See here is my problem with your excellent work (not sarcasm). Bowling Green beat Ohio, so they should never be ranked below Ohio, unless the season goes completely into the tank, obviously. However, as you've mentioned, your in-season rankings don't really reflect where the team will finish. Please keep posting these as the season goes along because I'm interested to see how it shakes out.

BTW, are there any other teams in other conferences like that? Teams that lost to another team, but ahead of them in your rankings?

Ask about NIU in a few weeks. NIU is not the NIU of a few years ago.

Little credibility?

I'm using my same OPPA++ rankings that correctly picked the win-loss outcomes of 78% of last week's FBS match-ups (w/o spreads). Jeff Sagarin picked one additional correct outcome to achieve 80%.

Eye test? I had Penn State and OSU separated by one position. They went to double OT. Sagarin had Penn State 35 positions behind OSU.

If I have little credibility, who has a significant amount??

I've discussed at length in other threads precisely why the eye test is invalid ranking methodology in college football.

What does your eye test tell the world about UMass's 7 pt loss at Toledo this weekend? It will tell you everything and nothing at the same time.

Proposed GoCats105 Rule:

I wasn't discussing the transitive property. I was discussing a head to head paradox.

Test:

Assumption: None of these team's seasons have gone completely in the tank (all of the involved teams currently have two (2) or less losses)

LSU beats Ole Miss:

1. LSU (Team A)
2. Ole Miss (Team B)

Auburn beats LSU:

3. Auburn (Team A')
4. LSU (Team B') (Team A)
5. Ole Miss (Team B)

If Ole Miss beats Auburn, there is no possible order to rank the three (3) teams based on head to head logic.

Ole Miss beats Auburn:

1. Auburn (Team A') (Team B''')
2. LSU (Team B') (Team A)
3. Ole Miss (Team B) (Team A''')

All teams are simultaneously Team As and Team Bs in these head to head matchups, therefore ranking these three teams in any order based on your proposed rule is impossible

Btw, I might add Auburn beat LSU 41-7. LSU beat Ole Miss by 4. Just like the eye test, margin of victory tells you everything and nothing at the same time.

Situation,

This is the problem with your rankings, well at least my issue: your ranking is probably going to be very, very accurate when the season is over. It just looks weird to me right now midseason. It would to anyone who watches college football for a living. Shit, Herbstreit's head might explode and Phil Steele would probably have aneurism. I understand what you're saying with the head-to-head paradox, but in reality it doesn't really compute that way only 8-9 games in. Once all 12 games have been played, it should be a little more telling I think.

Which technically should be the point of rankings right? When are they needed? When the season is over and it gets decided who goes where.

The rankings may look weird. But the numbers aren't that weird.

BGSU:

- 0-2 vs teams with winning records
- 0 wins against teams with a 0.500 record
- Beat #64 IU, #89 OHIO, #109 Buffalo, #119 UMass, FCS team
- BG's only qualifying OPPA bonus is the indirect accomplishment from IU beating Mizzou (currently valued at 0.2 pt)
* IU is 1-4 vs teams with winning records (1-1 vs Top 40)
* IU has wins over #26 Mizzou, #106 North Texas, FCS
* Mizzou win currently valued at 1.4 pts; that's why IU is #64, they are one outlier victory away from being nobody, and that's why BG doesn't benefit significantly
- BG gets some credit from beating IU who nabbed a (fluke?) victory over Mizzou, but let's say BG beat #63 Navy (4-4) instead (and that hypothetically Navy would remain 0.500). BG would've earned 1 pt for beating a 0.500 team (like Navy) instead of 0.2 pts from Indiana. Navy is currently 2-3 vs teams with winning records (who would've been the better win? That may be an interesting debate). BG in that scenario then be ranked #80, ahead of OHIO by 9 spots.


OHIO:

- 0-5 vs teams with winning records
- 1 win vs a team with 0.500 record, Akron (currently valued at 1 pt)
- Beat #86 Akron, #106 Idaho, #123 Kent, FCS team

There is no calculation in my ranking system for the paradoxical head to head constraints you described for the reasons I described above. BG has been ranked based on what they've earned, and as weird as it looks, the OHIO team they've beaten earned a more significant victory (so far) than them.

This all said another way:

Yeah BG beat the #89 OHIO team? So what does that tell me? A team ranked #80 or lower has lost 232 times so far this season. A team with a 0.500 record has only lost 52 times. Bottom 30% wins are a dime a dozen.

Indeed, though, your specific BG-Ohio example may well end up part of a paradox situation, since I expect Akron to beat BG. If, on the other hand, BG beats Akron, the situation will resolve in his system by moving BG ahead of Ohio.