Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 22, 2006 0:25:45 GMT -5
VT's master chartician, gorf, came up with the idea of using the VT Pick-the-Winners Contest as a reference for measuring the relative accuracy of the AVCA and Rich Kern polls and Pablo as tools for predicting winners of matches and compiled the following table. While the same data is not always available for all three because neither team is ranked by AVCA and/or RK for some of the matches, nonetheless it will be interesting to follow. Note the links to each teams webpage.
|
|
|
|
Post by Gorf on Sept 22, 2006 0:47:10 GMT -5
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 22, 2006 0:50:45 GMT -5
So, in head-to-head matches, so far, the talley is:
AVCA 2-1
Rich Kern 2-1
Pablo 1-2
|
|
|
|
Post by chipNdink on Sept 22, 2006 0:57:44 GMT -5
What about the previous weeks?
|
|
|
|
Post by Gorf on Sept 22, 2006 0:58:19 GMT -5
Pablo is 1-3, I included the current "record" for each in the table header (ie, "AVCA (2-1)")
|
|
|
|
Post by Gorf on Sept 22, 2006 1:00:08 GMT -5
What about the previous weeks? What previous weeks? I don't recall any previous weeks!  |
|
|
|
Post by Alberta on Sept 22, 2006 1:37:55 GMT -5
What about the previous weeks? What previous weeks? I don't recall any previous weeks! Funny how selective your memory is. |
|
|
|
Post by Ye Olde Dawg on Sept 22, 2006 2:22:19 GMT -5
Pablo can actually assign a probability to the match, not just pick a winner. For example, with Stanford at Washington tomorrow: -- Stanford's rating is 8600 -- Washington's is 8235 + 347 (HCA) = 8582 So, playing at Washington these teams are just 18 points apart. Stanford is favored, but it's very close to a toss-up -- like, 51%. Pablo is the only ranking that can predict the chance that it will pick the wrong team. So with all that it seems odd that we're giving Pablo credit or blame for picking this match right or getting it wrong. BTW, the Oregon match was similar, with Oregon adjusted to 77 pts over UCLA. But we can ding Pablo for that one because Oregon got swept.  |
|
|
|
Post by IdahoBoy on Sept 22, 2006 2:24:44 GMT -5
Pablo can actually assign a probability to the match, not just pick a winner. For example, with Stanford at Washington tomorrow: -- Stanford's rating is 8600 -- Washington's is 8235 + 347 (HCA) = 8582 So, playing at Washington these teams are just 18 points apart. Stanford is favored, but it's very close to a toss-up -- like, 51%. Pablo is the only ranking that can predict the chance that it will pick the wrong team. So with all that it seems odd that we're giving Pablo credit or blame for picking this match right or getting it wrong. BTW, the Oregon match was similar, with Oregon adjusted to 77 pts over UCLA. But we can ding Pablo for that one because Oregon got swept. C'mon... it's not horseshoes nor hand grenades. Who cares about close? Teams don't each go home with a "V".... It's a W or an L. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 22, 2006 8:38:07 GMT -5
This doesn't work since RK and the AVCA do not factor HCA.
Apples to oranges...
|
|
|
|
Post by The Bofa on the Sofa on Sept 22, 2006 8:39:57 GMT -5
VT's master chartician, gorf, came up with the idea of using the VT Pick-the-Winners Contest as a reference for measuring the relative accuracy of the AVCA and Rich Kern polls and Pablo as tools for predicting winners of matches and compiled the following table. While the same data is not always available for all three because neither team is ranked by AVCA and/or RK for some of the matches, nonetheless it will be interesting to follow. Note the links to each teams webpage. I did this exercise with the AVCA, BCR, and Pablo back in 2003 or so. When looking only matches that could be compared, Pablo finished about 7 games ahead of the AVCA at the end of the season, and 3 games ahead of BCR. Some weeks Pablo was better, some weeks BCR was better, some weeks the AVCA was better. Pablo was the best cumulatively. While it's worth doing, week by week analysis isn't very useful, because there aren't enough matches. Knock yourselves out, here, guys, but do it right. Go ahead and record the results. But don't put any significance in one week's results. Look at the results when you have 150 matches that you can compare. Then you can learn something. |
|
|
|
Post by The Bofa on the Sofa on Sept 22, 2006 8:41:29 GMT -5
This doesn't work since RK and the AVCA do not factor HCA. Apples to oranges... I wondered about that when I did my comparison. Perhaps the HCA should be 2 or 3 spots in the AVCA rankings? I never applied it, and it's only going to be an issue there when teams are very close in the rankings |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 22, 2006 8:46:58 GMT -5
AVCA and Kern both have points, so that's a better route. But who knows what the HCA would be? I just know that if you asked the voters in those polls, HCA would make a difference in their relative rankings of opponents.
OK. Maybe not _all_ of them.
All I know is that I went in and switched to San Jose State as the winner over Nevada because I looked at Pablo. I'm not doing that again for these picks.
|
|
|
|
Post by The Bofa on the Sofa on Sept 22, 2006 8:58:16 GMT -5
You know as well as I do that over the long term, you would do better to just go with Pablo.
Publius is the only one for whom that can't be said. At least last year.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 22, 2006 9:29:21 GMT -5
Nope. I'm kicking Pablo's behind this year.
(As long as we stick with the top 50 or so teams, that is.)
|
|
