(12-08-2018 10:45 PM)Hokie Mark Wrote: (12-08-2018 10:06 PM)BruceMcF Wrote: Last year's games don't count to this year's tie breakers, so that doesn't help. It helps conference cohesion, but accidents of scheduling could eg, put two teams in a CCG that have each lost to 3 and 4, but 1 & 2 happened to be in the creampuff richer cycle of their schedule.
I've heard that argument posed before, but it's a false argument. Consider this:
* For 1 & 2 to be at the top despite losing to 3 & 4, they must've won the remaining games (in the case of the Big Ten than means 8-1 in conference) AND they didn't play each other - thus eliminating 4 teams (1,2,3 and 4).
No, eliminating 3 teams, if you are counting "not playing yourself" as eliminating a team, eliminating 2 teams if counting more normally. You skip
4 teams playing 9 games in a 14 team division ...
Quote: * for 3 & 4 to be behind 1 & 2, they had to have lost at least 2 games each to teams other than 1 & 2, since you said they beat them - so for the sake of argument we'll say they lost to teams 5, 6, 7 and 8.
They were the teams that HAD the tougher schedule, so 4 lost to 3. They both played 5, 6, and 7, 3 lost to 5 and 7, 4 lost to 7 in addition to 3.
By the original assumption, these three include the other two skipped by 1 & 2, so 5 lost to 1, 6 lost to 2, and 7 lost to 3, 5, and 6.
Quote: * for teams 5 - 8 to be behind 3&4 they must've lost 3 or more games each.
Not true, not in a 9 games out of 13 opponents schedule. 8 could be one of the cupcakes that never beat anybody 1-7, and 5&6 could both be two loss schools in those spots by head to head.
Quote: NOW...
* which is really better: teams 1&2 who only lost 1 game each (to 3 & 4 respectively), OR teams 3 & 4 who lost 2 games each - and lost them to teams which themselves have 3 or more losses?
In this case, definitely 3&4, since they were undefeated OOC, including to ranked schools, while 2 was 2-1 winning both its buy games and losing to an unranked A5 schools, while 1 was 1-2 OOC, losing one of its buy games as well as losing to an unranked A5 school.
It's a lot
easier to paint scenarios where in-conference SOS differences put the strongest school in 3rd place on an in-conference-game-only basis (and this is the Big Ten, the odds that it
will be in-conference-only are very, very short odds), but skipping 4 games does make it feasible for both of the strongest schools to be 3 and 4th in the conference ladder due to a couple of unfortunate events in a couple of games.
Quote: See the fallacy? We might say that 1 & 2 played an "cream puff" schedule because their opponents had more total losses, but the fact is they beat those teams, whereas 3&4 did not...
3&4 don't necessarily have more total losses, just one more conference loss. In this scenario, the records could well be #1 9-3 (8-1), #2 10-2 (8-1), #3 10-2 (7-2), #4 10-2 (7-2), #5 9-3 (7-2), #6 9-3 (6-3){+}, #7 8-4 (6-3).
({+ Team #6 slipped up against one of the bottom 7 schools.})
Quote: BOTTOM LINE: There is nothing conceptually wrong with simply taking the 2 teams with the best records, assuming equal and significant number of pseudo-random conference games.
The more in-conference opponents are skipped, the more in-conference SOS makes that questionable. With only 2 or 3 games skipped, it's less of an issue. With 4 games skipped, the possibility of the best team getting left out starts to be noticeable. When it hits 6 games skipped, the possibility of missing the best team because of in-conference strength of record will start to be quite substantial.
The "BOTTOM LINE" is implicitly assuming that the "significant number" is a yes/no question, like flipping a light switch ... when in reality it is a continuum. 8/11 is more significant than 9/13, which is more significant than 9/15.