Pages

More Comparisons

2/23/2006

Check out the tournament resumes for the following four teams with identical records.

Team A Top Wins Losses
W-L 20-6 a#18 n#5
RPI 11 h#18 h#33
SOS 18 a#23 a#35
Last 10 6-4 n#27 h#35
W-L vs:   h#62 h#66
1-25 3-1 h#87 a#95
26-50 1-3  
51-100 3-2    
Team B Top Wins Losses
W-L 20-6 a#21 n#4
RPI 28 n#40 h#17
SOS 73 a#45 a#35
Last 10 6-4 h#45 h#35
W-L vs:   h#60 h#62
1-25 1-2 h#82 a#82
26-50 3-2  
51-100 2-2    
Team C Top Wins Losses
W-L 20-6 a#15 h#1
RPI 39 h#58 N#13
SOS 96 a#69 h#26
Last 10 8-2 h#69 a#47
W-L vs:   h#74 a#59
1-25 1-2 n#91 a#142
26-50 0-2  
51-100 5-1    
Team D Top Wins Losses
W-L 20-6 a#16 h#1
RPI 22 h#34 h#12
SOS 59 a#52 n#19
Last 10 8-2 h#52 a#42
W-L vs:   h#67 a#54
1-25 1-3 a#86 a#113
26-50 1-1  
51-100 6-1    

Team A is UCLA using RPI data. Team B is UCLA using JCI data. Look at the boost that UCLA is getting from the RPI data. #11 RPI, #18 SOS. Two wins against Arizona (#18 in the RPI) look like quality Top 25 wins. Based on that data, looks like they're a solid 4 seed.

Team C is Boston College using RPI data. Team D is Boston College using JCI data. BC's RPI profile is not very flattering: #39 overall, #96 SOS, only 1 top 50 win. However, their JCI paints a slightly rosier picture.

The variation in the RPI numbers are purely artificial, simply an outcome of inaccuracies in the RPI formulas and methodology. And even if the Selection Committee understands the basic shortcomings of the RPI, they won't be able to account for and adjust for each of the discrepancies that filter down to all of the RPI-based factors they look at. For instance, they might be able to recognize that UCLA's RPI is inflated, but they can only guess how far the number is off. And I doubt that they'd be able to reach any conclusion that would lead to Boston College being placed higher than UCLA, although the true rankings suggest that this should be the case.

Boston College is another victim of the 'bottom feeder' effect that I talked about earlier. BC played five games against opponents ranked above 250th while UCLA didn't play any (maybe the Pac-10 has figured this out, because Arizona did the same thing). UCLA's RPI SOS is 18th and BC's is 96th even though it would be tougher for a team to go 20-6 against BC's schedule than it would against UCLA's.

0 comments:

 
Wegoblogger #31 © 2011 | Designed by Bingo Cash, in collaboration with Modern Warfare 3, VPS Hosting and Compare Web Hosting