[BostonCard]

[quote author=Boston Card link=topic=15599.msg170309#msg170309 date=1473207453]
It plays off my preseason ballot, which was a projection of how everything will shake out, and makes adjustments based on results where applicable.    There have been several major changes, but the ballot cannot be wholly results-oriented because not every team played a meaningful opponent. . . .It will be several weeks before we have enough quality games to compile a ballot that's completely based on results.

So...

How is this different than what every other pollster does?  Everybody's polls are pretty much some blend of their priors and the results on the field?

If that is really what he does, then why did Notre Dame go from #10 in his pre-season poll to #6 after week 1 despite losing to a team he had unranked in his pre-season poll.  There is no universe in which that is some combination of his priors (preseason #10) and the results (a loss to a pre-season unranked team) result in the team's ranking improving.

In contrast, Stanford went from a pre-season #8 on his ballot to #13 on his ballot, despite beating a power 5 team.  I mean, I could understand if he said something like "we know no more about Stanford this week than last, because I expect the team they beat to be pretty bad this year", not moving Stanford up in its ranking (even though higher ranked teams ahead of Stanford such as LSU and Oklahoma).  And it's not just a Stanford thing.  He dropped Michigan from 7 to 10 even though they crushed Hawaii.  Again, I get if you want to discount that win as more or less not telling us much because a) Hawaii is not that good and b) Hawaii played a week before in Australia and you expect them to be jet lagged as hell.  So fine, you leave them where they were.  But how does that game tell us that Michigan was any worse than what you thought before it was played.

The man says the right things about his ranking, he just doesn't adhere to what he says.

BC

[BostonCard]

[quote author=Boston Card link=topic=15599.msg170309#msg170309 date=1473207453]
It plays off my preseason ballot, which was a projection of how everything will shake out, and makes adjustments based on results where applicable.    There have been several major changes, but the ballot cannot be wholly results-oriented because not every team played a meaningful opponent. . . .It will be several weeks before we have enough quality games to compile a ballot that's completely based on results.

So...

How is this different than what every other pollster does?  Everybody's polls are pretty much some blend of their priors and the results on the field?

If that is really what he does, then why did Notre Dame go from #10 in his pre-season poll to #6 after week 1 despite losing to a team he had unranked in his pre-season poll.  There is no universe in which that is some combination of his priors (preseason #10) and the results (a loss to a pre-season unranked team) result in the team's ranking improving.

In contrast, Stanford went from a pre-season #8 on his ballot to #13 on his ballot, despite beating a power 5 team.  I mean, I could understand if he said something like "we know no more about Stanford this week than last, because I expect the team they beat to be pretty bad this year", not moving Stanford up in its ranking (even though higher ranked teams ahead of Stanford such as LSU and Oklahoma).  And it's not just a Stanford thing.  He dropped Michigan from 7 to 10 even though they crushed Hawaii.  Again, I get if you want to discount that win as more or less not telling us much because a) Hawaii is not that good and b) Hawaii played a week before in Australia and you expect them to be jet lagged as hell.  So fine, you leave them where they were.  But how does that game tell us that Michigan was any worse than what you thought before it was played.

The man says the right things about his ranking, he just doesn't adhere to what he says.

BC