- From: George W Harris <gharrus@xxxxxxxxxxxxxx>
- Date: Tue, 18 Mar 2008 03:16:08 -0400
On Mon, 17 Mar 2008 21:29:53 -0700 (PDT), i_like_sockeye@xxxxxxxxxxxxxxx
:On Mar 17, 9:25 pm, George W Harris <ghar...@xxxxxxxxxxxxxx> wrote:
:> On Mon, 17 Mar 2008 11:27:02 -0700 (PDT), i_like_sock...@xxxxxxxxxxxxxxx
:> :On Mar 17, 1:08 pm, Donnie Barnes <djbSPAMSU...@xxxxxxxxxxxxxxxx>
:> :> I don't think it was as big as the Kansas/UNC situation this year, but
:> :> didn't Illinois murderize a lot more teams than UNC did in 2005, too?
:> :> I are too lazy to look it up.
:> :Clearly, MOV needs to be used to make predictions, but I don't think
:> :anyone knows exactly how it should be used to get the most reliable
:> :predictions. This is why I'm a little hesitant to jump completely on
:> :the Kansas bandwagon (there could be some noise in that MOV signal).
:> Well, it's using a much smaller sample size, but in my
:> ACC power ratings I've found that weighting games between
:> teams that are either close in rating or that were close in
:> score (or upsets) more heavily leads to more accurate
:> I'm not an actor, but I play one on TV!
:> George W. Harris For actual email address, replace each 'u' with an 'i'
:Have you tried a logistic model? For example, one that might view a
:3pt win as "slightly better than" a 1pt win, a 5 pt win as "better
:than" a 3 pt win, a 7 pt win as "a lot better than" a 5 pt win, a 9 pt
:win as "a lot better than" a 7pt win, an 11 pt win as "better than" a
:9 pt win, a 13 pt win as "slightly better than" an 11 pt win.
:Something a little more sophisticated than that, but something like
Yes, I did something like that, if I understand you
correctly (you're talking about diminishing returns, right?).
It wasn't any more accurate than a straight MOV method
that I could tell (I used it for a few years).
"Intelligence is too complex to capture in a single number." -Alfred Binet
George W. Harris For actual email address, replace each 'u' with an 'i'
- Re: YES!!!
- From: i_like_sockeye
- Re: YES!!!
- Prev by Date: Gauging NCAA Tournament Performance By Conference Relative To Seed
- Next by Date: NIT vs. CBI
- Previous by thread: Re: YES!!!
- Next by thread: Re: YES!!!