Actual SEC Tiebreaker Breakdown Thread

I zoned out typing that while reading something.

quote:

---

I think in this case LSU and Georgia would geaux.

Sucks because Texas will still get in the playoff and not have to take any additional damage.

---

Georgia and LSU do go in that scenario.

quote:

---

A) Head-to-head competition among the tied teams: 1. If there is a complete round robin (which there isn't), meaning all tied teams played each other: compare records to eliminate, promote, or move on to other tiebreaker rules.

---

ok so this is the part i’m questioning.

has the SEC come out and explained it needs to be a round robin?


the article on their website hasn’t said anything about that.

I am just trying to see if theSEC has clarified and if so where can I read it.

quote:

---

ok so this is the part i’m questioning.

has the SEC come out and explained it needs to be a round robin?


the article on their website hasn’t said anything about that.

I am just trying to see if theSEC has clarified and if so where can I read it.

---

It is in that link I have at the top of the thread from secsports.com website. It is explicitly laid out for us.

quote:

---

Cumulative Conference Wins / Total Cumulative Conference Games Played

---

But how did you calculate those numbers? What were your assumptions on records of all conference opponents at the end of the season?

quote:

---

But how did you calculate those numbers? What were your assumptions on records of all conference opponents?

---

The assumptions were all favorites win their games outside of the tied teams.

There's really no point in doing this yet. We have a lot of conference games between now and December.

I was just trying to explain to people how it works. There is a lot of people confusing the details in other threads I read.

quote:

---

I don't really understand how the cumulative conference winning percentage is calculated and all the assumptions that go into that.

How can you possibly predict the winning percentage of all the teams that Texas, Georgia, and Tennessee have played right now? Seems like that could vary a lot depending on what happened these last few weeks.

---

Long post ahead, but since you asked and I’m a nerd…

TL;DR - math. If you don’t care about the rest stop reading now.

———————————

First off, the only potential 3-way ties scenarios where opponent win% matters for LSU at this point are:
- LSU, UGA, and Texas
- LSU, UGA, and Texas A&M

The other potential 3-way ties involving LSU are:
- LSU, Tennessee, and Texas
- LSU, Tennessee, and Texas A&M
These two don’t go to opponent win% for SECCG selection because Tennessee lost to Arkansas, who all other tied teams would have beaten.

For the 3-way ties involving UGA, we only care about scenarios where UGA wins the first (3-way) tiebreaker outright. If the third team (A&M or Texas) wins the 3-way tiebreaker, we would have the 2-way tiebreaker over UGA by beating Bama. If UGA and LSU’s opponent win% are the same and both are at the top, both are in the SECCG anyway.

So all we care about are scenarios where UGA’s opponent win% is higher than LSU’s.

Based on 22 remaining SEC games, there are 4.2 million potential combinations of outcomes. That’s a little better than last week when there were 134 million potential combinations, but still way too many to run every single one.

Luckily we can narrow this down a lot.
- We are only worried about the games involving opponents of LSU and/or UGA. Those 13 teams play 22 games. Well shite, that actually didn’t narrow it down at all.
- Since we are considering a 3-way tie involving UGA and LSU, that means both teams have to win out. Now we are down to 15 games, or 32,768 outcome combinations.
- We don’t care about games where both teams play both UGA and LSU, since the outcomes of those games as a win and loss to both LSU’s and UGA’s opponent records regardless of who wins. That gets you down to 14 games, 16,384 outcomes.
- We also don’t care about games where both teams play LSU (but not UGA) or where both teams play UGA (but not LSU). That gets you to 12 games, or 4,096 outcome combinations.

Those 12 games are:
- Oklahoma @ Missouri
- Florida @ Texas
- Texas @ Arkansas
- Missouri @ South Carolina
- Texas A&M @ Auburn
- Missouri @ Mississippi State
- Alabama @ Oklahoma
- Mississippi State @ Ole Miss
- Auburn @ Alabama
- Arkansas @ Missouri
- Texas @ Texas A&M
- Tennessee @ Vanderbilt

These 12 games are the only games that can shift opponent win% between UGA and LSU, in a hypothetical 3-way tie between LSU, UGA, and Texas (or Texas A&M). Now you can cheat by establishing “chalk” for each game (denoted in bold above). Using this chalk makes LSU’s opponents 30-34 (0.4688) and UGA’s opponents 27-37 (0.4219).

At this point the only other outcomes we care about are those that either help UGA’s opponents win% or hurt LSU’s (or both). For example - if Oklahoma beats Alabama, that makes UGA’s opponents win% worse (they play Alabama) but doesn’t affect LSU’s at all (we play both). I don’t care about this outcome because the chalk already has LSU with a higher opponent win% than UGA. No need to look at how much higher it can go. The question is how UGA’s opponents win% can be higher than LSU’s.

The only outcomes other than chalk that can hurt LSU or help UGA are:

- Mississippi State beats Missouri (helps UGA)
- Missouri beats South Carolina (hurts LSU)
- Mississippi State beats Ole Miss (hurts LSU)
- Auburn beats Alabama (hurts LSU)
- Auburn beats Texas A&M (both)

The chalk has LSU’s opponent win% 3 games better than UGA’s. And if LSU + UGA tie, both still go to Atlanta. So for LSU to get left out UGA would need to make up 4 opponent wins. Texas A&M @ Auburn is potentially worth 2 wins (+1 for UGA and -1 for LSU) while the rest are worth 1 win (either +1 for UGA or -1 for LSU).

So in order for LSU to get left out in a 3-way tie, you would need the “both” upset plus two of the others -or- you would need all four of the “helps UGA” and “hurts LSU” upsets.

In all other 3-way tie scenarios, LSU is in Atlanta due to one of the following:
- Highest opponent win%
- Record vs. common opponents over Tennessee
- Record vs. common opponents over UGA

… that was a lot. Oops.