Latest Sagarin rankings - | The Boneyard

Latest Sagarin rankings -

Status
Not open for further replies.
VAMike23

VAMike23

The Virginian
Joined
Aug 26, 2011
Messages
8,512
Reaction Score
17,295
1. UCONN
2. DOOK
3. ND
4. Stanford
5. Louisville

LINK

Incidentally, if I read the page correctly, Jeff has UCONN's strength of schedule ranked 26th. St. Joseph's is #1. :confused:

Am I reading this correctly?

If so, it doesn't seem to make much sense at all. We have played two of the other teams in his top 5, as well as Maryland. I see that PSU has fallen in his rankings, but even so, I would think our schedule thus far is right up there with anyone else's.
 
VAMike23 said:
1. UCONN
2. DOOK
3. ND
4. Stanford
5. Louisville

LINK

Incidentally, if I read the page correctly, Jeff has UCONN's strength of schedule ranked 26th. St. Joseph's is #1. :confused:

Am I reading this correctly?

If so, it doesn't seem to make much sense at all. We have played two of the other teams in his top 5, as well as Maryland. I see that PSU has fallen in his rankings, but even so, I would think our schedule thus far is right up there with anyone else's.
Expand ...
I am not sure, but it appears the schedule strength includes games not played yet. Otherwise how would ND have a high SOS without playing a ranked team.
 
That is so true........This will be the issue going forward for UCONN.........Weak AAC and most of the Top 10 teams want no part of UCONN in the future. Reason #1, Get the hell out of this AAC............Big 10 here we come..........
 
VAMike, you not reading it wrong. Jeff is taking into account games not played yet in his SOS rating. But what he is NOT taking into account is that many of the teams on UConn's , or any other teams , schedule is that those other teams on the schedule could be a lot stronger and better than he is giving them credit for.
 
reno tony said:
I am not sure, but it appears the schedule strength includes games not played yet. Otherwise how would ND have a high SOS without playing a ranked team.
Expand ...
To you and the OP - the SOS is greatly influenced by the whole schedule of teams so while the Top of Uconn's schedule to date has been great, the bottom has been very low and the combination of those even with the strong top end drags it down. You would have a very high SOS if every team you played was ranked between 25 and 50. But having 1/3 ranked in the top 25, 1/3 in the 25-50 and 1/3 ranked below 200 will be mess up the overall SOS. Those bottom teams more than off-set the top 25 teams.
 
HGN said:
VAMike, you not reading it wrong. Jeff is taking into account games not played yet in his SOS rating. But what he is NOT taking into account is that many of the teams on UConn's , or any other teams , schedule is that those other teams on the schedule could be a lot stronger and better than he is giving them credit for.
Expand ...

I disagree. VAMike is reading it correctly and Sagarin does not include games yet to be played in his SOS calculations. But what Sagarin does do is calculate SOS as an average of the rankings of your opponents instead of weighing the tougher opponents more. Thus a schedule of teams ranked 2nd, 10th, and 288th (mean rank = 100th) is considered easier than a schedule with opponents ranked 50th, 75th, and 100th (average rank = 75th). No, it doesn't really make a lot of sense, except as I think Meyers or Phil pointed out in a earlier thread on this topic, a team ranked 200th that performed according to their ranking would finish the first schedule 1-2 but who lose all three games against the 75th ranked schedule. Of course as fans of great teams, we want SOS to measure schedule strength as perceived by great teams, not weaker teams, but that's not the way Sagarin's SOS works. However, his rating systems seem to utilize a more complex algorithm when determining the ratings (than just using record, SOS and MOV). Otherwise, Notre Dame would be ranked higher than UConn in ELO (which ignores MOV), as both teams are undefeated and Notre Dame has a higher calculated SOS (because they played fewer ultra-Twinkies than UConn).
 
.-.
Choke is right (but maybe I'm biased in favor of people who agree with me:)

The Sagarin approach does a fairly good job of earing team strength, thus the ranking, but not so hot at measuring schedule strength.

In my opinion, the optimal way of comparing schedules is to calculate the probability of loss - the higher the probability, the tougher the schedule.

The problem with this metric is that you cannot calculate it based upon the schedule itself. You need to measure it against an opponent. Which means that a particular schedule might be easier of tougher depending on who plays it. If anyone cares, I'll give an example.
 
Fightin Choke said:
I disagree. VAMike is reading it correctly and Sagarin does not include games yet to be played in his SOS calculations. But what Sagarin does do is calculate SOS as an average of the rankings of your opponents instead of weighing the tougher opponents more. Thus a schedule of teams ranked 2nd, 10th, and 288th (mean rank = 100th) is considered easier than a schedule with opponents ranked 50th, 75th, and 100th (average rank = 75th). No, it doesn't really make a lot of sense, except as I think Meyers or Phil pointed out in a earlier thread on this topic, a team ranked 200th that performed according to their ranking would finish the first schedule 1-2 but who lose all three games against the 75th ranked schedule. Of course as fans of great teams, we want SOS to measure schedule strength as perceived by great teams, not weaker teams, but that's not the way Sagarin's SOS works. However, his rating systems seem to utilize a more complex algorithm when determining the ratings (than just using record, SOS and MOV). Otherwise, Notre Dame would be ranked higher than UConn in ELO (which ignores MOV), as both teams are undefeated and Notre Dame has a higher calculated SOS (because they played fewer ultra-Twinkies than UConn).
Expand ...
Basically what I posted and spot on. It is the same issue with recruiting class rankings that I have as well. I have always thought that SOS should be calculated based on 'buckets' of teams that vary based on relative strength. A top five team should not get any bump for playing the #75 team in the country vs. the #225 team in the country - in both cases if they do not blow the team off the floor, they have played poorly. So, go ahead and use actual ranks for the top 15 teams for them, but then treat teams 16-25 as one value, teams 26-50 as another value and teams 51 - 100 and teams 101+ as pooled values. And this sliding evaluation should be used down the rankings as well so say a team at #15 uses 6 - 25 actual ranks, 1-5 gets a single value because beating any one of them is a huge upset, and then 26-40 gets one value 41-65 another, etc. With that kind of an SOS you are evaluated against your close brethren but against greatly superior or inferior opponents you are not overly rewarded or penalized.
 
jsloanjr said:
That is so true...This will be the issue going forward for UCONN....Weak AAC and most of the Top 10 teams want no part of UCONN in the future. Reason #1, Get the hell out of this AAC..Big 10 here we come.....
Expand ...
Wow, I never realized how easy it was for us to get into the Big 10. What a bunch of dummies we must have running things at UConn. I too thought UConn should leave the AAC, but thought the best we could do was to start a new league made up of the strongest mid-majors. BTW, is that for just the women's basketball team, or can all the UConn sports teams get into the BIG 10 that easily?
 
UcMiami said:
Basically what I posted and spot on. It is the same issue with recruiting class rankings that I have as well. I have always thought that SOS should be calculated based on 'buckets' of teams that vary based on relative strength. A top five team should not get any bump for playing the #75 team in the country vs. the #225 team in the country - in both cases if they do not blow the team off the floor, they have played poorly. So, go ahead and use actual ranks for the top 15 teams for them, but then treat teams 16-25 as one value, teams 26-50 as another value and teams 51 - 100 and teams 101+ as pooled values. And this sliding evaluation should be used down the rankings as well so say a team at #15 uses 6 - 25 actual ranks, 1-5 gets a single value because beating any one of them is a huge upset, and then 26-40 gets one value 41-65 another, etc. With that kind of an SOS you are evaluated against your close brethren but against greatly superior or inferior opponents you are not overly rewarded or penalized.
Expand ...


I like that approach. It only works for top teams, but let's face it, the 225th team is not really arguing with the 226th team about who has a tougher schedule. Or maybe they are, but that conversation isn't happening here.

Given a point for playing a triple digit ranked team (or maybe zero) 2 for 51-100, 3 for 26-50, etc. That would be a relatively eay way to compae schedules.
 
Phil said:
In my opinion, the optimal way of comparing schedules is to calculate the probability of loss - the higher the probability, the tougher the schedule.
Expand ...

And potentially means that when UCONN has a team head and shoulders above everyone else in WCBB and the probability of loss approaches zero no matter who they play they by definition play the easiest schedule.
 
Last edited:
Phil said:
I like that approach. It only works for top teams, but let's face it, the 225th team is not really arguing with the 226th team about who has a tougher schedule. Or maybe they are, but that conversation isn't happening here.

Given a point for playing a triple digit ranked team (or maybe zero) 2 for 51-100, 3 for 26-50, etc. That would be a relatively eay way to compae schedules.
Expand ...
But it would work for any team because you would still use the rankings for the 15 - 25 teams around them, but then just reduce the impact of the games against clearly superior or inferior teams. By the time you get to 200+ ranked teams, maybe you expand the real ranking universe to 50 or 100 teams because the difference between teams really does become minimal. For Uconn/Duke/TN/Louisville/Stanford I would see something like:
Playing top 15 = real ranking 1-15
Playing any team in the 15-25 = 20 per game
Playing 26-50 = 35
Playing 51-100 = 60
Playing 101 - anything = 100

For a team ranked #50
Playing any top 15 team = 5
Playing 16-35 = 20
Playing 36 - 70 = real ranking 36-70
Playing 71 - 100 = 75
Playing 100 - 150 = 120
Playing 151 - anything = 150
Probably need to make adjustments and anyone on the cusp of one of the breaks might have significant fluctuations week to week but I think it would make for a better true evaluation of strength of schedule relative to your peers. If it caught on teams might also actually consider the scheduling of games in a more competitive way.
 
.-.
Yeah, to say that any system would use games not played yet in the SOS is just crazy, and one look at some of the still horrific SOS's of some B12 teams like Baylor at #305 would instantly disprove any such strange idea.

Ditto to everything Phil, Choke, and UcMiami have said, with the note that finding a fair and simple SOS rating system is tough because there are weighting factors involved. Playing a #100 and #300 team is pretty much the same for UConn, but it is much different for a #200 team. But to create an SOS system that fairly weights SOS for all teams is not easy. As UcMiami notes, weighting values for teams close to a team would be better, but there are still many factors involved with any range since the difference between a #16 team and a #35 team is huge compared to a #101 and #120, and there are always gradations involved. Also, some would say that allowances also have to be made for the fact that you can play a #5 team early on that loses its top two players and slides back to #50 by the end of the season.

I have no idea exactly how SOS is compiled for Sagarin because much of how Sagarin works is shrouded in a bit of "special recipe" secrecy, and anyone who tells you they know exactly how it all works is lying to you unless they are Jeff Sagarin and maybe Jim Sukup (not sure if he just compiles game results or works also behind the scene with the system). I am NOT saying that the Sagarin system merely adds up the opponents' ratings and divides by the number of games to compile the SOS and SOS rankings, as he actually makes some adjustments but it is fairly close to the mean average of the opponents. For St. Joseph's, ND, and UConn, the current numbers are as follows:

...Official...............................StJ................................................ND......................................................UConn
Sag SOS Rank:.....................6....................................................24.........................................................25
Sag SOS Score:..................79.51..............................................76.41.....................................................76.37
....Computed estimates
Sag Opp Rank Avg:.....88.14............................................109.23....................................................110.00
Sag Opp Score Avg:.....78.35.............................................76.74.....................................................77.88

I have no idea what and why Sagarin makes adjustments, but it appears that St. Joe's gets a bonus for having a higher percentage of sub 100 ranked teams while maybe UConn is penalized a little for having the most over 150 ranked teams. Hard to say.

Note that if the #150 team played the St. Joe, ND, and UConn schedules and everything had gone true to form by the ratings (lose to a #149 team and beat a #151 team), it would be 1-13 playing St. Joe's schedule, 3-10 playing ND's schedule, and 5-10 playing UConn's schedule. So everything is relative, and though UConn has by far the toughest schedule for a top 10 team to play, a weak team would actually do better with it than it would with St. Joe's.
 
Last edited:
As of January 6, 2014, UConn is 10th in SOS, using this ranking service:
http://www.wbbstate.com/standings/14/sos

As of January 6, 2014, UConn is 25th in SOS, using this ranking service:
http://www.rpiratings.com/womrate.php

And frankly, UConn could be 100th in SOS and still be the overwhelming favorite to win the 2014 national title. At this point, Duke could finish #1 in SOS (something Coach P mentioned in a post-game press conference last week, when Duke was #1 in SOS) and it is akin to the Washington Mystics hanging the attendance banner in the WNBA.
 
CamrnCrz1974 said:
As of January 6, 2014, UConn is 10th in SOS, using this ranking service:
http://www.wbbstate.com/standings/14/sos

As of January 6, 2014, UConn is 25th in SOS, using this ranking service:
http://www.rpiratings.com/womrate.php

And frankly, UConn could be 100th in SOS and still be the overwhelming favorite to win the 2014 national title. At this point, Duke could finish #1 in SOS (something Coach P mentioned in a post-game press conference last week, when Duke was #1 in SOS) and it is akin to the Washington Mystics hanging the attendance banner in the WNBA.
Expand ...
Yeah, there are many different ranking and SOS values, and a fan of any team can shop around to find the best rank for his team and proclaim it #10 in SOS rather than #30. SOS tends to be more volatile than rankings when comparing systems, but even rankings can vary hugely as witnessed by the men's Syracuse team that is rated #1 in some systems listed in Vaporia and #22 in others. We can all argue about what ratings factors are the most valid, but at the end of the season none of these ratings right now mean anything.
 
CamrnCrz1974 said:
As of January 6, 2014, UConn is 10th in SOS, using this ranking service:
http://www.wbbstate.com/standings/14/sos

As of January 6, 2014, UConn is 25th in SOS, using this ranking service:
http://www.rpiratings.com/womrate.php

And frankly, UConn could be 100th in SOS and still be the overwhelming favorite to win the 2014 national title. At this point, Duke could finish #1 in SOS (something Coach P mentioned in a post-game press conference last week, when Duke was #1 in SOS) and it is akin to the Washington Mystics hanging the attendance banner in the WNBA.
Expand ...

As of today, 1/6/14 UConn is 7th in SOS using this ranking service
http://warrennolan.com/basketballw/2014/sos
 
ochoopsfan said:
As of today, 1/6/14 UConn is 7th in SOS using this ranking service
http://warrennolan.com/basketballw/2014/sos
Expand ...
In Sagarin it is Central Michigan with the top SOS and OSU is at #18, but CMU is only #32 in the Nolan ratings and OSU is #1. On the other hand, Nolan has Baylor at almost 200 spots better than Sagarin's position. Just goes to show how much different systems can vary. One thing both systems agree upon is that Gardner-Webb has a horrible SOS and is probably the weakest team with a winning record.
 
huskybill said:
Wow, I never realized how easy it was for us to get into the Big 10. What a bunch of dummies we must have running things at UConn. I too thought UConn should leave the AAC, but thought the best we could do was to start a new league made up of the strongest mid-majors. BTW, is that for just the women's basketball team, or can all the UConn sports teams get into the BIG 10 that easily?
Expand ...
The funniest post I have read in a long time.
 
.-.
The Washington Capitals had the highest SOS in their league because they played a number one undefeated team every game. :p:D
 
KnightBridgeAZ said:
The funniest post I have read in a long time.
Expand ...
I think it takes a Rutgers fan to find the heights of humor in HB's remark.

But he is wrong, as I think there is something about being a member of the AAU that is required for B10 admission. Or wait a minute, maybe it's the AAC that a school needs on its resume, in which case HB is right. B10 here we come!
 
He's not taking into account the poll rankings. We've played Duke, Stanford, Maryland, Penn St., California and we will play Baylor and Louisville twice. Not sure how he comes up with his rankings, but they have nothing to do with the polls.
 
UConn has never striven to have the top SOS, though some years like the last one they certainly did by some measures. Winning big games and always being solidly in the hunt for an NC or the overwhelming favorite in a few recent years is really what UConn is about. So getting at all worked up about why ND has a slightly higher rated SOS right now than UConn is not that sensible since the rating system was not developed for the Huskies' benefit and by the one-size-fits-all system for Sagarin the ratings for UConn are quite reasonable.

It is only down in Knoxville that the dictum that the #1 SOS makes champions is considered an indisputable truism, though getting either that top SOS or an NC has been a bit out of reach for the last 6 years.
 
The Massey ratings (click) has UConn's SoS as 1st, with an anticipated end of season SoS of 2nd.

Massey is an interesting one - at times it has done a good job of predicting margin of victory in games. You can see the predictions by clicking on the team names. I don't know how they calculate SoS, but for UConn to be number 1, elite opponents must be more heavily weighted, somehow?
 
jsloanjr said:
That is so true...This will be the issue going forward for UCONN....Weak AAC and most of the Top 10 teams want no part of UCONN in the future. Reason #1, Get the hell out of this AAC..Big 10 here we come.....
Expand ...

No thank you! Stay in the AAC, or go to ACC. :p
 
.-.
DobbsRover2 said:
It is only down in Knoxville that the dictum that the #1 SOS makes champions is considered an indisputable truism, though getting either that top SOS or an NC has been a bit out of reach for the last 6 years.
Expand ...
Was, not is.
 
iri10 said:
The Massey ratings (click) has UConn's SoS as 1st, with an anticipated end of season SoS of 2nd.

Massey is an interesting one - at times it has done a good job of predicting margin of victory in games. You can see the predictions by clicking on the team names. I don't know how they calculate SoS, but for UConn to be number 1, elite opponents must be more heavily weighted, somehow?
Expand ...
Massey? Now you're getting into something a little too cool, reasonable, and nuanced to be socially acceptable. Massey is set up to focus on relative comparisons that help the top elite teams instead of simply serving the average schlubs of the teeming masses.

But yes, Massey has an SOS system built much on the "good" SOS system described in the posts above, where a top team playing another top team gets an SOS reward and is not penalized so much for playing a Monmouth, while a poor team like Oakland does not necessarily get a great SOS (Massey has them at #202) for playing top teams that earn them the #27 SOS in Sagarin. Massey spelled it all out a decade and half ago as to what for a UConn fan would be considered a sensible SOS rating system. So UConn is now #1 in SOS, and with a sprinkling of Louisvilles and Baylors plus some decent AAC competition, it will be passed only by Duke by the end of the season according to projections.

That decent AAC competition does not extend to tonight obviously, as the Massey predictor has the score at 91-29 for UConn. Sagarin has the margin at around 45, so that will serve as an interesting comparison. One of the cool things about Massey it is very open about itself so that you can click previously to see past predictions. UConn was predicted as a 91-39 winner over Memphis in a 90-49 game. I'm not sure that Massey is a better predictor than Sagarin though, as Sagarin claims to have an 89% success rate for its ELO Chess ratings and Massey got about 77% correct on Saturday, though that might not have been a normal day. Some of the Massey archives stretch back decades, and you can see that it predicted UConn by 75-68 over ND in the semis and by 77-59 over Louisville in the NC. Not that much for WCBB overall though.
 
Last edited:
.-.
Status
Not open for further replies.

Forum statistics

Threads
168,506
Messages
4,579,325
Members
10,489
Latest member
Djw06001


Top Bottom