RebKell's Junkie Boards
Board Junkies Forums
 
Log in Register FAQ Memberlist Search RebKell's Junkie Boards Forum Index

RPI rankings
Goto page Previous  1, 2
 
Post new topic   Reply to topic    RebKell's Junkie Boards Forum Index » NCAA Women's Basketball - General Discussion
View previous topic :: View next topic  
Author Message
ClayK



Joined: 11 Oct 2005
Posts: 10545



Back to top
PostPosted: 12/19/19 11:25 am    ::: Reply Reply with quote

It's also possible to cap the MOV at, say, 25, to give no advantage to running up the score.



_________________
Oṃ Tāre Tuttāre Ture Svāhā
GlennMacGrady



Joined: 03 Jan 2005
Posts: 6327
Location: Heisenberg


Back to top
PostPosted: 12/19/19 1:01 pm    ::: Reply Reply with quote

pilight wrote:

The best ratings systems use . . . .


"Best system" makes me think of an empirical question and also an empirical suggestion for you, pilight.

Which of the rating/ranking systems have been the most historically accurate in predicting the outcome of the NCAA tournament in your "fun bracket" over the years? Of course, that measures their predictive accuracy only as of their end-of-season rankings/ratings in March, when all the systems tend to converge.

The suggestion would be to take a snapshot of these ranking/rating systems at the ends of November, December, January and February -- and then compare how those earlier season ratings compare with the later NCAA tournament outcome. This would give some empirical evidence of which system is best in early and mid-season.
pilight



Joined: 23 Sep 2004
Posts: 61486
Location: Where the action is


Back to top
PostPosted: 12/19/19 1:09 pm    ::: Reply Reply with quote

Massey and Sagarin have been the best in the Fun Bracket over the years, with several wins each and consistently finishing near the top. RPI has only won once in 16 years (2008) and is usually well off the pace.



_________________
The power of the police to fulfill their functions and duties is dependent on public approval of their existence, actions and behavior, and on their ability to secure and maintain public respect
Conway Gamecock



Joined: 23 Jan 2015
Posts: 979
Location: Here


Back to top
PostPosted: 01/30/20 4:32 pm    ::: Reply Reply with quote

With some time now passing during the season, perhaps the RPI ratings will have more substance to them. Lets see:

Kenneth Massey Ratings: https://www.masseyratings.com/cbw/ncaa-d1/ratings
Warren Nolan Ratings: http://warrennolan.com/basketballw/2020/rpi-live


calbearman76



Joined: 02 Nov 2009
Posts: 4404
Location: Carson City


Back to top
PostPosted: 01/30/20 7:46 pm    ::: Reply Reply with quote

Looking at the ratings today we can see the true outliers in the top 32 of the RPI:

Central Michigan RPI 20 Massey 47. The Chippewas have not beaten a top 50 team. Their best RPI win is to #67 Marist. Three of their 4 losses are to teams ranked lower than them, including losses to #60 Central Florida and #145 Green Bay. They do not have any top 50 games remaining. Their best chance at having a top 50 RPI win at the end of the season is that if Marist wins out they could climb into the top 50.

Ohio St RPI 25 Massey 41. It is rare to have a major conference team so overrated in the RPI. The Buckeyes are 11-8 and have just 1 win against the top 45 (Michigan (46) is their only other top 50 win. They have lost to #93 Michigan St, #72 Ohio and #51 South Dakota (South Dakota is very underrated at this time. They could climb up to about 40 if they win out, but they have a far better resume than Ohio St.)

Western Kentucky RPI 29 Massey 74. The Hilltoppers are very interesting from an RPI standpoint. They have 3 top 50 wins over three vastly overrated teams. RPI#20 Central Michigan is #47 in Massey, RPI#47 Oklahoma is #67 in Massey and #50 Middle Tennessee is #84 in Massey. They are currently 6th in CUSA, behind both #124 UAB and #215 UTEP. They have losses to both #176 St Mary's and #224 North Texas. There is a good chance that 2 of their top 50 wins will not be top 50 by the end of the year, and their ranking will drop even if they win out because of their weak remaining schedule.

Old Dominion 31 RPI 52 Massey. The Monarchs may be the team best positioned to steal a bid because of the deficiencies of RPI. Not only does ODU have an inflated RPI but some of the teams they played do as well. Virginia (RPI 35 Massey 57), WKU (RPI 29 Massey 74) and Middle Tennessee (RPI 50 Massey 83) ODU has lost to #106 VCU and #124 UAB and their only other top 100 wins are over #71 Charlotte and #97 Idaho. Right now their resume (based on RPI) looks like they could get a bid. But if Virginia and MTSU both drop out of the top 50, which is expected, they will probably be left out even if they climb into the top 30.


Conway Gamecock



Joined: 23 Jan 2015
Posts: 979
Location: Here


Back to top
PostPosted: 01/31/20 12:03 am    ::: Reply Reply with quote

calbearman76 wrote:
Looking at the ratings today we can see the true outliers in the top 32 of the RPI:

Central Michigan RPI 20 Massey 47. The Chippewas have not beaten a top 50 team. Their best RPI win is to #67 Marist. Three of their 4 losses are to teams ranked lower than them, including losses to #60 Central Florida and #145 Green Bay. They do not have any top 50 games remaining. Their best chance at having a top 50 RPI win at the end of the season is that if Marist wins out they could climb into the top 50.

Ohio St RPI 25 Massey 41. It is rare to have a major conference team so overrated in the RPI. The Buckeyes are 11-8 and have just 1 win against the top 45 (Michigan (46) is their only other top 50 win. They have lost to #93 Michigan St, #72 Ohio and #51 South Dakota (South Dakota is very underrated at this time. They could climb up to about 40 if they win out, but they have a far better resume than Ohio St.)

Western Kentucky RPI 29 Massey 74. The Hilltoppers are very interesting from an RPI standpoint. They have 3 top 50 wins over three vastly overrated teams. RPI#20 Central Michigan is #47 in Massey, RPI#47 Oklahoma is #67 in Massey and #50 Middle Tennessee is #84 in Massey. They are currently 6th in CUSA, behind both #124 UAB and #215 UTEP. They have losses to both #176 St Mary's and #224 North Texas. There is a good chance that 2 of their top 50 wins will not be top 50 by the end of the year, and their ranking will drop even if they win out because of their weak remaining schedule.

Old Dominion 31 RPI 52 Massey. The Monarchs may be the team best positioned to steal a bid because of the deficiencies of RPI. Not only does ODU have an inflated RPI but some of the teams they played do as well. Virginia (RPI 35 Massey 57), WKU (RPI 29 Massey 74) and Middle Tennessee (RPI 50 Massey 83) ODU has lost to #106 VCU and #124 UAB and their only other top 100 wins are over #71 Charlotte and #97 Idaho. Right now their resume (based on RPI) looks like they could get a bid. But if Virginia and MTSU both drop out of the top 50, which is expected, they will probably be left out even if they climb into the top 30.


Central Michigan I see RPI 19 Massey 42. Perhaps they just moved up in the Massey ratings. Of course the Massey ratings are not RPI ratings, and do not use the RPI ratings' formula. Central Mich. has played a lot of Q2-level opponents thus far, and that's why they are rated as high in the RPI. They lost the first 3 games of their season, and have gone 15-1 since, going 6-1 against Q2 opponents. This is a moot point, as it looks like most of their remaining opponents are outside the top 100 in RPI with several outside the top 200, so they will steadily drop in RPI regardless if they win or lose.

Ohio State RPI 29 Massey 41. Ohio State has played the 3rd-toughest SOS thus far, with 6 games already played against opponents with top 10 RPI. Remember, in the RPI formula, your team's win-loss accounts for only 1/4th of the entire rating. The win-loss % of your opponents count for twice as much as that. The purpose of the RPI was to ensure that teams with great win-loss records were teams that earned those great win-loss records against solid schedules of opponents. The selection committees for the NCAAT wanted to be sure that they were filling the NCAAT every year with quality teams that would compete against other quality teams - the cream of the crop - and not put in teams with lots of wins, only to see them lose early in the tournament because they only padded their records with cupcakes.

Just looking at the face value of a win-loss record is easy for a committee member to do. But they didn't want to get sucked into some team that's 27-3, but all of those wins were against puffball opponents. Then they'd go up against a 19-11 P5 team in the 1st round, and lose by 25 points to them. Ohio State's RPI rating is high due to their tremendously tough schedule, and if they had any decent record - like 15-5 instead of 11-9 - they'd be a top-25 team and their RPI would most likely be top 10 instead of 29. The RPI rating shows if a team with a great record got it by playing cream puffs, or in the case of Ohio State, if a above-mediocre team will have a .500 record due to playing a very difficult schedule. If Ohio State's SOS was closer to 50 than 1, it would probably be 15-5 or 16-4 instead of 11-9, and it would be coasting into the NCAATs.

It's similar to Virginia: they have the #1 SOS, but their RPI is 43 because against that SOS they are 8-13. So they'll get a salute from the NCAA for playing such a schedule, but they ain't playing in the NCAAT. Most likely, not even in the WNIT. As it is, Massey also has UVA's SOS #1, and he has Ohio State's SOS #9, so in terms of that portion of the RPI equation, Massey's ratings isn't that far off. But Massey's ratings is not RPI - they are not the same thing.....


calbearman76



Joined: 02 Nov 2009
Posts: 4404
Location: Carson City


Back to top
PostPosted: 01/31/20 1:01 pm    ::: Reply Reply with quote

Conway Gamecock wrote:
calbearman76 wrote:
Looking at the ratings today we can see the true outliers in the top 32 of the RPI:

Central Michigan RPI 20 Massey 47. The Chippewas have not beaten a top 50 team. Their best RPI win is to #67 Marist. Three of their 4 losses are to teams ranked lower than them, including losses to #60 Central Florida and #145 Green Bay. They do not have any top 50 games remaining. Their best chance at having a top 50 RPI win at the end of the season is that if Marist wins out they could climb into the top 50.

Ohio St RPI 25 Massey 41. It is rare to have a major conference team so overrated in the RPI. The Buckeyes are 11-8 and have just 1 win against the top 45 (Michigan (46) is their only other top 50 win. They have lost to #93 Michigan St, #72 Ohio and #51 South Dakota (South Dakota is very underrated at this time. They could climb up to about 40 if they win out, but they have a far better resume than Ohio St.)

Western Kentucky RPI 29 Massey 74. The Hilltoppers are very interesting from an RPI standpoint. They have 3 top 50 wins over three vastly overrated teams. RPI#20 Central Michigan is #47 in Massey, RPI#47 Oklahoma is #67 in Massey and #50 Middle Tennessee is #84 in Massey. They are currently 6th in CUSA, behind both #124 UAB and #215 UTEP. They have losses to both #176 St Mary's and #224 North Texas. There is a good chance that 2 of their top 50 wins will not be top 50 by the end of the year, and their ranking will drop even if they win out because of their weak remaining schedule.

Old Dominion 31 RPI 52 Massey. The Monarchs may be the team best positioned to steal a bid because of the deficiencies of RPI. Not only does ODU have an inflated RPI but some of the teams they played do as well. Virginia (RPI 35 Massey 57), WKU (RPI 29 Massey 74) and Middle Tennessee (RPI 50 Massey 83) ODU has lost to #106 VCU and #124 UAB and their only other top 100 wins are over #71 Charlotte and #97 Idaho. Right now their resume (based on RPI) looks like they could get a bid. But if Virginia and MTSU both drop out of the top 50, which is expected, they will probably be left out even if they climb into the top 30.


Central Michigan I see RPI 19 Massey 42. Perhaps they just moved up in the Massey ratings. Of course the Massey ratings are not RPI ratings, and do not use the RPI ratings' formula. Central Mich. has played a lot of Q2-level opponents thus far, and that's why they are rated as high in the RPI. They lost the first 3 games of their season, and have gone 15-1 since, going 6-1 against Q2 opponents. This is a moot point, as it looks like most of their remaining opponents are outside the top 100 in RPI with several outside the top 200, so they will steadily drop in RPI regardless if they win or lose.

Ohio State RPI 29 Massey 41. Ohio State has played the 3rd-toughest SOS thus far, with 6 games already played against opponents with top 10 RPI. Remember, in the RPI formula, your team's win-loss accounts for only 1/4th of the entire rating. The win-loss % of your opponents count for twice as much as that. The purpose of the RPI was to ensure that teams with great win-loss records were teams that earned those great win-loss records against solid schedules of opponents. The selection committees for the NCAAT wanted to be sure that they were filling the NCAAT every year with quality teams that would compete against other quality teams - the cream of the crop - and not put in teams with lots of wins, only to see them lose early in the tournament because they only padded their records with cupcakes.

Just looking at the face value of a win-loss record is easy for a committee member to do. But they didn't want to get sucked into some team that's 27-3, but all of those wins were against puffball opponents. Then they'd go up against a 19-11 P5 team in the 1st round, and lose by 25 points to them. Ohio State's RPI rating is high due to their tremendously tough schedule, and if they had any decent record - like 15-5 instead of 11-9 - they'd be a top-25 team and their RPI would most likely be top 10 instead of 29. The RPI rating shows if a team with a great record got it by playing cream puffs, or in the case of Ohio State, if a above-mediocre team will have a .500 record due to playing a very difficult schedule. If Ohio State's SOS was closer to 50 than 1, it would probably be 15-5 or 16-4 instead of 11-9, and it would be coasting into the NCAATs.

It's similar to Virginia: they have the #1 SOS, but their RPI is 43 because against that SOS they are 8-13. So they'll get a salute from the NCAA for playing such a schedule, but they ain't playing in the NCAAT. Most likely, not even in the WNIT. As it is, Massey also has UVA's SOS #1, and he has Ohio State's SOS #9, so in terms of that portion of the RPI equation, Massey's ratings isn't that far off. But Massey's ratings is not RPI - they are not the same thing.....


My point is that the RPI ranks teams in a way that creates absurd results. You are correct that CMU lost their first 3 games, but that is not taken into account by either the RPI or Massey; all games are treated equally. My issue is that a team that has not beaten anyone in the top 50 and has lost a game to #145 (All of the rankings change continuously so I won't bother to update the rankings should not be a top 20 team. And while CMU does not play a top 50 team the rest of the way they will remain at least a top 25, and more likely a top 20 team in the RPI.

As for Ohio St, I would not have listed them if they had not been in the top 25. Their loss to Maryland dropped them by 5 slots, another example of where the results often don't have what would be the appropriate RPI effect. A team ranked #25 should not significantly losing to the number 6 team. As to your point, you are absolutely right that W-L record is not a good individual statistic to choose, strength of schedule must be factored in. But the RPI uses a terrible metric for SOS such that a game against #188 Binghamton or #194 Cleveland St is considered much tougher than a game versus #42 Virginia. Incidentally, the use of quadrants is even more problematic for evaluating teams because it draws a big distinction between a team in the 40s and a team in the 50s. Right now South Dakota (who easily handled Ohio St on a neutral court) is not even a top 50 team despite only 2 losses to top teams (Missouri St and South Carolina) and 3 top 50 wins (Drake, Creighton and Ohio St). South Carolina is deprived of a top 50 win for now because South Dakota is so underrated.

As we go through the rest of the season look at how the good teams from lesser conferences naturally drift upwards despite just beating mediocre competition. Watch how teams like James Madison, Central Florida, Bucknell, Marist, Stephen F Austin, Penn and Abilene Christian all rise even though they aren't beating anyone good teams.

In short the RPI is a terrible system, and it is indefensible that a group of institutions of higher learning would use such a system to rank sports teams. A freshman statistics student could easily build a better set of ratings in a week.


tfan



Joined: 31 May 2010
Posts: 8117



Back to top
PostPosted: 01/31/20 4:51 pm    ::: Reply Reply with quote

calbearman76 wrote:

But the RPI uses a terrible metric for SOS such that a game against #188 Binghamton or #194 Cleveland St is considered much tougher than a game versus #42 Virginia.


How did you determine that?


summertime blues



Joined: 16 Apr 2013
Posts: 5908
Location: Shenandoah Valley


Back to top
PostPosted: 01/31/20 6:27 pm    ::: Reply Reply with quote

tfan wrote:
calbearman76 wrote:

But the RPI uses a terrible metric for SOS such that a game against #188 Binghamton or #194 Cleveland St is considered much tougher than a game versus #42 Virginia.


How did you determine that?


Yeah, really......have you SEEN Virginia this year? #42 is an awfully high ranking for them this year, IMNSHO.



_________________
Don't take life so serious. It ain't nohows permanent.
It takes 3 years to build a team and 7 to build a program.--Conventional Wisdom
calbearman76



Joined: 02 Nov 2009
Posts: 4404
Location: Carson City


Back to top
PostPosted: 01/31/20 7:34 pm    ::: Reply Reply with quote

summertime blues wrote:
tfan wrote:
calbearman76 wrote:

But the RPI uses a terrible metric for SOS such that a game against #188 Binghamton or #194 Cleveland St is considered much tougher than a game versus #42 Virginia.


How did you determine that?


Yeah, really......have you SEEN Virginia this year? #42 is an awfully high ranking for them this year, IMNSHO.


The calculation of a team's RPI is based on 3 factors:

25% Win-Loss record
50% Win-Loss record of opponents
25% Win-Loss record of opponents' opponents

The second and third factors combined determine the strength of schedule.


The math becomes a little messy at this point because the records get combined with all other games, but essentially the record of the opponent is valued twice as much as the record of the opponent's opponent.

Therefore the simplified rating impact of any one game is (2*(winning percentage of the opponent) + 1*(winning percentage of opponent's opponents))/3

Based on that here are the calculations for the three subject teams(team, won-loss record (%), won-loss record of opponents (%)):

Virginia 8-13 (0.381) 293-112 (0.723)
2*0.381 + 0.723 = 1.485/3 = 0.495

Binghamton 14-6 (0.7) 139-221 (0.444)
2*0.7 + 0.444 = 1.844/3 = 0.615

Cleveland St 13-6 (0.684) 128-218 (0.37)
2*0.684 + 0.37 = 1.758/3 = 0.586


Therefore a game against Binghamton or Cleveland St results in a significantly higher Strength of Schedule. And while I agree with Summertime Blues that Virginia is overrated in the RPI, I still believe they are better than the Bearcats or the Vikings (although a sword fight between a Cavalier and a Viking could be intriguing.)


SpaceJunkie



Joined: 10 Sep 2012
Posts: 4224
Location: Minnesota


Back to top
PostPosted: 01/31/20 8:13 pm    ::: Reply Reply with quote

calbearman76 wrote:
Therefore a game against Binghamton or Cleveland St results in a significantly higher Strength of Schedule. And while I agree with Summertime Blues that Virginia is overrated in the RPI, I still believe they are better than the Bearcats or the Vikings (although a sword fight between a Cavalier and a Viking could be intriguing.)


RPI sucks no matter what, but would it be better if the ratios were changed so opponents' winning percentage was lower and opponents' oppoents' winning perecentage higher?


Conway Gamecock



Joined: 23 Jan 2015
Posts: 979
Location: Here


Back to top
PostPosted: 02/01/20 12:21 am    ::: Reply Reply with quote

calbearman76 wrote:
summertime blues wrote:
tfan wrote:
calbearman76 wrote:

But the RPI uses a terrible metric for SOS such that a game against #188 Binghamton or #194 Cleveland St is considered much tougher than a game versus #42 Virginia.


How did you determine that?


Yeah, really......have you SEEN Virginia this year? #42 is an awfully high ranking for them this year, IMNSHO.


The calculation of a team's RPI is based on 3 factors:

25% Win-Loss record
50% Win-Loss record of opponents
25% Win-Loss record of opponents' opponents

The second and third factors combined determine the strength of schedule.


The math becomes a little messy at this point because the records get combined with all other games, but essentially the record of the opponent is valued twice as much as the record of the opponent's opponent.

Therefore the simplified rating impact of any one game is (2*(winning percentage of the opponent) + 1*(winning percentage of opponent's opponents))/3

Based on that here are the calculations for the three subject teams(team, won-loss record (%), won-loss record of opponents (%)):

Virginia 8-13 (0.381) 293-112 (0.723)
2*0.381 + 0.723 = 1.485/3 = 0.495

Binghamton 14-6 (0.7) 139-221 (0.444)
2*0.7 + 0.444 = 1.844/3 = 0.615

Cleveland St 13-6 (0.684) 128-218 (0.37)
2*0.684 + 0.37 = 1.758/3 = 0.586


Therefore a game against Binghamton or Cleveland St results in a significantly higher Strength of Schedule. And while I agree with Summertime Blues that Virginia is overrated in the RPI, I still believe they are better than the Bearcats or the Vikings (although a sword fight between a Cavalier and a Viking could be intriguing.)



Hahaha. Laughing I think you need to replace the AA batteries in your calculator. And then check your basis for your math - it needs some help.... Laughing


pilight



Joined: 23 Sep 2004
Posts: 61486
Location: Where the action is


Back to top
PostPosted: 02/01/20 12:48 am    ::: Reply Reply with quote

calbearman76 wrote:
The math becomes a little messy at this point because the records get combined with all other games


Actually, they don't. Opponent's winning percentage in RPI is actually the average of your opponents winning percentages. OOWP is also an average.

The NCAA didn't want scheduling shenanigans, so they tweaked the system early on to make each opponent count equally towards OWP and OOWP just like they do for WP.



_________________
The power of the police to fulfill their functions and duties is dependent on public approval of their existence, actions and behavior, and on their ability to secure and maintain public respect
summertime blues



Joined: 16 Apr 2013
Posts: 5908
Location: Shenandoah Valley


Back to top
PostPosted: 02/01/20 12:30 pm    ::: Reply Reply with quote

Also mid-majors get screwed over on RPI because most P5s are reluctant to schedule them, fearing that a loss to a good mid-major will hurt their precious RPI.



_________________
Don't take life so serious. It ain't nohows permanent.
It takes 3 years to build a team and 7 to build a program.--Conventional Wisdom
calbearman76



Joined: 02 Nov 2009
Posts: 4404
Location: Carson City


Back to top
PostPosted: 02/01/20 4:09 pm    ::: Reply Reply with quote

pilight wrote:
calbearman76 wrote:
The math becomes a little messy at this point because the records get combined with all other games


Actually, they don't. Opponent's winning percentage in RPI is actually the average of your opponents winning percentages. OOWP is also an average.

The NCAA didn't want scheduling shenanigans, so they tweaked the system early on to make each opponent count equally towards OWP and OOWP just like they do for WP.


Thanks for that clarification. While I won't redo the calculation the result would not change materially. The only effect would be to make opponents that have played fewer games slightly more significant.


Conway Gamecock



Joined: 23 Jan 2015
Posts: 979
Location: Here


Back to top
PostPosted: 02/01/20 5:21 pm    ::: Reply Reply with quote

calbearman76 wrote:
pilight wrote:
calbearman76 wrote:
The math becomes a little messy at this point because the records get combined with all other games


Actually, they don't. Opponent's winning percentage in RPI is actually the average of your opponents winning percentages. OOWP is also an average.

The NCAA didn't want scheduling shenanigans, so they tweaked the system early on to make each opponent count equally towards OWP and OOWP just like they do for WP.


Thanks for that clarification. While I won't redo the calculation the result would not change materially. The only effect would be to make opponents that have played fewer games slightly more significant.


I think your above formula is simply flawed. You do not determine SOS based on Team A's W-L % + Team A's OWP. You determine it based on Team A's OWP + Team A's OOWP, and it's 2(OWP) + (OOWP)/3. That throws your whole math off - Virginia's SOS is .5853, not .4950. I didn't even bother to check Binghamton's or Cleveland St.'s formula's, but online sites have their SOS as .4177 and .3953, not .6150 and .5860 as your formula shows....

It's a strength of SCHEDULE, not strength of Team, so why include the W-L % of one single team?? Unless Team A plays Team B 25-30 times in a season, and has its schedule made up of pretty much one single opponent. Otherwise it needs to involve the entire schedules, not a single team like your formula suggests...


calbearman76



Joined: 02 Nov 2009
Posts: 4404
Location: Carson City


Back to top
PostPosted: 02/02/20 3:07 am    ::: Reply Reply with quote

Conway Gamecock wrote:
calbearman76 wrote:
pilight wrote:
calbearman76 wrote:
The math becomes a little messy at this point because the records get combined with all other games


Actually, they don't. Opponent's winning percentage in RPI is actually the average of your opponents winning percentages. OOWP is also an average.

The NCAA didn't want scheduling shenanigans, so they tweaked the system early on to make each opponent count equally towards OWP and OOWP just like they do for WP.


Thanks for that clarification. While I won't redo the calculation the result would not change materially. The only effect would be to make opponents that have played fewer games slightly more significant.


I think your above formula is simply flawed. You do not determine SOS based on Team A's W-L % + Team A's OWP. You determine it based on Team A's OWP + Team A's OOWP, and it's 2(OWP) + (OOWP)/3. That throws your whole math off - Virginia's SOS is .5853, not .4950. I didn't even bother to check Binghamton's or Cleveland St.'s formula's, but online sites have their SOS as .4177 and .3953, not .6150 and .5860 as your formula shows....

It's a strength of SCHEDULE, not strength of Team, so why include the W-L % of one single team?? Unless Team A plays Team B 25-30 times in a season, and has its schedule made up of pretty much one single opponent. Otherwise it needs to involve the entire schedules, not a single team like your formula suggests...


I am not trying to calculate the Strength of Schedule (SOS) for any of these teams. Instead I am calculating the affect of playing any of those teams in their own SOS. So, for instance, if George Washington were to play Virginia the impact on George Washington's SOS would be:

(2*(Virginia's W-L%) + Virginia's opponent's W-L%)/3

Virginia's SOS is actually .6625 (per Real Time RPI), not .5853. While I won't attempt to do the math, I will note that my calculation of the W-L% for Virginia's opponents was .7231. While not quite accurate (as Pilight noted) this amount, which I have used for the opponents' opponents W-L%, is actually the 50% portion of Virginia's RPI, and 2/3 of Virginia's SOS. Virginia's opponent's opponents W/L% calculation would therefore be approximately .5413.

3*0.6625=1.9875 - (2*0.7231) = 0.5413

The same can be done for Binghamton:

3*0.4165 = 1.2495 - (2*0.3861) = 0.4773

(note: I had an error in my previous post. The record of opponents for Binghamton is .3861, not .444. This would reduce the impact for Binghamton from 0.615 to 0.595, still well above Virginia's 0.495

or Cleveland St:

3*0.3930 = 1.1790 - (2*0.3700) = 0.4390

The point remains the same; playing Cleveland St or Binghamton results in a higher SOS than playing Virginia under the RPI formula.


Display posts from previous:   
Post new topic   Reply to topic    RebKell's Junkie Boards Forum Index » NCAA Women's Basketball - General Discussion All times are GMT - 5 Hours
Goto page Previous  1, 2
Page 2 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB 2.0.17 © 2001- 2004 phpBB Group
phpBB Template by Vjacheslav Trushkin