|
Post by vbbetterthanbb on Dec 22, 2014 17:48:34 GMT -5
...Pablo nailed it with Washington BYU and Penn St... Pablo has always been the leader among the predictive indexes, but it was ridiculous in 2014. Care to share any details? (and hopefully in a layman manner that most of us can half way understand NCAA VT Pablo Richkern AVCA RPI 2014 79.4% 87.3% 80.3% 75.0% 71.4% 2013 79.4% 73.0% 78.7% 76.7% 77.8% 2012 76.2% 76.2% 74.6% 73.3% 73.0% PTW-Wkly 2014 75.1% 76.6% 73.4% 73.3% 71.5% 2013 71.5% 68.8% 68.2% 66.7% 66.1% 2012 69.3% 68.4% 68.7% 68.8% 66.1%
Did 2014 turn out to be more predictable than previous years?
|
|
|
Post by gogophers on Dec 22, 2014 17:52:13 GMT -5
While you're asking: what the heck happened in 2013? RPI beating Pablo? Never knew that.
|
|
|
Post by hammer on Dec 22, 2014 17:55:10 GMT -5
Simply really: Pablo has been hanging out with Rampager.
|
|
|
Post by badgerbreath on Dec 22, 2014 17:56:18 GMT -5
Is that for the whole year? Or just the NCAAs?
|
|
|
Post by vbbetterthanbb on Dec 22, 2014 17:57:29 GMT -5
...what the heck happened in 2013? RPI beating Pablo?... Largely because RPI picked PSU to win and Pablo picked Stanford.
|
|
|
Post by vbbetterthanbb on Dec 22, 2014 17:58:29 GMT -5
Is that for the whole year? Or just the NCAAs? NCAA was the bracket contest, PTW-Wkly was for whole year.
|
|
|
Post by badgerbreath on Dec 22, 2014 18:14:45 GMT -5
Is that for the whole year? Or just the NCAAs? NCAA was the bracket contest, PTW-Wkly was for whole year. So there are 63 matches in the tournament, I think (32+16+8+4+2+1=63), but I don't get round numbers when I multiply most of these fractions by 63. How did you calculate this?
|
|
|
Post by vbbetterthanbb on Dec 22, 2014 19:04:10 GMT -5
So there are 63 matches in the tournament, I think (32+16+8+4+2+1=63), but I don't get round numbers when I multiply most of these fractions by 63. How did you calculate this? Calculated the same way as any entry to most bracket contests, such as PTW-NCAA. Pablo couldn't enter itself into the contest, so publius did that for Pablo, and selected a winner for each NCAA match-up based on pablo's ratings for the 2 teams. For example, Pablo picked Washington to win during the first 5 rounds (then lose in the final to PSU). For those 5 picks on Washington, Pablo got 2 of 5 points since Washington lost in the 3rd round. Pablo picked PSU in all 6 rounds. PSU's 6 wins got Pablo 6 of 6 points. Pablo got 55 correct out of the 63 matches, therefore 87.3%.
|
|
|
Post by badgerbreath on Dec 22, 2014 19:16:07 GMT -5
So there are 63 matches in the tournament, I think (32+16+8+4+2+1=63), but I don't get round numbers when I multiply most of these fractions by 63. How did you calculate this? Calculated the same way as any entry to most bracket contests, such as PTW-NCAA. Pablo couldn't enter itself into the contest, so publius did that for Pablo, and selected a winner for each NCAA match-up based on pablo's ratings for the 2 teams. For example, Pablo picked Washington to win during the first 5 rounds (then lose in the final to PSU). For those 5 picks on Washington, Pablo got 2 of 5 points since Washington lost in the 3rd round. Pablo picked PSU in all 6 rounds. PSU's 6 wins got Pablo 6 of 6 points. Pablo got 55 correct out of the 63 matches, therefore 87.3%. That's right, but 80.3% of 63 for Kern is 50.59 matches. And 75% of 63 for AVCA is 47.25 matches. Just trying to make sure.
|
|
|
Post by hebrooks87 on Dec 22, 2014 19:31:26 GMT -5
Calculated the same way as any entry to most bracket contests, such as PTW-NCAA. Pablo couldn't enter itself into the contest, so publius did that for Pablo, and selected a winner for each NCAA match-up based on pablo's ratings for the 2 teams. For example, Pablo picked Washington to win during the first 5 rounds (then lose in the final to PSU). For those 5 picks on Washington, Pablo got 2 of 5 points since Washington lost in the 3rd round. Pablo picked PSU in all 6 rounds. PSU's 6 wins got Pablo 6 of 6 points. Pablo got 55 correct out of the 63 matches, therefore 87.3%. That's right, but 80.3% of 63 for Kern is 50.59 matches. And 75% of 63 for AVCA is 47.25 matches. Just trying to make sure. But if neither team was ranked in AVCA or Kern, there wouldn't be a prediction.
|
|
|
Post by badgerbreath on Dec 22, 2014 19:32:20 GMT -5
Ah...that must be it. It's important then that ranking systems only be compared based on matches for which they both made a prediction. Pablo makes predictions for all of the big mismatches early in the tournament, and the other ranking systems tend to exclude some of those because the underdog is not ranked. That would bias the results toward Pablo simply because it predicts some results for some real mismatches. Pablo-RPI comparisons would not have this problem.
|
|
|
Post by vbbetterthanbb on Dec 22, 2014 19:33:40 GMT -5
That's right, but 80.3% of 63 for Kern is 50.59 matches. And 75% of 63 for AVCA is 47.25 matches. Just trying to make sure. VT, Pablo and RPI got to play all 63 matches. But since there were match-ups involving 2 unranked teams by Kern and AVCA, neither of those 2 polls can ever pick all 63 matches. Therefore Kern's score was 49/61, not 49/63. And AVCA's score was 45/60, not 45/63. By the way, here are Pablo's 8 misses, arguably all close calls: Rd Match Pablo's pick 1 Kansas v. UALR (Kansas) 1 Hawaii v. Duke (Duke) 1 Oklahoma v. LSU (Oklahoma) 2 Oregon State v. UALR (Kansas instead of UALR) 2 Kentucky v. Ohio State (Kentucky) 3 Washington v. Nebraska (Washington) 4 Nebraska v. BYU (Washington instead of Nebraska) 5 BYU vs. Texas (Washington instead of BYU)
It was also mentioned that Pablo rated BYU higher than Nebraska and Texas.
|
|