In college football (American), a ton of people start calling every year for a playoff system to replace the BCS. (Quick primer: usually, a champion of a sports league is determined by a playoff between teams who did the best during the season. Instead, the BCS determines the top two teams by a combination of human and computer polls, essentially creating a 1 game playoff.)
What people don’t think about is that a playoff system is not very likely to crown the best team in the country as champion.
Let’s take an 8-team playoff, for example, and have University of A to be “objectively” the best team in the country. And let’s assume the UA team is so good, it will defeat other elite ( top 8 ) teams 75% of the time. UA will have to win three games in a row, which it has a 42% chance of doing. This speaks highly of UA, but their chances of being recognized as champions are still worse than a coin flip.
The 75% number is higher than can be expected. 60% may be closer to realistic, giving the best team a 21% chance of winning the crown.
If that’s the system people want, then by all means they can clamor for it. As long as they recognize it’s unlikely that the best team in the country will be known as the champions. (College basketball is even worse for determining the #1 team, but everyone likes gambling on March Madness so no one brings it up).
My biggest beefs with the BCS system are:
1. Human pollsters, who cannot have knowledge of every single game played, cannot possibly hope to compare all teams adequately, and have ingrained biases, are given greater weight than the computers.
2. The computer polls are forced not to considered margin of victory in their calculations. This can throw them out of whack, as this story on Jeff Sagarin’s rankings indicates (at one point, North Dakota State was ranked in the top 20). I guess for some reason, the BCS thought that maybe these programmers wouldn’t have developed algorithms that made sure a 52-point victory didn’t mean more than a 35-point victory.
In other words, the BCS has to give computers less weight, because their own rules make computers less accurate in predicting the top two teams.
If it were up to me, I’d eliminate computer poll restrictions, and completely ignore the humans. The only check would be at the end of the year, a council would get together that could veto the computers’ selection if 75% agreed, at which point the human polls would be used to determine the game.
This is mostly off-the-cuff, so blast away with holes in this thinking.