What people don't understand about computer models is that they are probabilistic averages. For instance, my rag tag Bull**** model has Mercer losing to Furman 27-26. Will that happen? probably not. But it tells you that the average expectation is a very close Furman win. It's still a coin flip and in some ways is worthless (but so are all forecasts for close games). The averages tell you what you can expect or if, given outcomes, whether or not a team over/under-performs or if their strong/weak schedule has led the data to indicate that a team is better/worse than they are. You won't know until the dataset is complete...at the end of the season.
And football, particularly college football is different, because 99% of teams don't ever rematch, so it's a much less stable model for outcomes. Contrast that with the NBA or NFL or college basketball, which rematches everyone in conference at least once a season, and the information for the computer ranking is less reliable for college football.
And the Socon is weird because Wofford has played most of the top 6 in the conference, Mercer has played most everyone and has the best sample but of the other mid-tier of the conference, Furman has played Wofford and the three bottom tier and Samford has only played 3 games in conference.
So any computer rankings at this point are going to be flawed or unreliable. I'll be more confident about them (any of them) after this week because you'll have a better sample of Furman and Samford.
I'm only confident in saying Wofford and Western Carolina will probably make the playoffs; everything else is up in the air and margin of error
Bookmarks