhere you look, with 3 of the existing bowls becoming more meaningful, and an additional three games with all the accompanying revenue.

Posted

There was an article in the USA Today yesterday about the BCS possibly adding a fifth bowl after the current four are played in 2006. The teams would be ranked and the top two would play in the "championship" bowl. It still wouldn't be perfect, but it would be an improvement.

Posted
There was an article in the USA Today yesterday about the BCS possibly adding a fifth bowl after the current four are played in 2006. The teams would be ranked and the top two would play in the "championship" bowl. It still wouldn't be perfect, but it would be an improvement.

Couldn't be the "Championship" Bowl. Needs a sponsor.

How about the Time-Warner AOL Microsoft WInston Salem Bowl?

Posted
How about the Time-Warner AOL Microsoft Winston Salem Bowl?

.... sponsored by Nextel. Coming to you live from the Honda Broadcast Booth via the SkyTel up-link here are your hosts .....

Posted

Wouldn't this take too long? Some kids would be playing football into the spring semester!

No. They could start earlier, plus these guys are doing something football related all year anyways. Christmas break is quite long at most schools, plus football teams rarely miss school. They might leave Thursday night at the earliest and most college students don't have Friday classes. If college bball teams can be gone for a month during March Madness, then football can play one more week, or lose a meaningless nonconference game from the schedule.

Posted

No. They could start earlier, plus these guys are doing something football related all year anyways. Christmas break is quite long at most schools, plus football teams rarely miss school. They might leave Thursday night at the earliest and most college students don't have Friday classes. If college bball teams can be gone for a month during March Madness, then football can play one more week, or lose a meaningless nonconference game from the schedule.

I'd rather see a full scale play-off, but a play-off between the bowl winners does seem like a reasonable fix for now.

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...