|
Post by horns1 on Nov 20, 2022 18:23:57 GMT -5
Stanford loses two RPI Top 25 wins with USC falling to #26. There's been a lot of movement for so many teams in that #20-30 range; more to come, I'm sure.
|
|
|
Post by mikegarrison on Nov 20, 2022 18:28:29 GMT -5
Stanford loses two RPI Top 25 wins with USC falling to #26. There's been a lot of movement for so many teams in that #20-30 range; more to come, I'm sure. I'll just say again that it must have been a complete idiot who came up with the idea of a bonus for 1-25 and a smaller bonus for 26-50 but seemingly didn't realize that this means arbitrary discontinuities at 25-26 and 50-51. There is no reason these bonuses could not have been done as a simple linear scale between 1 and 50. Or even, let's say, a flat bonus from 1-20, a linear decrease from 21-30, a flat bonus from 31-45, and a linear decrease from 46-55. There are many, many ways they could have avoided this problem, and they chose to do none of them.
|
|
|
Post by horns1 on Nov 20, 2022 18:34:31 GMT -5
Stanford loses two RPI Top 25 wins with USC falling to #26. There's been a lot of movement for so many teams in that #20-30 range; more to come, I'm sure. I'll just say again that it must have been a complete idiot who came up with the idea of a bonus for 1-25 and a smaller bonus for 26-50 but seemingly didn't realize that this means arbitrary discontinuities at 25-26 and 50-51. There is no reason these bonuses could not have been done as a simple linear scale between 1 and 50. Or even, let's say, a flat bonus from 1-20, a linear decrease from 21-30, a flat bonus from 31-45, and a linear decrease from 46-55. There are many, many ways they could have avoided this problem, and they chose to do none of them. I really like what men's and women's hoops started doing a year or two ago with assigning wins/losses to quadrants (Q1, Q2, Q3, etc.) based on games being played at home, on the road, or at neutral sites. Of course, RPI is still a big factor. And, there still have to be cutoffs for each quadrants. Hoping volleyball moves to something close to this.
|
|
|
Post by jayj79 on Nov 20, 2022 18:43:51 GMT -5
I'll just say again that it must have been a complete idiot who came up with the idea of a bonus for 1-25 and a smaller bonus for 26-50 but seemingly didn't realize that this means arbitrary discontinuities at 25-26 and 50-51. There is no reason these bonuses could not have been done as a simple linear scale between 1 and 50. Or even, let's say, a flat bonus from 1-20, a linear decrease from 21-30, a flat bonus from 31-45, and a linear decrease from 46-55. There are many, many ways they could have avoided this problem, and they chose to do none of them. I really like what men's and women's hoops started doing a year or two ago with assigning wins/losses to quadrants (Q1, Q2, Q3, etc.) based on games being played at home, on the road, or at neutral sites. Of course, RPI is still a big factor. And, there still have to be cutoffs for each quadrants. Hoping volleyball moves to something close to this. making a distinction between home/neutral/away results is certainly an improvement, but it still sets arbitrary benchmarks same as the RPI bonuses.
|
|
trojansc
Legend
All-VolleyTalk 1st Team (2022, 2021, 2020, 2019, 2018, 2017), All-VolleyTalk 2nd Team (2016), 2021, 2019 Fantasy League Champion, 2020 Fantasy League Runner Up, 2022 2nd Runner Up
Posts: 28,639
|
Post by trojansc on Nov 20, 2022 18:44:01 GMT -5
I’ll try to get a Bracketology out late tonight.
|
|
|
Post by horns1 on Nov 20, 2022 18:51:04 GMT -5
I really like what men's and women's hoops started doing a year or two ago with assigning wins/losses to quadrants (Q1, Q2, Q3, etc.) based on games being played at home, on the road, or at neutral sites. Of course, RPI is still a big factor. And, there still have to be cutoffs for each quadrants. Hoping volleyball moves to something close to this. making a distinction between home/neutral/away results is certainly an improvement, but it still sets arbitrary benchmarks same as the RPI bonuses. If there was a perfect method to all this someone would have already identified it!
|
|
trojansc
Legend
All-VolleyTalk 1st Team (2022, 2021, 2020, 2019, 2018, 2017), All-VolleyTalk 2nd Team (2016), 2021, 2019 Fantasy League Champion, 2020 Fantasy League Runner Up, 2022 2nd Runner Up
Posts: 28,639
|
Post by trojansc on Nov 20, 2022 18:56:26 GMT -5
I like looking at Team Sheets/Nitty Gritty with grouping 1-25, 26-50, etc. and viewing records against those teams. I don’t like bonus points being so dependent on a spot or two in RPI, mike’s solution to that is much better.
And when looking at things like Nitty Gritty, it’s important to go deeper. Like the last 10 metric is completely useless without context. A lot of these AQ’s will be 8-2, 9-1, or 10-0 in their last 10.
3-1 vs. Top 25 is not necessarily better or worse than 5-4 vs. Top 25 in RPI. Those 3 wins could all be better/worse, Top 10 or 20-25, and same with the losses.
It’s easy to use those numbers to make arguments though.
|
|
|
Post by rainbowbadger on Nov 20, 2022 19:03:59 GMT -5
According to Figstats, WIS/tOSU/NEB are now 5/6/7 in RPI.
This week’s gonna be BONKERS.
|
|
|
Post by c4ndlelight on Nov 20, 2022 19:04:48 GMT -5
I like looking at Team Sheets/Nitty Gritty with grouping 1-25, 26-50, etc. and viewing records against those teams. I don’t like bonus points being so dependent on a spot or two in RPI, mike’s solution to that is much better. And when looking at things like Nitty Gritty, it’s important to go deeper. Like the last 10 metric is completely useless without context. A lot of these AQ’s will be 8-2, 9-1, or 10-0 in their last 10. 3-1 vs. Top 25 is not necessarily better or worse than 5-4 vs. Top 25 in RPI. Those 3 wins could all be better/worse, Top 10 or 20-25, and same with the losses. It’s easy to use those numbers to make arguments though. We had fun with 48 hours of the Huskies on the bubble, but UW is in. What I don't know is how teams like Michigan and Utah (who IMO are better than several at-larges whp will make it) will be treated by the committee against teams like Texas St. and the SEC's mass of mediocrity.
|
|
|
Post by jayj79 on Nov 20, 2022 19:12:32 GMT -5
Stanford loses two RPI Top 25 wins with USC falling to #26. There's been a lot of movement for so many teams in that #20-30 range; more to come, I'm sure. I'll just say again that it must have been a complete idiot who came up with the idea of a bonus for 1-25 and a smaller bonus for 26-50 but seemingly didn't realize that this means arbitrary discontinuities at 25-26 and 50-51. There is no reason these bonuses could not have been done as a simple linear scale between 1 and 50. Or even, let's say, a flat bonus from 1-20, a linear decrease from 21-30, a flat bonus from 31-45, and a linear decrease from 46-55. There are many, many ways they could have avoided this problem, and they chose to do none of them. why limit the linear scale to an arbitrary limit at all? why not have a linear scale for the whole field? or just get rid of bonuses altogether.
|
|
|
Post by hornshouse23 on Nov 20, 2022 19:16:16 GMT -5
I’ll try to get a Bracketology out late tonight. Yes!
|
|
|
Post by jwvolley on Nov 20, 2022 19:29:51 GMT -5
I’ll try to get a Bracketology out late tonight. Yes! We are spoiled
|
|
|
Post by BigDigEnergy on Nov 20, 2022 19:39:49 GMT -5
I’ll try to get a Bracketology out late tonight. Yes please, thank you!!!
|
|
|
Post by hornshouse23 on Nov 20, 2022 19:41:36 GMT -5
|
|
|
Post by mikegarrison on Nov 20, 2022 19:46:11 GMT -5
I'll just say again that it must have been a complete idiot who came up with the idea of a bonus for 1-25 and a smaller bonus for 26-50 but seemingly didn't realize that this means arbitrary discontinuities at 25-26 and 50-51. There is no reason these bonuses could not have been done as a simple linear scale between 1 and 50. Or even, let's say, a flat bonus from 1-20, a linear decrease from 21-30, a flat bonus from 31-45, and a linear decrease from 46-55. There are many, many ways they could have avoided this problem, and they chose to do none of them. why limit the linear scale to an arbitrary limit at all? why not have a linear scale for the whole field? or just get rid of bonuses altogether. The bonuses are just a kludge to try to overcome the problem that RPI is bad at giving teams credit for beating high-level opponents. The whole RPI system is a stack of cards built on a shaky foundation, but the NCAA keeps using it. I think they like how it incentivizes teams to schedule a certain way in the OoC matches (that is, it incentivizes teams from top conferences to schedule against the better teams from lower-level conferences). But it's just a badly designed metric, and manipulating bad data by adjusting it with more bad data (trying to fix RPI by using RPI) is not really a good idea.
|
|