Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 28, 2018 23:43:42 GMT -5
Not a lot of respect for Illinois. Lowest winning odds of the To 4 seeds, and below lower seeds PSU and Kentucky. FWIW, computers don't really have "respect" for anybody. You don’t say? It was just a general comment about the odds. Considering, I think, Illinois finished 2nd in RPI, the odds probably say more about the relative strength of the other team(s) in the group of 4. |
|
|
|
Post by pavsec5row10 on Nov 28, 2018 23:52:32 GMT -5
Why does THAT team always get this friggin' pass to a regional? Location, Location, Location. |
|
|
|
Post by jayj79 on Nov 29, 2018 0:01:23 GMT -5
FWIW, computers don't really have "respect" for anybody. You don’t say? It was just a general comment about the odds. Considering, I think, Illinois finished 2nd in RPI, the odds probably say more about the relative strength of the other team(s) in the group of 4. RPI and Pablo calculate different things. RPI is just a reflection of a conglomeration of your own winning percentage, all of your opponents' winning percentages, and your opponents' opponents' winning percentages (with a few modifiers for OOC scheduling and "good wins" and "bad losses" mixed in for flavor). RPI isn't designed to be predictive at all. Pablo on the other hand, was designed to be predictive. It factors in set/match scores and other parameters beyond just wins and losses, along with some voodoo magic. So it is a totally different beast than RPI. But yes, these odds of advancement definitely do reflect the varying strength/difficulty of the other teams in the subregional pods, not just the "respect" for the various seeded teams. A good seeded team in a pod with one or two stronger teams may have lower odds of advancement than a weaker seed in a pod with the other teams being weaker than those in the stronger seed's pod. |
|
|
|
Post by Fight On! on Nov 29, 2018 1:15:13 GMT -5
By the by, why is it that the team with the 2nd best odds of winning their sub-regional is not the 2nd best seed? In fact, THAT team is not even a top 4 seed. Why does THAT team always get this friggin' pass to a regional? The other day I grouped the field into four tiers, with the 16 seeds being Tier 1 and everyone else placed in RPI order within Tiers 2, 3 and 4. The ideal bracket (in the absence of geographical restrictions) would include a 1/2/3/4 grouping at each site, with 1 vs. 4 and 2 vs. 3 in the 1st round. This year only seven of the groups include a 1/2/3/4 mix: Minnesota, Illinois, Wisconsin, Nebraska, Creighton, Kentucky and Washington State. Underweighted groups include Stanford at 1/3/4/4, BYU at 1/3/3/4 and Penn State at 1/3/3/4. Overweighted groups include Texas at 1/2/2/3, UCF at 1/2/2/3 and Marquette at 1/2/2/4. I only count 13. USC? Oregon? Pittsburgh? Thanks tho!! |
|
|
|
Post by FreeBall on Nov 29, 2018 8:33:39 GMT -5
The other day I grouped the field into four tiers, with the 16 seeds being Tier 1 and everyone else placed in RPI order within Tiers 2, 3 and 4. The ideal bracket (in the absence of geographical restrictions) would include a 1/2/3/4 grouping at each site, with 1 vs. 4 and 2 vs. 3 in the 1st round. This year only seven of the groups include a 1/2/3/4 mix: Minnesota, Illinois, Wisconsin, Nebraska, Creighton, Kentucky and Washington State. Underweighted groups include Stanford at 1/3/4/4, BYU at 1/3/3/4 and Penn State at 1/3/3/4. Overweighted groups include Texas at 1/2/2/3, UCF at 1/2/2/3 and Marquette at 1/2/2/4. I only count 13. USC? Oregon? Pittsburgh? Thanks tho!! I didn't list them because in my mind they weren't quite as bad. Here they are: USC at 1/2/4/4, Pittsburgh at 1/2/4/4 and Oregon at 1/2/3/3. |
|
|
|
Post by FreeBall on Nov 29, 2018 8:41:02 GMT -5
I noticed one interesting thing this year in the composition of the four tiers. There were more highly ranked (by RPI) AQ's, so two of the at-large teams dropped into Tier 4. San Diego has the 49th highest RPI in the field and LMU was 51st. This seems like another example of why RPI is considered an imperfect metric for ranking teams.
|
|
|
|
Post by jayj79 on Nov 29, 2018 9:27:03 GMT -5
I noticed one interesting thing this year in the composition of the four tiers. There were more highly ranked (by RPI) AQ's, so two of the at-large teams dropped into Tier 4. San Diego has the 49th highest RPI in the field and LMU was 51st. This seems like another example of why RPI is considered an imperfect metric for ranking teams. The #1 seed/#1 RPI team is also an AQ, as are many others. So I'm not understanding how the RPI of AQ teams really indicates anything wrong with the feasibility of the RPI metric. |
|
|
|
Post by FreeBall on Nov 29, 2018 9:39:48 GMT -5
I noticed one interesting thing this year in the composition of the four tiers. There were more highly ranked (by RPI) AQ's, so two of the at-large teams dropped into Tier 4. San Diego has the 49th highest RPI in the field and LMU was 51st. This seems like another example of why RPI is considered an imperfect metric for ranking teams. The #1 seed/#1 RPI team is also an AQ, as are many others. So I'm not understanding how the RPI of AQ teams really indicates anything wrong with the feasibility of the RPI metric. Two of the at-large teams this year are in Tier 4 and that is what I found to be unusual. This is the first time I recall this happening in the years I've analyzed the tournament field in this manner. Do you really think San Diego and Loyola Marymount are both among the weakest 16 teams in this year's tournament? |
|
|
|
Post by pavsec5row10 on Nov 29, 2018 10:23:03 GMT -5
By the by, why is it that the team with the 2nd best odds of winning their sub-regional is not the 2nd best seed? In fact, THAT team is not even a top 4 seed. Why does THAT team always get this friggin' pass to a regional? The other day I grouped the field into four tiers, with the 16 seeds being Tier 1 and everyone else placed in RPI order within Tiers 2, 3 and 4. The ideal bracket (in the absence of geographical restrictions) would include a 1/2/3/4 grouping at each site, with 1 vs. 4 and 2 vs. 3 in the 1st round. This year only seven of the groups include a 1/2/3/4 mix: Minnesota, Illinois, Wisconsin, Nebraska, Creighton, Kentucky and Washington State. Underweighted groups include Stanford at 1/3/4/4, BYU at 1/3/3/4 and Penn State at 1/3/3/4. Overweighted groups include Texas at 1/2/2/3, UCF at 1/2/2/3 and Marquette at 1/2/2/4. I couldn't understand how Stanford got such weak teams in their subregional. I mean Duke at 44th RPI as their 2nd tier team was so far off from the 32nd team an ideal bracket would have. Also, Wisconsin technically would be 1/1/3/4 with UNI at #16 in RPI. In terms of Seed/RPI rank the ideal total would be 130 (1,32,33,64,etc). From toughest to easiest sub-regional... 1. Texas 99 (5,22,30,42) 2. UCF 104 (13,18,27,46) 3. Wisconsin 116 (6,16,35,59) 4. Kentucky 117 (10,20,33,54) 4. Marquette 117 (14,24,29,50) 6. Oregon 124 (15,17,45,47) 7. Illinois 128 (3,28,37,60) 8. Nebraska 129 (7,31,39,52) 9. Creighton 134 (9,25,43,57) 9. BYU 134 (4,34,40,56) 11. Washington St 135 (16,26,38,55) 12. USC 141 (11,23,49,58) 13. Minnesota 145 (2,32,48,63) 14. Penn St 147 (8,36,41,62) 14. Pittsburgh 147 (12,21,53,61)
16. Stanford 160 (1,44,51,64)
|
|
|
|
Post by The Bofa on the Sofa on Nov 29, 2018 10:53:05 GMT -5
The other day I grouped the field into four tiers, with the 16 seeds being Tier 1 and everyone else placed in RPI order within Tiers 2, 3 and 4. The ideal bracket (in the absence of geographical restrictions) would include a 1/2/3/4 grouping at each site, with 1 vs. 4 and 2 vs. 3 in the 1st round. This year only seven of the groups include a 1/2/3/4 mix: Minnesota, Illinois, Wisconsin, Nebraska, Creighton, Kentucky and Washington State. Underweighted groups include Stanford at 1/3/4/4, BYU at 1/3/3/4 and Penn State at 1/3/3/4. Overweighted groups include Texas at 1/2/2/3, UCF at 1/2/2/3 and Marquette at 1/2/2/4. I couldn't understand how Stanford got such weak teams in their subregional. I mean Duke at 44th RPI as their 2nd tier team was so far off from the 32nd team an ideal bracket would have. Also, Wisconsin technically would be 1/1/3/4 with UNI at #16 in RPI. In terms of Seed/RPI rank the ideal total would be 130 (1,32,33,64,etc). From toughest to easiest sub-regional... 1. Texas 99 (5,22,30,42) 2. UCF 104 (13,18,27,46) 3. Wisconsin 116 (6,16,35,59) 4. Kentucky 117 (10,20,33,54) 4. Marquette 117 (14,24,29,50) 6. Oregon 124 (15,17,45,47) 7. Illinois 128 (3,28,37,60) 8. Nebraska 129 (7,31,39,52) 9. Creighton 134 (9,25,43,57) 9. BYU 134 (4,34,40,56) 11. Washington St 135 (16,26,38,55) 12. USC 141 (11,23,49,58) 13. Minnesota 145 (2,32,48,63) 14. Penn St 147 (8,36,41,62) 14. Pittsburgh 147 (12,21,53,61)
16. Stanford 160 (1,44,51,64)
This isn't a correct way to rate subregionals for any individual team. For example, consider Pittsburgh. While they play the 61st team in the first round matters, yes, but for the most part, all of the seeded teams are much better than their first round opponents, so the marginal difference is pretty small. In all cases, the seeded teams are highly favored in the first round (in RPI - not the case in Pablo). However, when it comes to the second round for Pittsburgh, the fact that the other team in the second round is 53 is pretty meaningless, since that team is not likely to be their opponent. What matters most in the second round is the quality of the better team, and that means the Pitt likely faces #21 in the second round. Your approach implicitly assumes that Pitt is equally likely to face the 21st or 53rd team. Of course, they aren't. Meanwhile, Penn St gets to face, at best, the 36th best team. There is no way that Pitt has an easier subregional than Penn St, even according to RPI. See the thread on Pablo Analyzes the Brackets to see how to evaluate this in a meaningful way. On the whole, RPI is not a great tool to use to do it, but even when using it, you have to assess it smartly. |
|
|
|
Post by noblesol on Nov 29, 2018 11:16:54 GMT -5
The #1 seed/#1 RPI team is also an AQ, as are many others. So I'm not understanding how the RPI of AQ teams really indicates anything wrong with the feasibility of the RPI metric. Two of the at-large teams this year are in Tier 4 and that is what I found to be unusual. This is the first time I recall this happening in the years I've analyzed the tournament field in this manner. Do you really think San Diego and Loyola Marymount are both among the weakest 16 teams in this year's tournament? In previous years (2014 and 2015) I've looked at the standard deviation of un-adjusted Sel RPI scores. Based on that, the tiers of relative strength groupings of teams in the tournament are not all the same size. Based on standard deviation groupings, Tier 1 would be the 3-sigma teams (super strong) of which there are usually about four teams. 2-sigma (strong) teams will fall into about the 5 - 52 un-adjusted Sel RPI scores group. I'd break the 2-sigma up into 2 tiers. Tier 2 = seed teams 5 - 16, and Tier 3 = remaining teams with un-adjusted Sel RPI up to ~52 (the tournament gives home advantage to the Tier 2 teams in the sub-regional, and home teams are typically stronger than away teams). Typically the weakest at large team's un-adjusted Sel RPI would fall within Tier 3 (2-sigma). The mid-point of 1-sigma (average) teams is ~168. I'd put the top half of 1-sigma teams in a Tier 4, and the bottom half in Tier 5. I'd put the remaining (-2)-sigma (weak) and (-3)-sigma (super weak) teams in Tiers 6 and 7. I haven't seen a Tier 6 or 7 team in the tournament, but I've only looked at a few years.
So, I'd expect the following relative strength groupings and sizes in a typical tournament: Tier 1 = 4 'super strong' 3-sigma teams, typically the top four seeds; Tier 2 = 12 'home strong' 2-sigma teams, the remaining seeds; Tier 3 = the 'away' strength 2-sigma teams ranked from 17 to about 52 in un-adjusted Sel RPI scores, which typically includes all remaining at-large teams; Tier 4 = the top half of the 'average' strength 1-sigma teams, typically about 5 AQs; Tier 5 = the bottom half of the average strength teams, about 3 AQs. |
|
|
|
Post by pavsec5row10 on Nov 29, 2018 11:17:23 GMT -5
I couldn't understand how Stanford got such weak teams in their subregional. I mean Duke at 44th RPI as their 2nd tier team was so far off from the 32nd team an ideal bracket would have. Also, Wisconsin technically would be 1/1/3/4 with UNI at #16 in RPI. In terms of Seed/RPI rank the ideal total would be 130 (1,32,33,64,etc). From toughest to easiest sub-regional... 1. Texas 99 (5,22,30,42) 2. UCF 104 (13,18,27,46) 3. Wisconsin 116 (6,16,35,59) 4. Kentucky 117 (10,20,33,54) 4. Marquette 117 (14,24,29,50) 6. Oregon 124 (15,17,45,47) 7. Illinois 128 (3,28,37,60) 8. Nebraska 129 (7,31,39,52) 9. Creighton 134 (9,25,43,57) 9. BYU 134 (4,34,40,56) 11. Washington St 135 (16,26,38,55) 12. USC 141 (11,23,49,58) 13. Minnesota 145 (2,32,48,63) 14. Penn St 147 (8,36,41,62) 14. Pittsburgh 147 (12,21,53,61)
16. Stanford 160 (1,44,51,64)
This isn't a correct way to rate subregionals for any individual team. For example, consider Pittsburgh. While they play the 61st team in the first round matters, yes, but for the most part, all of the seeded teams are much better than their first round opponents, so the marginal difference is pretty small. In all cases, the seeded teams are highly favored in the first round (in RPI - not the case in Pablo). However, when it comes to the second round for Pittsburgh, the fact that the other team in the second round is 53 is pretty meaningless, since that team is not likely to be their opponent. What matters most in the second round is the quality of the better team, and that means the Pitt likely faces #21 in the second round. Your approach implicitly assumes that Pitt is equally likely to face the 21st or 53rd team. Of course, they aren't. Meanwhile, Penn St gets to face, at best, the 36th best team. There is no way that Pitt has an easier subregional than Penn St, even according to RPI. See the thread on Pablo Analyzes the Brackets to see how to evaluate this in a meaningful way. On the whole, RPI is not a great tool to use to do it, but even when using it, you have to assess it smartly. Actually it's correct for the purpose stated. Wasn't trying to claim that the Pablo percentages aren't better. |
|
|
|
Post by The Bofa on the Sofa on Nov 29, 2018 11:19:43 GMT -5
This isn't a correct way to rate subregionals for any individual team. For example, consider Pittsburgh. While they play the 61st team in the first round matters, yes, but for the most part, all of the seeded teams are much better than their first round opponents, so the marginal difference is pretty small. In all cases, the seeded teams are highly favored in the first round (in RPI - not the case in Pablo). However, when it comes to the second round for Pittsburgh, the fact that the other team in the second round is 53 is pretty meaningless, since that team is not likely to be their opponent. What matters most in the second round is the quality of the better team, and that means the Pitt likely faces #21 in the second round. Your approach implicitly assumes that Pitt is equally likely to face the 21st or 53rd team. Of course, they aren't. Meanwhile, Penn St gets to face, at best, the 36th best team. There is no way that Pitt has an easier subregional than Penn St, even according to RPI. See the thread on Pablo Analyzes the Brackets to see how to evaluate this in a meaningful way. On the whole, RPI is not a great tool to use to do it, but even when using it, you have to assess it smartly. Actually it's correct for the purpose stated. Wasn't trying to claim that the Pablo percentages aren't better. I was basing it on your "toughest" and "easiest" subregional statement. Your approach does not assess the difficulty of the subregionals. |
|
|
|
Post by Fight On! on Nov 29, 2018 12:09:49 GMT -5
I only count 13. USC? Oregon? Pittsburgh? Thanks tho!! I didn't list them because in my mind they weren't quite as bad. Here they are: USC at 1/2/4/4, Pittsburgh at 1/2/4/4 and Oregon at 1/2/3/3. Appreciate it  |
|
|
|
Post by tomclen on Nov 29, 2018 12:11:23 GMT -5
Am I the only one who confuses Creighton with Marquette? Why is that?
|
|
