The real top 25 ( via computer rankings )

[rmilia1]

Took the Big 5 computer ranking systems ( RPI, BPI, Pomeroy, Sagarin, Massey ) and took the top 30 teams in all the rankings. Aside points to all teams based off of where they were ranked ( ie 30 pts for 1st, 29 pts for 2nd and on down the line ). I then added together the scores for ALL tems ranked in any of the 5 metrics and the top 25 is as follows

1. Arizona 140
2. Ohio State 135
3. Wisconsin 131
4. Kansas 126
5. OK State 124
6. Syracuse 122
7. Iowa State 119
8. Nova 117
9. Michigan State 103
10. UMass 91
11. Louisville 88
11. Oregon 88
13. Kentucky 84
14. Florida 82
15. Wichita State 80
16. Iowa 79
17. Duke 66
18. Pitt 65
19. Memphis 55
20. Gonzaga 53
21. Colorado 52
22. Creighton 48
23. Baylor 44
24. Mizzou 29
25. Tennessee 28

- Just missed Michigan 23, UNC 19, Illinois 19, Arkansas 18, FSU 15
- Little further back UCLA 12, LSU 12, Harvard 9, Boise 8, Xavier 6

 

[gordontrue]

Nicely done and a good idea.

 

[rmilia1]

An interesting note is that the RPI ( while the metric widely considered the most relevant metric when it comes to at large selections ) was BY FARRRR the biggest outlier of all the systems. The RPI had 7! teams ranked in its top 30 that were not ranked by ANY of the other rankings ( Illinois 13th, Harvard 22nd, Boise 23rd, Xavier 25th, North Dakota State 26th, Minnesota 27th, UCSB 29th ). Every toher metric had only 1 team ranked that wasnt ranked by any of the other systems ( VCU 30th- Pomeroy, St Marys 29th- BPI, UConn 26th- Massey, Cincinnati 29th- Sagarin). I think that goes to show what we have ALL believed for some time. The RPI is by far the worst of the metrics yet is used more, it seems, than almost all other evaluation tools.

 

[CatsTopPac]

I've gone with the BPI since its inception. It just gives a more complete picture than any other single marker, imo.

 

[SteelersPride]

glad to see pitt in there

 

[rmilia1]

I've gone with the BPI since its inception. It just gives a more complete picture than any other single marker, imo.

The BPI is, in fact, the closest single ranking system to the combined averages of the 5 systems I used. It deviates only an average 3.3 spots from the average of the systems combined with its largest variances being Baylor ( 23rd in combined 41 in BPI and Tennessee being 24 combined and 47 BPI ). It appears to be far and away the most accurate measure when compared to the group averages.

 

[rmilia1]

glad to see pitt in there

Pitt likely would have been higher BUT the Panthers are not ranked in the top 30 in RPI despite being no worse than 22nd in any of the other 4 rankings.

 

[Shanemansj13]

 

[SteelersPride]

funny pitt is top 9 in BPI, 18 in RPI, not ranked by espn power rankings, AP, or USA today

and an 8-9 seed by bracketologists

seems like no one knows what they got

 

[SteelersPride]

Pitt likely would have been higher BUT the Panthers are not ranked in the top 30 in RPI despite being no worse than 22nd in any of the other 4 rankings.

where are u getting your info i got pitt at 18 rpi, and not ranked in usa today and ap poll?
NCAA College Basketball Polls, College Basketball Rankings, NCAA Basketball Polls - ESPN

 

[rmilia1]

funny pitt is top 9 in BPI, 18 in RPI, not ranked by espn power rankings, AP, or USA today

and an 8-9 seed by bracketologists

seems like no one knows what they got

Used realtimerpi which was updated this morning. Pitt is 32nd

RealTimeRPI.com Men's College Basketball Rating Percentage Index (RPI) Ratings - A leading sports ratings and resources community on the Internet

 

[SteelersPride]

but they arent top 22 in any of the other rankings

 

[SteelersPride]

and im not saying that RPI is wrong just seems odd the RPI on espn that is daily updated and that site is off by 14 spots.....

 

[SteelersPride]

and the site your using actually shows pitt increasing their RPI

 

[rmilia1]

and im not saying that RPI is wrong just seems odd the RPI on espn that is daily updated and that site is off by 14 spots.....

Another reason RPI sucks. There are 3 different RPI sites and they all are different. Pitt is 11th in Pomeroy, 9th in BPI, 17th in Massey and 22nd in Sagarin just so you know

 

[SteelersPride]

ohhhhhhh those 4 lol no usa today or ap got yah...........sorry didnt know what you were using

 

[rmilia1]

ohhhhhhh those 4 lol no usa today or ap got yah...........sorry didnt know what you were using

Yeah not using the polls because the polls are inherently biased by name recognition and previous success. Just basing it solely off of ( supposed ) non biased metrics.

 

[TrollyMcTroller]

Took the Big 5 computer ranking systems ( RPI, BPI, Pomeroy, Sagarin, Massey ) and took the top 30 teams in all the rankings. Aside points to all teams based off of where they were ranked ( ie 30 pts for 1st, 29 pts for 2nd and on down the line ). I then added together the scores for ALL tems ranked in any of the 5 metrics and the top 25 is as follows

Good idea. Poor execution. All of the systems you've listed rate teams. The ranking is a much less meaningful number. Ranking will always show an even distribution of teams. Ratings show the true relative strength based on the metrics used. The top 5 teams might only be separated by .004 points, and the drop to 6th might be several points. But just using the ranking it appears that the difference between 1 and 2 is the same as 5 and 6 (which is incorrect)

To properly do what you're trying to do, you need to base your system off of the rating and then you need to normalize all of your data so that the top team has the same wight in each index, and the bottom team has the same wight in each index, then you can take your averages and get a much more meaningful rating.

I used to have an Excel sheet that would import the daily info from Sagarin and Kenpom and would give an average based on the Sagarin's ELO_CHESS and PREDICTOR, as well as KenPom's AdjO and AdjD (after normalizing all of the values) but I haven't updated it to pull the figures from this season.

When I put it together, BPI didn't update often enough, Massey didn't publish anything in an import-friendly format, and I just didn't want to fool with the RPI since it isn't even trying to evaluate the same thing that the other indexes are.

 

[rmilia1]

Yeah I agree there are more variables but did it at work so just couldnt get as into all the data as necessary. I utilized a point structure at least and not just an averaged ratings of the teamd

 

[Great Dayne]

Update