This is what helps teams that play in conference title games, such as Florida in 1996 (pre-BCS, but the same principle).

Of course, teams that lose in November/December also need a good deal of help -- e.g., Florida 1996, Nebraska 2001, OSU/LSU 2007.

The one team that lost late, got no help and still made the BCS title game was Oklahoma in 2003. If I recall correctly, they were saved by the computers.

I looked at this about a year ago. I only used one year of data (2008), and I analyzed it on a more global scale rather than purely from the perspective of a BCS champion. You can see the results at the link below.

http://clashmoremike.com/2008/12/pre-season-polls-what-happened-to-the-bulldogs/

I plan on doing something similar this year, provided there is time.

My argument to MD was regarding the scheduling of Maryland, that contending teams are less likely to schedule major opponents late in the season--November specifically--except for cases of historical rivalries(Florida-Miami-FSU, ND-Southern Cal, etc.). I pointed to examples from Texas and Florida's schedules next season where they are playing Florida Atlantic and App. State respectively in their open weeks in November. The key stat, in my opinion, is the number of teams who lost games in November that cost them a shot at the title and what those losses were. LSU in 07 lost two games but made it to the title--one in November--but both losses were in triple overtime, so I think that voter forgiveness does play a factor. I'll look up these stats though when I get a free minute or 60.

Although it is probably a smaller factor, I think voters are more forgiving of a relatively young team that loses early vs a more experienced team. The theory would be that young teams will improve much more over the course of the season than experienced teams and thus an early loss may be more forgivable.

I'd be curious to see if the data back this up or not.

Let's look at this year's ND team. If ND opened with a loss to Purdue, we would immediately be viewed as being a non-contender and subsequent wins over Michigan, MSU, etc., would probably not be enough to change that perception unless we ran the table and then beat a strong SC team to close the season. Conversely, a close loss to MSU after a couple of strong wins might be viewed as a setback, but not a fatal one, if we ran the table.

Similarly, a string of big wins through the first 2/3 of the season, with a 2-point loss against a Utah team that was itself undefeated at the time (but which subsequently lost) from which we rebounded to wax Army and a good SC team while all other remaining contenders wound up losing at least one would maybe be enough to get us in. Conversely, running the table and then losing to a mediocre SC team in the last game of the year would almost certainly relegate us to being also-rans.

So it's when you lose, to whom you lose, and what you have left after the loss to make up for it.

And that's not even taking into account what else happens the week you lose. Consider that other teams lost on the day Michigan lost to Applachian State, but we really can't remember who those teams were -- so their losses wouldn't hurt them as much in the overall scheme of things. Similarly, I can recall weekends in November where it seemed like each set of games throughout the day (noon ET, 3:30 ET, 7:30-8:00 ET) provided greater upsets, with the upshot being that the team that lost early in the day was overshadowed by day's end.

Moco and I were discussing timing of non-conference opponents (particularly, good ones)

Moco believes in the often held position that late losses are far more costly than early losses. However, I pointed out that due to conference title games, most contenders have the ability to recover from a late non-conference loss and since mostly undefeated teams make the game, it's not that big of a deal.

For evidence I collected the losses of every BCS title team over its decade long headache.

Of the teams that made the game there were 12 losses prior to the game itself

3 in September
5 in October
3 in November
1 in December

Moco's response questioned whether or not any of the teams that lost late got in over teams that lost early. In at least one case, they did. OU was in the game after losing in December over USC that lost in September, both teams finishing with 1-loss.

I think what's happened is there's a bias to remember teams that fumble the season late compared to teams that fumbled the season early. In mid-October we choose the contenders, the teams still in it, and then we watch them fall 1 by 1. However, when we make that list we disregard all the teams that lost in September until everyone else loses and all of a sudden they become relevant again. In the end though, it appears that there isn't a selection bias towards losing in September teams there's just a bias towards thinking that. In the end, the polls/computers don't seem to care nearly as much as everyone assumes.

I know of at least one reason my stat might be misleading and that's if there's a LOT more losses by contenders in November because they're playing conference and rivalry games instead of playing "buy" games. However, looking through the schedules that doesn't appear to be the case but I dont have the data necessary to prove that.

My conclusion, according to the data collected, is that everyone is exaggerating the value of a "timely" loss. I did however see lots of OT losses and close losses by these teams.

Just my thoughts on an often held belief/misconception