In the Sun Belt, the tiebreaker only restarts after the top team among the tied teams is determined, after which that team is placed in that seed for future comparisons and the rest all start the tiebreak over. It does not restart otherwise. The Sun Belt does not have this fully documented.
I've completed checking my results against official brackets, and other than some date/time issues and one case where I didn't mark an ineligible team ineligible, the Sun Belt and SoCon issues are the only ones that needed fixing.
I neglected to put the Sun Belt issue on BlueSky. Here's the recap:
1. B10 mbb 26.6%
2. SEC m 16.0
3. B12 m 11.4
4. ACC m 10.9
5. BEast m 6.0
6. B10 w 4.1
7. MWC m 3.1
8. MVC m 2.5
9. B12 w 1.8
10. Sun Belt m 1.6
11. AAC m 1.4
12. MAC m 1.3
13. A10 m 1.3
14. CAA m 1.0
15. ACC w 0.9
16. Horizon m 0.8
17. Southern m 0.8
All others: 8.6%
(2/2)
Since Jan 1, around 448k basketball tiebreak scenarios were generated at the site. This is virtually the same as each of the past two seasons have had. By conference: (1/2)
And then it was offline again, but I think it's fixed now
I'm not sure why the site was offline for a while there. The moment I realized it and tried to do something about it, it was back online.
Of the conferences that finished up yesterday, the one mismatch I saw compared to my site was SoCon MBB. I haven't seen a published SoCon tiebreaker in a few years, but apparently they break h2h ties first before going on to other tiebreakers. This is now fixed.
My code is set up to ignore round-robin comparisons that aren't full round robins. A code issue meant this wasn't always working correctly in the code, but I think it's fixed now. This impacts conferences that don't have every team play every other team (e.g. Sun Belt).
My tiebreaker logic has been updated to reflect this.
MBB says: "If a team or teams are separated from the group based on step a, seeding for remaining teams among the group is not determined by head-to-head record vs. the remaining teams, but rather by taking all remaining teams to next tiebreaker"
img.boostsport.ai/boost-cms/20...
A note on the Big 10 tiebreakers:
I discovered today that the MBB and WBB tiebreakers are not identical in the Big 10.
WBB says: "Once a seed is awarded, all remaining tied teams shall restart the tiebreak process at the first criteria".
img.boostsport.ai/boost-cms/20...
*its
When comparing against teams top-down in the seeding, 2-0 is better than 1-0 and 0-1 is better than 0-2 ONLY IF comparing record vs a group of teams, not a single team. I'm not sure why they interpret it that way, but I'm going with that logic.
Based on some inside info, I have updated the Atlantic 10 tiebreaker to reflect how the conference believes it's tiebreaker should be interpreted:
I've now added a large neon banner that shows up near the "Calculate Seeds" button if a past game has the wrong result selected, so the user should now be aware when that's happening. (2/2)
I've removed the auto-refreshing the game selection page on my site used to do. The initial purpose was to fix a problem where some browsers would stubbornly maintain old game selections even after game results updated, so completed games would have wrong results selected. (1/2)
That's exactly what it is. When there are teams with zero games, you get divide-by-zero errors. I'm not going to bother fixing it because I think the effort involved would be way more than it's worth.
9. Sun Belt 0.2% (2.0k; 2.4k ls)
10. Pac 12 0.04% (388; 327 ls)
Last season SEC and Big 12 were 94.4% of all generated scenarios. This season they were only 61.9% of generated scenarios, as the site got more traction among other conferences.
Overall, by conference:
1. SEC 37.7% (331k; 640k last season)
2. ACC 28.6% (251k; 18k ls)
3. Big 12 24.2% (213k; 298k ls)
4. Big 10 4.8% (43k; 23k ls)
5. MAC 1.6% (14k; 6.8k ls)
6. AAC 1.4% (12k; 1.4k ls)
7. MWC 1.1% (9.4k; 2.1k ls)
8. CUSA 0.3% (2.7k; 1.6k ls)
Football season roundup:
For the season (Oct 1 through Nov 30) there were around 878k user scenarios generated. This was around 12% less than last season. SEC was way down but still #1, while ACC was WAY up.
My changes to CUSA didn't get saved earlier today, but it should be good now.
It was pointed out that the CUSA tiebreaker apparently changed this season. The "Modified" date in the pdf properties is 11/16, so it was likely uploaded on or after that date. I've updated my CUSA logic to match.
... for second place where it's not clear that the teams in that tiebreaker are heavily dependent on SS ranking. (6/6)
I could easily change my site to match that, but since I don't have the SportsSource rankings anyway, I think it's less confusing to just leave it as a three-way tie broken by SS ranking. Otherwise, I'm showing a two-team tiebreaker ... (5/ )
...the SportsSource rankings step. For some reason, the MAC only selectes the top team via SS ranking and then heads back to the two-team tiebreaker for the other two teams rather than rank all of them in that step. (4/ )
I've updated my logic to match. Two important thing to note:
1. I do not have access to SportsSource rankings. My site basically just places them alphabetically.
2. In MAC scenarios 9 and 10, there is a three way tie for first place heading into ... (3/ )
The tiebreaker last year was 1. H2H (if full round robin or one team beat all the others), 2. common opponents, 3. common opponents top-down in standings, 4. opponent record, 5. SportsSource rating. They have apparently streamlined it by removing steps 3 and 4. (2/ )
The MAC removed the tiebreaker document from their site (from 2024) and released tiebreaker scenarios. It's clear from the scenarios that they changed the tiebreaker logic in 2025 without publicly releasing it. getsomemaction.com/news/2025/11... (1/ )
I've changed my logic to match that, so now I think my results match the conference. I guess it depends on the definition of "remaining teams" in the tiebreaker procedure.
This should only impact ties involving four or more teams. (4/4)
So what I think is happening is they are not restarting the tiebreaker for the other three after ASU was selected, though it says "After one team has an advantage and is “seeded”, all remaining teams in the multiple-team tiebreaker will repeat the tie-breaking procedure." (3/ )