bluepenquin
Hall of Fame
![*](//storage.proboards.com/forum/images/stars/star_blue.png) ![*](//storage.proboards.com/forum/images/stars/star_blue.png) ![*](//storage.proboards.com/forum/images/stars/star_blue.png) ![*](//storage.proboards.com/forum/images/stars/star_blue.png) ![*](//storage.proboards.com/forum/images/stars/star_blue.png)
4-Time VolleyTalk Poster of the Year (2019, 2018, 2017, 2016), All-VolleyTalk 1st Team (2021, 2020, 2019, 2018, 2017, 2016)
Posts: 12,572
|
Post by bluepenquin on Sept 15, 2021 11:55:40 GMT -5
Stanfords RPI is juicy and silly. Just for fun - if we think that Stanford is the same as Washington (per Pablo) and better than everyone else in the conference. Then their RPI Futures would #12. Stanford has the most RPI upside in the conference because of their schedule. USC has upside - and Utah's downside doesn't drop real fast because they have good schedules. Oregon and Colorado don't have very good schedules - which will hurt their chances for a seed. Washington is a complete and utter disaster. There is almost no upside for them short of 18+ conference wins. Iowa, Portland, and Ohio are RPI killers. Most teams try and avoid one RPI killer in the non conference - Washington has 3 of them. Then throw in Utah/Colorado are the two teams they don't play twice - among the top projected w/l% in conference. Yikes.
|
|
|
Post by slxpress on Sept 15, 2021 12:02:46 GMT -5
Remember it's not actually a prediction. It's a probability exercise. As Stanford goes through the season, the more they win, the more that probability of 13 total losses decreases. It's intended as a tool to show where teams stand in probabilistic RPI at the end of the year, as opposed to where they stand at this exact moment, which is wildly inaccurate as a measure. If viewed through the right lens it's very interesting. If viewed through the wrong lens is very mockable. But that's more a matter of viewing it wrong than a problem with the exercise itself. I’m aware of how these futures are generated, thanks. You're welcome.
|
|
trojansc
Legend
![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png)
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: 29,071
|
Post by trojansc on Sept 15, 2021 12:08:19 GMT -5
Remember it's not actually a prediction. It's a probability exercise. As Stanford goes through the season, the more they win, the more that probability of 13 total losses decreases. It's intended as a tool to show where teams stand in probabilistic RPI at the end of the year, as opposed to where they stand at this exact moment, which is wildly inaccurate as a measure. If viewed through the right lens it's very interesting. If viewed through the wrong lens is very mockable. But that's more a matter of viewing it wrong than a problem with the exercise itself. I’m aware of how these futures are generated, thanks. I don't think many may realize your 'LOL' has to do more with Pablo's view of Stanford rather than how RPI futures works (At least that's what I am assuming). The issue is not RPI Futures probability in these early rankings, it's that the ranking system (Pablo) isn't nearly as close to an accurate picture yet as it should be, but it's getting closer !
|
|
|
Post by slxpress on Sept 15, 2021 12:19:49 GMT -5
I’m aware of how these futures are generated, thanks. I don't think many may realize your 'LOL' has to do more with Pablo's view of Stanford rather than how RPI futures works (At least that's what I am assuming). The issue is not RPI Futures probability in these early rankings, it's that the ranking system (Pablo) isn't nearly as close to an accurate picture yet as it should be, but it's getting closer ! This happens every year, though, trojansc. Pablo isn't really a prediction, either. The point is still that all of this is simply to give a more accurate snapshot of what the final RPI could look like. Being critical of the numbers because of some notion that it's doing something else - like predicting win-loss records - is completely missing the purpose.
|
|
|
Post by ay2013 on Sept 15, 2021 12:32:04 GMT -5
Stanfords RPI is juicy and silly. Just for fun - if we think that Stanford is the same as Washington (per Pablo) and better than everyone else in the conference. Then their RPI Futures would #12. Stanford has the most RPI upside in the conference because of their schedule. USC has upside - and Utah's downside doesn't drop real fast because they have good schedules. Oregon and Colorado don't have very good schedules - which will hurt their chances for a seed. Washington is a complete and utter disaster. There is almost no upside for them short of 18+ conference wins. Iowa, Portland, and Ohio are RPI killers. Most teams try and avoid one RPI killer in the non conference - Washington has 3 of them. Then throw in Utah/Colorado are the two teams they don't play twice - among the top projected w/l% in conference. Yikes. Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records.
|
|
trojansc
Legend
![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png)
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: 29,071
|
Post by trojansc on Sept 15, 2021 12:44:25 GMT -5
I don't think many may realize your 'LOL' has to do more with Pablo's view of Stanford rather than how RPI futures works (At least that's what I am assuming). The issue is not RPI Futures probability in these early rankings, it's that the ranking system (Pablo) isn't nearly as close to an accurate picture yet as it should be, but it's getting closer ! This happens every year, though, trojansc. Pablo isn't really a prediction, either. The point is still that all of this is simply to give a more accurate snapshot of what the final RPI could look like. Being critical of the numbers because of some notion that it's doing something else - like predicting win-loss records - is completely missing the purpose. This happens in a different perspective this year because of last year’s incomplete season. Pablo is used in this instance to predict W/L records, and is a predictive measurement of probabilities. The purpose is not missed. It was simply laughing at the W-L record.
|
|
|
Post by n00b on Sept 15, 2021 12:48:41 GMT -5
Just for fun - if we think that Stanford is the same as Washington (per Pablo) and better than everyone else in the conference. Then their RPI Futures would #12. Stanford has the most RPI upside in the conference because of their schedule. USC has upside - and Utah's downside doesn't drop real fast because they have good schedules. Oregon and Colorado don't have very good schedules - which will hurt their chances for a seed. Washington is a complete and utter disaster. There is almost no upside for them short of 18+ conference wins. Iowa, Portland, and Ohio are RPI killers. Most teams try and avoid one RPI killer in the non conference - Washington has 3 of them. Then throw in Utah/Colorado are the two teams they don't play twice - among the top projected w/l% in conference. Yikes. Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records. Except one of the upsides of RPI (I’m not sure if it was intentional) is that it incentivizes top teams to schedule top mid-majors. That is an absolute plus in my opinion. Otherwise, Power 5 schools wild be incentivized to never play mid-majors that could get at larges so nobody can get quality wins. Texas vs Rice is good for the sport. USC and Kentucky in their annual tournament with Creighton and UNI is good for the sport.
|
|
trojansc
Legend
![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png) ![*](//storage.proboards.com/forum/images/stars/star_yellow.png)
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: 29,071
|
Post by trojansc on Sept 15, 2021 12:52:24 GMT -5
Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records. Except one of the upsides of RPI (I’m not sure if it was intentional) is that it incentivizes top teams to schedule top mid-majors. That is an absolute plus in my opinion. Otherwise, Power 5 schools wild be incentivized to never play mid-majors that could get at larges so nobody can get quality wins. Texas vs Rice is good for the sport. USC and Kentucky in their annual tournament with Creighton and UNI is good for the sport. It incentivizes teams to schedule good W/L record mid majors. There have been plenty of good WCC teams that are bad to schedule and you’re better off scheduling east coast cupcakes are more likely to put up good W/L records, whether they are good or not.
|
|
|
Post by n00b on Sept 15, 2021 13:05:48 GMT -5
Except one of the upsides of RPI (I’m not sure if it was intentional) is that it incentivizes top teams to schedule top mid-majors. That is an absolute plus in my opinion. Otherwise, Power 5 schools wild be incentivized to never play mid-majors that could get at larges so nobody can get quality wins. Texas vs Rice is good for the sport. USC and Kentucky in their annual tournament with Creighton and UNI is good for the sport. It incentivizes teams to schedule good W/L record mid majors. There have been plenty of good WCC teams that are bad to schedule and you’re better off scheduling east coast cupcakes are more likely to put up good W/L records, whether they are good or not. I think those differences are more marginal because there is upside of getting a quality win. You’re never going to get a quality win out of Iowa. I don’t know any of the WCC coaches. Is scheduling tough for them? I look at the teams predicted to be in the middle of the conference, and they seem to attract the right levels of opponent. PEP/LMU/SC get teams like Oregon, Purdue, South Carolina, Washington this year.
|
|
|
Post by c4ndlelight on Sept 15, 2021 13:12:02 GMT -5
Just for fun - if we think that Stanford is the same as Washington (per Pablo) and better than everyone else in the conference. Then their RPI Futures would #12. Stanford has the most RPI upside in the conference because of their schedule. USC has upside - and Utah's downside doesn't drop real fast because they have good schedules. Oregon and Colorado don't have very good schedules - which will hurt their chances for a seed. Washington is a complete and utter disaster. There is almost no upside for them short of 18+ conference wins. Iowa, Portland, and Ohio are RPI killers. Most teams try and avoid one RPI killer in the non conference - Washington has 3 of them. Then throw in Utah/Colorado are the two teams they don't play twice - among the top projected w/l% in conference. Yikes. Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records. Lipstick on a pig is still a pig. These tweaks don't address that RPI's root formula is crappy. If the ACC and SEC (short conference schedules + geographic bias) actually achieve parity with the PAC and B1G, the RPI is going to have persistently ridiculous outcomes. The RPI sucks and has sucked, but since it has generally only screwed West Coast teams, and the powers-that-be in VB embrace screwing West Coast teams ("grow the game"), that's been considered acceptable. We'll see what happens when the screwing is more equitably shared on a long-term basis.
|
|
|
Post by n00b on Sept 15, 2021 13:16:02 GMT -5
Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records. Lipstick on a pig is still a pig. These tweaks don't address that RPI's root formula is crappy. If the ACC and SEC (short conference schedules + geographic bias) actually achieve parity with the PAC and B1G, the RPI is going to have persistently ridiculous outcomes. The RPI sucks and has sucked, but since it has generally only screwed West Coast teams, and the powers-that-be in VB embrace screwing West Coast teams ("grow the game"), that's been considered acceptable. We'll see what happens when the screwing is more equitably shared on a long-term basis. The RPI really isn't that bad. Is there another easily-replicable formula out there that is better? I think that level of transparency is important.
|
|
|
Post by ay2013 on Sept 15, 2021 13:20:20 GMT -5
Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records. Except one of the upsides of RPI (I’m not sure if it was intentional) is that it incentivizes top teams to schedule top mid-majors. That is an absolute plus in my opinion. Otherwise, Power 5 schools wild be incentivized to never play mid-majors that could get at larges so nobody can get quality wins. Texas vs Rice is good for the sport. USC and Kentucky in their annual tournament with Creighton and UNI is good for the sport. Top mid majors are top 100 RPI teams so all their matches would still count and are highly relevant. USC and Kentucky would absolutely get incentivized and rewarded for playing and beatings teams like Creighton and UNI (when UNI is solid). What I disagree with is USC getting whacked in RPI because they also choose to play a local team from down the street (CSUN). For the selection committee, how USC faired against Kentucky, Creighton, and UNI (usually they are a top 100 RPI Team) is relevant. The fact that they can beat CSUN is NOT relevant - yet, we make it relevant by putting CSUN's record 50% of USC's RPI. Washington travelled to Ohio to play Ohio State - a top 25 team. That tournament was rounded out by Northen Kentucky ("ok" for RPI) and Ohio (crummy for RPI). So, what is Washington supposed to do, pull an Ohio State and just NOT play one of the teams also in the gym because RPI says they are too bad? My point is who cares if Ohio is bad? playing them and beating them should not matter to the committee. The committee should concern themselves of who was better between Washington and Ohio State, not that both Washington and Ohio State can beat Northern Kentucky and that Washington can win against Ohio.y
|
|
|
Post by c4ndlelight on Sept 15, 2021 13:21:01 GMT -5
Lipstick on a pig is still a pig. These tweaks don't address that RPI's root formula is crappy. If the ACC and SEC (short conference schedules + geographic bias) actually achieve parity with the PAC and B1G, the RPI is going to have persistently ridiculous outcomes. The RPI sucks and has sucked, but since it has generally only screwed West Coast teams, and the powers-that-be in VB embrace screwing West Coast teams ("grow the game"), that's been considered acceptable. We'll see what happens when the screwing is more equitably shared on a long-term basis. The RPI really isn't that bad. Is there another easily-replicable formula out there that is better? I think that level of transparency is important. It really is though, particularly at the margins that matter for NCAA tourney selection (Top 5, Top 16/20, Top 50). There's a big fallacy that a ranking system is good because stats geeks look at it and see it mostly puts 300-odd teams in the right decile, but that's not what we're asking the system to do. Pablo is just as transparent, so is Massey. They at least provide better outcomes. I'm sure if NCAA hates those, they could create an ELO-based formula that would also be superior.
|
|
|
Post by ay2013 on Sept 15, 2021 13:23:34 GMT -5
Well, there isn't anything inherently wrong with the teams they play - the problem is who those teams play. I do think that the RPI inputs generally yields the best 100 teams or so. But for tournament purposes, they should scrap all the impact of sub 100 RPI teams. I look at a team like Washington who scheduled Iowa. Iowa actually isn't THAT bad, BUT they scheduled competitively in the non-conference and they play in the best conference. If they didn't have to run the Big 10 gauntlet, would their projected record really be 7-23? no. Then I look at a team like Purdue that has a very lofty SOS record, and I see teams like Jacksonville State, RPI #140, but with a nice 21-9 actual record and Lipscomb, RPI #144 and a respectable 15-11 record. When it comes time to determining which teams "earned" a seed or a bid, why should it matter to the committee that Washington/Purdue can beat these teams? None of them are remotely close to tournament teams per RPI, yet Purdue will get a healthier SOS boost than Washington because 50% of their RPI is Jacksonville State and Lipscomb's record meanwhile Washington gets an RPI SOS knock because Iowa's record is 7-23. I think that if the selection RPI gets readjusted to JUST include matches against top 100 teams, you can still reward those teams who play a very tough schedule without punishing teams who also have a few teams with abysmal records. Lipstick on a pig is still a pig. These tweaks don't address that RPI's root formula is crappy. If the ACC and SEC (short conference schedules + geographic bias) actually achieve parity with the PAC and B1G, the RPI is going to have persistently ridiculous outcomes. The RPI sucks and has sucked, but since it has generally only screwed West Coast teams, and the powers-that-be in VB embrace screwing West Coast teams ("grow the game"), that's been considered acceptable. We'll see what happens when the screwing is more equitably shared on a long-term basis. I think this would be a really nice shade of lipstick though. A brand new RPI reran with just the records of the top 100 teams.
|
|
|
Post by c4ndlelight on Sept 15, 2021 13:25:44 GMT -5
Lipstick on a pig is still a pig. These tweaks don't address that RPI's root formula is crappy. If the ACC and SEC (short conference schedules + geographic bias) actually achieve parity with the PAC and B1G, the RPI is going to have persistently ridiculous outcomes. The RPI sucks and has sucked, but since it has generally only screwed West Coast teams, and the powers-that-be in VB embrace screwing West Coast teams ("grow the game"), that's been considered acceptable. We'll see what happens when the screwing is more equitably shared on a long-term basis. I think this would be a really nice shade of lipstick though. A brand new RPI reran with just the records of the top 100 teams. W-L will still be super %*$#ty. 96. (100) Coastal Carolina - Sun Belt (19-7) - .5423 - 226 97. (93) Arizona - Pac 12 (15-16) - .5418 - 55
|
|