No, that's not what I'm doing. You just automatically assume that's the only thing others do.
They get credit for winning games. But when trying to determine which conferences are better, it does no good to look at intra-conference games. You have to see how they did against outside competition to really know anything. But the SEC chooses to play precious few games OOC to protect themselves.
There is no measure of what has actually taken place on the field this year that says the SEC is better than the Pac-12. The Pac-12 was 8-3 against other Power 5 teams this year; the SEC was 5-6. The Pac-12 is currently 4-1 in bowl games; the SEC is 4-3. The computers are nice in that they give you a way to compare teams that largely haven't played each other. But once you have actual on-field performance to look at, the computers don't mean much.
How did the Sagarin predictor fare in telling us what would happen in the Peach Bowl or the Orange Bowl? Or do we not care what actually happened in the game because the computer said it shouldn't have happened that way?
This is like beating my brain against the wall. You have, again, picked "Power 5 teams" as if that means something (ignores the fact that some P5 teams are horrible while some non-P5 teams are good) while also completely ignores the individual matchups themselves, and pointed to records in 5 games, with no margin of victory figures included in your analysis whatsoever, and attempted to draw a conclusion from it
Meanwhile, Sagarin draws the data from every matchup between every team, and as a part of its very analysis calculates what team ratings best fit the results on the field that have actually occurred, in every game, using the actual game score data - yes, margin of victory.
You also don't seem to understand variance - Ohio State outperformed their Sagarin/Vegas expectation in the B1GCCG by 63 points - something that you'd actually expect to occur once out of every 100 or so games anyway. That's why I rightfully put virtually no stock into any one individual game outcome - it could be meaningful or meaningless. The only way to know is to look at it as part of a larger data set. Like, a whole entire season for example.
"But once you have actual on-field performance to look at, computers don't mean much" is a total non sequitur. It does not make sense as a statement. The computers draw their inferences directly from the on-field performances.
Anything other obvious things that I need to point out for the thousandth time?