<p>Big game in Chapel Hill this coming Saturday, as the Tar Heels take on Dook. Normally the game is a mere afterthought to UNC (UNC has won 18 of the last 19 against Dook), but for the first time in about 30 years Dook football has been able to win more than 2 games in a given season, so all bets are off as to what happens this year.</p>
<p>
It is impossible for Florida and Alabama to play against each other for the BCS title because they will play each other for the SEC title. Whoever is the victor will most likely play against Texas.</p>
<p>^ The key word is “or.”</p>
<p>And I think that Florida/Alabama going to the NC is the surest thing there is at this point.</p>
<p>
</p>
<p>LSU could beat Bama and then beat Florida in the SEC title game.</p>
<p>^^I hope that happens, then Texas will have a much better shot at winning it all!</p>
<p>
</p>
<p>I only said that Florida or Alabama in the NC was the most LIKELY outcome, not that it would happen for sure. They are currently ranked #1 and #3, so it’s much more probable that either would make it over LSU.</p>
<p>And maybe LSU could beat one or the other, but both? IDK. Remember, LSU already lost to Florida at home.</p>
<p>LSU beating both Bama and Florida would be quite a feat, I’m not sure that Texas would want to face that afterward :)</p>
<p>LSU doesn’t have the offense to beat Bama. They’re on the road, Bama plays home after a bye week. I don’t see LSU beating Bama or even play a close game. If that’s the case, Bama will play Florida in the SEC champ. </p>
<p>If Boise St and TCU win out, it looks like Boise is gonna get cut from the BCS bowl again…</p>
<p>
</p>
<p>How is that unfair? The teams from those conferences have to play another tough game against a good opponent. It’s a disadvantage because it’s just another opportunity to lose. Personally, I believe every conference should have a championship game so that everyone has an even chance.</p>
<p>It’s unfair because it gives those conferences an a priori advantage even before the season begins (as a result of the way conferences schedule games)</p>
<p>Also, I’m not saying having the championship game is unfair. I’m saying that intentionally overweighting later games in the computers is unfair, because then the champion of those conferences will <em>always</em> gain an automatic boost because they beat the #2 team from their conference later. On the other hand, the “pac-10 championship game” between #1/#2 could happen any saturday (and probably was USC vs. Oregon this year)</p>
<p>I’m just saying that the computers shouldn’t a priori favor a particular scheduling scheme (round robin, division/championship, etc), and probably shouldn’t weight wins by early/late at all (even though humans do)</p>
<p>LSU probably won’t win, but Bama can’t play poorly in that game and expect to pull out the win, either.</p>
<p>In some cases, whether or not a computer weights later games is a verifiable fact. Looking at the websites for the six BCS computers, here’s what I’ve found:</p>
<ol>
<li>Sagarin’s ratings are called “ELO-CHESS”. The Elo ratings used in chess do weight more recent games more heavily, but Sagarin could be using a modification that eliminates this. He doesn’t state his exact methodology.</li>
<li>Massey explicitly states that recent games are weighted. Massey also states that they take into account the score of the game, which I thought they were prohibited from doing. However, that page is dated August 2000, so the information may be outdated.</li>
<li>Billingsley takes into account time of the game. In addition, Billingsley’s methodology is designed so that, for example, Boise does not get a bonus for Oregon doing well after Boise beat them. They only get a bonus for how well Oregon was ranked at the time of the game. Billingsley also assigns teams a starting rank, so all teams do not begin the season equal.</li>
<li>Colley Matrix and Wolfe both publish their full methodology, and it does not take into account time of the game.</li>
<li>Anderson and Hester don’t seem to publish anything about their methodology.</li>
</ol>
<p>It’s not entirely clear what’s going on (outside of Colley and Wolfe)</p>
<p>Empirical evidence implies Sagarin probably does do weighting as I described. Massey actually uses a slightly different rating system than he describes (as you are right, he is not allowed to use margin-of-victory data) that he doesn’t explicitly say uses weighting (though it is likely)</p>
<p>Billingsley is kind of an odd beast as it is designed to rank teams based on how humans do rankings (only in an unbiased fashion), taking the typical factors that humans take into account. I’m not going to venture and speculate whether these factors are unfair or not, since it’s much less clear if they give specific conferences an a priori advantage (though it’s still possible)</p>
<p>Also, Billingsley actually has some good arguments in defense of his choices of factors.</p>
<p>You know, that kind of came off like I hate the computer rankings :)</p>
<p>I actually think they can be quite valuable, especially considering the muck that is the coaches poll (2 loss ohio state over 2 loss USC? Are you serious people?)</p>
<p>Mostly I made the comment because I follow Sagarin’s computer rankings (not just the ELO_CHESS used by the BCS, but the margin-of-victory based “predictor”, and the strength of schedule as well) and I have been quite impressed so far. It’s just little things like the recent games weighting that bug me about it.</p>
<p>The weighting for recent games makes sense if the point of the rankings is to pick who wins a certain game. I do agree that it probably shouldn’t be in a more general ranking, and not allowing the computers to include margin of victory is just silly.</p>
<p>^Well, how often are first string players taken out of the game to rest or protect them, and then the second string gives up a touchdown or two?</p>
<p>^ Those instances are relatively inconsequential. Generally speaking, the greater the margin of victory, the greater the dominance of the winning team. It should be included somehow.</p>
<p>Well, if margin of victory was considered, then coaches would game plan for it and run up the score whenever possible (admittedly, sometimes they do try to manipulate the human polls like this)</p>
<p>Still, it changes the game fundamentally. Right now, the fundamental goal of a football game is to score more points than the other team. That would change if margin of victory were considered.</p>
<p>Yes, the human polls take margin of victory into account, but I would hope that (say) kicking a meaningless field goal instead taking a knee at the end of a game, doesn’t give an advantage to that team.</p>
<p>
</p>
<p>They deserve to… TCU plays a much tougher schedule than Boise. And what do you mean again?</p>
<p>I doubt it would become any more of a problem than it already is. Teams still frequently run up the score, and if there were any egregious cases they would be severely hurt in terms of PR. It certainly wouldn’t fundamentally change the game.</p>
<p>It’s not like College Football was completely screwed for those few years when the BCS did allow margin of victory. Margin of victory is important to any good objective rating though.</p>
<p>Purdue Beats UMich!!! WOOO! Aww yehhhhh OSU And UMich!</p>