An additional point of note/clarification: It does not matter to SOR which games on a schedule were won or lost; only the number of wins/losses.

For example, Alabama's SOR would be the same as it is now if we reversed the results of their FSU and UGA games.

Finally, SOR employs the cdf for the Poisson binomial distribution to combine each of the individual exp win pct values and a team's results into a single chance-of-record-or-better value. That's it!

The expected win pct values incorporate the opponent's up-to-date FPI strength and advantages related to home field, travel distance, and rest.

You can see how each of LSU's opponents' chance to win values vary based on those factors (very little, except for hfa).

Let's use 8-0 Texas A&M as an example.

This is their schedule, so far, and the expected win pct that an average top 25 team would have against each team on the schedule (AVGTOPTM_EXP_WINPCT):

If water is not a drink, why would water with ice in it be a drink?

SOS matters for Army but not Bama is not logically consistent.

Is there a logically-consistent approach where SMU should be in over Army and Bama?

The whole point is that the “just win your games” is dumb when one team is playing a schedule where every team they play is seven points better.

You can’t say “just win your games” to justify SMU over Bama while also having SMU in over, say, Army.

bsky.app/profile/wess...

Would it be easy to generate the same plot for last year?