(01-05-2026 06:09 PM)Crayton Wrote: (01-05-2026 04:52 PM)ken d Wrote: (01-05-2026 04:12 PM)Crayton Wrote: The takeaway is if something ‘like’ this could replace all those opaque computer formulas and media polls, making conference and national races more accessible to fans.
Teams’ SOS adjustments will be more-or-less “known” after 4 weeks, allowing fans to track which teams are on track for a CCG or playoff berth. In the ACC, for example, Georgia Tech and Pittsburgh fans will get the benefit of playing a stronger schedule and ‘know’ they’ll win the tiebreaker over Virginia if they finish with the same record.
Not all 10-2 records are created equal, even in the same conference, and right now we all wait until Sunday to find out which teams get postseason berths. Like the NFL and other sports, discrete standings would allow this information to be known beforehand.
How does anyone know what their team's SOS will be after only four weeks? And, even if they could know it, what good would it do them? Few teams have yet played a meaningful number of conference games, and teams that look like sure playoff bets after four weeks have been known to fade in the stretch. Just look at LSU this year.
Instead of Texas fans saying they need to beat A&M, hope for “some” upsets ahead of them, and enlist their coach to stump the media for enhanced standing, Texas fans should be able to say “we need to beat A&M and have 3 of the following 6 teams get upset.”
Can we know SOS after only 4 weeks?
That is a great question. I should run the OP standings at that point in the season and see how valid my presumption is. As you note, teams like LSU (and FSU) can see dramatic dips in fortunes after impressive season starts. There are a few mitigating factors though:
1) most teams ‘are’ as strong at the beginning of the year as they are at the end.
2) the teams which experience upsets after 4 weeks are generally upset by conference-mates, having a balancing affect on other conference teams’ SOS.
3) any unexpected result is ultimately only 1 of 12 games on a team’s schedule and only 1 of 144 when looking at another team’s SOS. Any unexpected results will generally balance.
But the proof is in the pudding. Either tonight or this weekend, I’ll look to see how SOS changes week-to-week. If it changes radically, my purpose may be defeated.
So, does a team's SOS stay the same after only 4 games? A note that I did not mention to ken earlier, but looking at ALL scheduled games (not just opponents played) helps reduce variance week-to-week. Texas doesn't play A&M until the end of the season, but if we want a metric that stays as consistent as possible, we should include A&M in the SOS calculation from the beginning.
Here are the SOS for the SEC each week of the past year. A couple observations:
1) My claim was not perfectly accurate. SOS is 'not' static after the first few weeks. Teams' relative SOS 'do' fluxuate week to week. BUT, most teams relative position only flxuates by a few positions. After the first few weeks we can see Florida, South Carolina, and Arkansas will have the toughest SOS; Ole Miss, Vandy, and A&M have the easiest. We 'do' see what ken is calling out when we look at Alabama: the Tide started with the #2 toughest SOS but saw it drop all the way to #10 as Florida State's early season strength crumbled.
2) There is a slight decline across the entire conference over time. This is mostly because there is no calibration to 'be' consistent from week to week. This offseason while I create a new hand-calculable Power metric, I will consider adding a modification that keeps these SOS numbers more flat over time.
(12-31-2025 11:33 PM)Crayton Wrote: Opportunities / Questions?
1) Because not all teams play 12 games (Army-Navy, CCG teams, mid-season standings), Win% would be more accurate than Win Count. On the other hand, "Wins" are more discrete and fan-accessible, I think. Would "adjusted Win %" still be as accessible as "Adjusted Wins"? Another option is to +/- half wins for CCGs and teams who play less than 12 games.
3) I threw the "-1" into the SOS calculation to balance the fact that I am only looking at the 8 most difficult opponents on a team's schedule. Is it nice to be able to see that Virginia, for example, has a pretty 'mid' schedule close to 0.0, or would it be better to save the extra subtraction step and only rarely come across teams with negative SOS adjustments?
Through the magic of algebra, these two "Opportunities" could be addressed through a single solution.
Rather than a -1, subtracting GamesOver10 divided by 1.5 would fix both. Most teams will play 2 games over 10. This amounts to -1.33, even closer to the actual -1.31. Teams that play 11 or 13 games will have a bonus/penalty of 0.66. This is roughly equivalent to a 10-3 team. 11-2 and 12-1 teams will experience a slight bonus; 8-3 and 9-4 teams will experience a slight penalty.
With the anticipated reworking of the SRS/Power rating, the exact factors are likely to change. Thank you everyone for asking good questions so that I can identify what needs to be addressed. If I like the product, I may have a thread tracking it next year.
Another "Opportunity" may be that Loss Count is more often what fans use to mentally sort teams rather than Win Count. Perhaps SOS-Losses would be more intuitive than SOS+Wins? I probably wouldn't use the -1 rule I just talked about (or its replacement) so that I can keep SOS-Losses in the positive for the Top 25/40 teams.
EDIT: yes, SOS-Losses would also allow a simpler SOS formula:
SOS = (AvgTop8SRS + GamesScheduled) / 8
Rather simple? It’d yield the same values as the OP, just -9.5. This means the Top 25 would still be > 0, which is aesthetically pleasing.