| View previous topic :: View next topic |
| Author |
Message |
pilight
Joined: 23 Sep 2004
Posts: 66911
Location: Where the action is
Back to top |
 Posted: 12/02/16 11:57 am ::: |
Reply  |
|
Conference RPI top 10
1 SEC
2 Pac 12
3 ACC
4 Big 12
5 American
6 Big East
7 Big 10
8 CAA
9 WCC
10 MAC
_________________
I'm a lonely frog
I ain't got a home
|
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
FollowtheCardinalRule
Joined: 12 Oct 2011
Posts: 5153
Location: Denver
Back to top |
| Posted: 12/06/16 12:07 am ::: |
Reply |
|
| So, can someone explain to me what made Boise St ant Utah jump so high?
|
|
Fighting Artichoke
Joined: 12 Dec 2012
Posts: 4040
Back to top |
| Posted: 12/06/16 9:17 am ::: |
Reply |
|
| FollowtheCardinalRule wrote: |
| So, can someone explain to me what made Boise St ant Utah jump so high? |
The RPI is 25% your winning percentage, 50% your opponents winning percentage, and 25% your opponents' opponents' winning percentage. Early in the season, you can have a very high RPI by beating up on bad teams that have had a very easy schedule so far (garnering gaudy winning percentages), thus giving you a very high rating. But as the season progresses, teams eventually have to play decent teams, or at least their opponents do, and thus their RPI drops.
Utah and Boise State are undefeated thus far, and have appears to have played (in the RPI system's eyes) a tough schedule. That will change unless they are truly good and their schedule is actually challenging.
Opponents of the RPI note that 75% of the rating is based on the strength of your schedule rather a team's own success with that schedule. These opponents have an excellent point. The RPI is useful as an indication of SOS, but is not very useful in determining the best team.
|
|
pilight
Joined: 23 Sep 2004
Posts: 66911
Location: Where the action is
Back to top |
| Posted: 12/06/16 9:25 am ::: |
Reply |
|
| Fighting Artichoke wrote: |
| Opponents of the RPI note that 75% of the rating is based on the strength of your schedule rather a team's own success with that schedule. These opponents have an excellent point. The RPI is useful as an indication of SOS, but is not very useful in determining the best team. |
Opponent's records have far less differentiation than individual team records, so while the SOS provides 75% of the rating in theory in practice a team's own record is a far greater determinant of its RPI.
RPI has pretty consistently underperformed other computer based rankings in my annual tournament competition.
_________________
I'm a lonely frog
I ain't got a home
|
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
| Posted: 12/06/16 10:04 am ::: |
Reply |
|
I'm not hating on ND but I'm genuinely curious how
GB + WAS +IOWA = #2 SOS
There must be some built in element
_________________
Nnekalonians 1:14 - Thou shalt not accept that which is not earned
|
|
purduefanatic
Joined: 10 Aug 2011
Posts: 2819
Location: Indiana
Back to top |
| Posted: 12/06/16 10:40 am ::: |
Reply |
|
| Shades wrote: |
I'm not hating on ND but I'm genuinely curious how
GB + WAS +IOWA = #2 SOS
There must be some built in element |
Here's why (and why RPI means nothing this time of year):
Central Michigan - 6-2 RPI #171
Fordham - 6-2 RPI #69
Green Bay - 6-1 RPI #11
Washington - 8-1 RPI #22
La Lafayette - 2-2 RPI #78
TCU - 5-2 RPI #119
Iowa - 6-3 RPI #71
Valparaiso - 3-4 RPI #250
That translates into 5 Top 100 wins, including 2 Top 25. In addition, the only squad that has a losing record is Valpo, at 3-4. When you beat teams that are also winning, it boosts up the RPI and thus the SOS at the same time.
|
|
Fighting Artichoke
Joined: 12 Dec 2012
Posts: 4040
Back to top |
| Posted: 12/06/16 10:51 am ::: |
Reply |
|
| purduefanatic wrote: |
| Shades wrote: |
I'm not hating on ND but I'm genuinely curious how
GB + WAS +IOWA = #2 SOS
There must be some built in element |
Here's why (and why RPI means nothing this time of year):
Central Michigan - 6-2 RPI #171
Fordham - 6-2 RPI #69
Green Bay - 6-1 RPI #11
Washington - 8-1 RPI #22
La Lafayette - 2-2 RPI #78
TCU - 5-2 RPI #119
Iowa - 6-3 RPI #71
Valparaiso - 3-4 RPI #250
That translates into 5 Top 100 wins, including 2 Top 25. In addition, the only squad that has a losing record is Valpo, at 3-4. When you beat teams that are also winning, it boosts up the RPI and thus the SOS at the same time. |
Exactly, purduefanatic. The records of those teams will even out as the season progresses unless they are truly good teams. But Notre Dame's RPI will not drop precipitously because their conference schedule will help them.
|
|
ArtBest23
Joined: 02 Jul 2013
Posts: 14550
Back to top |
| Posted: 12/06/16 11:06 am ::: |
Reply |
|
| Shades wrote: |
I'm not hating on ND but I'm genuinely curious how
GB + WAS +IOWA = #2 SOS
There must be some built in element |
The RPI SOS score is 2/3 a team's opponents' winning percentage and 1/3 its opponents' opponents' winning percentage. ND's opponents - the largest element - are 42-17, which is high for this point of the season. For comparison, UConn's opponents are a combined 39-21. The RPI doesn't care what a team's name or AP ranking is, and three of UConn's opponents have losing records compared to one of ND's. Compare Maryland which has a SOS of 124. Why? Well five of their opponents have losing records, two have only one win, and their combined record is 26-32. The record of #2 LSu's opponents is 42-15, so their opponents' opponents records must be slightly worse than ND's opponents' opponents. I didn't dig that deeply.
It's really early yet. Limiited data produces strange results and wild swings. Teams can move daily based on a single opponent's opponent losing a game at this point in the season. Records of opponents will be more realistic and reflective of quality by the end of January or so, and become even more sensible and reliable the more games are played. The biggest distinguishing characteristic (and conscious choice) in the RPI is to exclude consideration of margin of victory that other models include. It's just simple math.
|
|
pilight
Joined: 23 Sep 2004
Posts: 66911
Location: Where the action is
Back to top |
| Posted: 12/06/16 11:31 am ::: |
Reply |
|
| ArtBest23 wrote: |
| The biggest distinguishing characteristic (and conscious choice) in the RPI is to exclude consideration of margin of victory that other models include. It's just simple math. |
Victory margin was excluded to discourage teams from running up the score against overmatched opponents, not to make RPI a more accurate measure of team strength.
_________________
I'm a lonely frog
I ain't got a home
|
|
ArtBest23
Joined: 02 Jul 2013
Posts: 14550
Back to top |
| Posted: 12/06/16 11:59 am ::: |
Reply |
|
| pilight wrote: |
| ArtBest23 wrote: |
| The biggest distinguishing characteristic (and conscious choice) in the RPI is to exclude consideration of margin of victory that other models include. It's just simple math. |
Victory margin was excluded to discourage teams from running up the score against overmatched opponents, not to make RPI a more accurate measure of team strength. |
I understand that.
But including MOV has its own problems because it doesn't take into account pace or other style-of-play factors. Relative MOV often has nothing to do with relative strength.
Consider for example two teams that play an identical 30 team schedule and win all 30 games. At the end of the season they should have the identical RPI. But team A plays a methodical half court offense at which they excel, and on defense generally drops back into a tightly packed zone. Team B isn't worth a damn in the half court, scores mostly in transition and from the arc, and never zones, playing a full court man press for forty minutes. As a result primarily of their differing styles, Team A averages 55 ppg and has an average MOV of 12, while Team B averages 84 ppg, and has an average MOV of 26. So at the end of the season, Team B is probably higher in nearly every computer and human poll. Their MOV is higher and they probably have a higher "eye test". But are they actually any stronger? Does that higher MOV realistically reflect a probability that B would defeat A? I doubt it. It's probably a toss up.
Or consider two coaches, both of which have a roster with nine HS AAs. Coach X's approach is to stick with the nine on her roster, while Coach Y likes to fill out her roster to 15 with some projects and walk ons. When X gets a big lead, there's no choice but to keep playing starters and other AAs. When Y gets a big lead, she empties her bench which likely holds down the MOV. All else being equal, X probably has a lot higher avg MOV. But the reality is that if X plays Y, those last 6 players that drove the difference in avg MOV will never take off their warmups and are completely irrelevant.
Some of those factors can be reduced by capping the MOV considered in the model, but then you have to arbitrarily choose where to set that cap, how to factor in MOV, and in the process you necessarily create other distortions.
There's no perfect system.
The RPI was developed by the NCAA for use in tournament selection and seeding, is used in many of its tournament championship selection processes, not just basketball, and as you say the driving factor for excluding MOV was to keep teams from running up the score, or as the NCAA terms it, from "manipulating" scores. But there's no single "best" way to factor in MOV, and excluding it or including it both create distortions. The RPI was never really intended to be used as a real time substitute for the AP poll or the like. It's just another number.
|
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
elsie
Joined: 08 Apr 2016
Posts: 276
Back to top |
| Posted: 12/10/16 12:46 am ::: |
Reply |
|
I see tonite that Gonzaga is rated below GB and SDS in the mid majors........how can a team who has beaten two Pac 12 teams as well as loose to a national power such as FSU and another Big 10 team such as Michigan not have a very high RPI?
the only big team that GB has played is ND, and they lost, although it was close...
is it better to keep playing virtually nobody just to keep our "wins" up?.
|
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Fighting Artichoke
Joined: 12 Dec 2012
Posts: 4040
Back to top |
| Posted: 12/10/16 9:28 am ::: |
Reply |
|
| elsie wrote: |
I see tonite that Gonzaga is rated below GB and SDS in the mid majors........how can a team who has beaten two Pac 12 teams as well as loose to a national power such as FSU and another Big 10 team such as Michigan not have a very high RPI?
the only big team that GB has played is ND, and they lost, although it was close...
is it better to keep playing virtually nobody just to keep our "wins" up?. |
In the RPI, the winning percentage of the teams you've played greatly affects your rating (especially early in the season). While Gonzaga has played 4 top 100 teams, they have also played many teams that are ranked low and have poor records, which lowers your RPI (3-4 Nicholls St., 1-7 UC Irvine, 1-8 Winthrop, and 0-7 Presbyterian). Green Bay's only opponents with losing records are 5-6 Chattanooga and 2-6 Bradley.
Green Bay's only loss was to Notre Dame, and it was in South Bend and by only 4 points. Last night they beat South Dakota State by 25 in Brookings. Green Bay is ranked 10th by both Sagarin and Massey whereas Gonzaga is rated 23rd and 39 respectively. If the Zags beat Northwestern on Monday, they should move ahead of SDSU into 2nd in the midmajor rankings.
|
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
tfan
Joined: 31 May 2010
Posts: 9624
Back to top |
|
Shades
Joined: 10 Jul 2006
Posts: 63770
Back to top |
|
|
|
