(02-10-2026 09:53 AM)JSchmack Wrote: (02-10-2026 08:31 AM)thrill_house Wrote: There's websites out there that track the accuracy of these ranking services
Shame we don't have computer rankings for college football, but that's what happens when the vast majority of college football (and sports) fans have an IQ that is too low to analyze teams any further than record and performance in a couple of games.
Sure, but even that, it's like double or triple counting factors. It's like, "Oh, you have to give like 3.5 points leeway for home field" but all the math you did on results to make your formula had a home and road team like 98% of the time.
And of course, the most massively counted thing is "OOC by conference," because of the nature of conference play (67% or more of your games are conference teams, and every conference is .500 against itself).
Predictive metrics are dumb. Except for "trying to make better sports gambling bets."
But "NCAA Tournament Selection" should reject the premise of them and use an actual results-based system to measure resumes. Something less SOS based where only losing 14 games to strong competition isn't a GOOD thing.
Miami Ohio is a texbook example. If you have a win percentage over .900, you deserve to be in. Regardless of who it's against. I have a pretty good idea that someone 10th in the P5 league probably isn't winning the NCAA Tournament. But does anyone really know how good Miami Ohio is? Let's find out.
Changing the philosophy of Selection Committees to THAT would go a long way towards making college sports better. It would put more value on the regular season, increase the recruiting pitches of more schools and be a self-fulfilling prophesy in the right direction instead of the wrong direction.
The NCAA should use a Swiss Pairs format to determine the teams.
Teams would play two games each month: November, December, January, February, March. The two games would be in the form of a four-team mini-tournament played over two nights: Monday-Wednesday, Tuesday-Thursday, Wednesday-Friday, Thursday-Saturday, Friday-Sunday, at one of the four schools.
Everybody starts with zero wins. A computer draws the tournaments, with emphasis on travel distance and inter-conference play:
For each school determine 10? 20? nearest schools in different conference, and pick one with the greatest total distance to the other schools. This would probably be Hawai'i. Its 20 nearest schools not in the Big West, would include other California schools in the Big Ten, ACC, Mountain West, WAC, Big Sky, WCC and extending out into Arizona, Nevada, Oregon, as necessary.
Randomly choose one of the schools, (e.g., UCLA). Find the 20 nearest schools, excluding Big Ten and Big West schools (so no USC or Oregon). Pick the third school (e.g., Sacramento State), and a fourth (e.g. Pepperdine).
So our first tournament is: Hawai'i-UCLA-Sacramento State-Pepperdine. Choose the pairings randomly (e.g. Hawai'i-Sacramento State, and UCLA-Pepperdine). These games might be played Wednesday, with the winners and losers playing on Friday. So one team will have two wins, two teams will have one win, and one will have zero wins.
Repeat across the country, and 92 will have two wins, 184 one win, 92 zero wins. So the December round will match up the 92 teams with two wins, and so on.
We can keep track of conference records, and award conference bonuses (4 conferences get two extra, 8 get one extra, 8 gets 1/2 extra (play-in), and 12 get 1/4 extra (double play-in). Conferences choose how to awards these.