[BostonCard]

[quote author=Boston Card link=topic=15599.msg170261#msg170261 date=1473189671]
Indeed.  We might call him John "Circular Reasoning" Wilner:

I am a little torn on Mr. Circular Reasoning.

I like the fact that he tries to bring logic to his choices, doesn't stick slavishly to what he did the week before, and wants to primarily base his choices on results. But, he doesn't seem to quite understand what he is doing, opening himself up to both circular reasoning and arbitrary data selection.

The key problem right now is that he wants his ballot to be completely results based despite the fact it is week #1, so only a few teams have played reasonable opponents and even among the others we don't really know how good any team is yet.

So there are all sorts of odd things in there.
I'll choose the one affecting Stanford for now.

Once we account for the teams with marquee victories, the guesswork truly begins with Ohio State, Michigan, TCU, Washington and Stanford in the No. 9 – 13 spots — all teams that I believe are top shelf but didn’t record substantive results.

Well, one of those teams played a Power 5 opponent that was in a bowl game last year, and generally expected to be better this year. Naturally that team came in at the end of this list. Another struggled with an FCS team; they came in mid-group.

[/quote]

Here's the thing that makes no sense.  Although I don't think it makes sense to throw away your priors (pre-season expectations) after only a game, I can respect trying to make the season rankings based on results and willing to adjust quickly to new data (and WIlner is willing to do this).  The issue is that is not what he is doing.  If you are going to make your rankings based on results and not based on your pre-season expectations, then Stanford beating Kansas State is functionally the same as Texas beating Notre Dame since both teams beat power 5 opponents in the first week.  Yet Wilner discounts Stanford's victory over KState (which is fine) and gives Texas credit for beating a "good" Notre Dame.  But that means that his priors (KState is a middle-of-the-road Big12 team while Notre Dame is a top-10 team) are not being thrown away.  In other words, Wilner is claiming to have one methodology, but his actual methodology is not what he says it is.

BC

[BostonCard]

[quote author=Boston Card link=topic=15599.msg170261#msg170261 date=1473189671]
Indeed.  We might call him John "Circular Reasoning" Wilner:

I am a little torn on Mr. Circular Reasoning.

I like the fact that he tries to bring logic to his choices, doesn't stick slavishly to what he did the week before, and wants to primarily base his choices on results. But, he doesn't seem to quite understand what he is doing, opening himself up to both circular reasoning and arbitrary data selection.

The key problem right now is that he wants his ballot to be completely results based despite the fact it is week #1, so only a few teams have played reasonable opponents and even among the others we don't really know how good any team is yet.

So there are all sorts of odd things in there.
I'll choose the one affecting Stanford for now.

Once we account for the teams with marquee victories, the guesswork truly begins with Ohio State, Michigan, TCU, Washington and Stanford in the No. 9 – 13 spots — all teams that I believe are top shelf but didn’t record substantive results.

Well, one of those teams played a Power 5 opponent that was in a bowl game last year, and generally expected to be better this year. Naturally that team came in at the end of this list. Another struggled with an FCS team; they came in mid-group.

[/quote]

Here's the thing that makes no sense.  Although I don't think it makes sense to throw away your priors (pre-season expectations) after only a game, I can respect trying to make the season rankings based on results and willing to adjust quickly to new data (and WIlner is willing to do this).  The issue is that is not what he is doing.  If you are going to make your rankings based on results and not based on your pre-season expectations, then Stanford beating Kansas State is functionally the same as Texas beating Notre Dame since both teams beat power 5 opponents in the first week.  Yet Wilner discounts Stanford's victory over KState (which is fine) and gives Texas credit for beating a "good" Notre Dame.  But that means that his priors (KState is a middle-of-the-road Big12 team while Notre Dame is a top-10 team) are not being thrown away.  In other words, Wilner is claiming to have one methodology, but his actual methodology is not what he says it is.

BC