Conference Rankings Revisited

Tinman17 — Thu, 20 Mar 2008 07:45:03 -0000

Pac-10 Dominance Confirmed

As everyone is counting down the minutes to the first tip-off of madness while putting the finishing touches on their brackets, here is some good old-fashion data to chew on.

Bruinhoo posted a very interesting diary on March 1st entitled "Does Conference RPI tell us Anything Meaningful?" that he mostly got ripped for. Bruinhoo had the insight to throw out the highest and lowest RPI scores from each conference in order to eliminate the outliers that may have been affecting the average RPI for each conference and skewing the result. While this may seem academic and pointless to most, it was great to see that I was able to find someone of greater than, or equal, sports nerdiness right here in the Bruin Nation.

You see, over the past 4-5 years I've been tinkering with a College Basketball Tournament Bracket Calculator that assigns a raw score to each team participating. The team in each game with the higher raw score wins the game and so-on until you have a tournament winner and with any luck office bragging rights and a fist full of $20s.

My calculator is heavily dependent on quantifying and ranking all the tournament conferences because my theory is that a team will be most prepared for the tournament, and most likely to win any individual tournament game, if they have been winning games consistently in a competitive conference all year. I believe that the strength of your conference means more, in terms of doing well in the tourney, than just your strength of schedule. Basically, the whole is equal to more than the sum of the parts when it comes to testing teams for a tournament atmosphere.

By this point you are probably staring cross-eyed at your computer saying to yourself, "Why do I care"? Well, you care because with some inspiration from Bruinhoo and some Excel assistance from my accountant wife, I was able to confirm what you probably already know; the Pac-10 is FAR superior to any other conference this year.

Now I stopped using conference RPI a few years ago because there are some real statistical problems with it and also because it's difficult to work with a four digit decimal number. I use the far superior data from TeamRankings.com to rank the conferences but like Bruinhoo, I've wanted to get rid of the outliers in a meaningful way. I didn't want to necessarily just throw out the highest and lowest teams from each conference because those teams may be perfectly representative of that conference. What I ended up doing was using a common statistical definition of an outlier (++defined below) to throw out those truly non-representative teams like the putrid, Lavin-Loving Beavers of Oregon State.

Here are the results for the top 11 conferences (ranked according to my adjustments. Sorry first time poster the formatting stinks):

Rank. Conference - Orig Power Ranking - Tin Power Ranking

1.Pacific 10 - 85.3 - 87.2
2.Big 12 - 84.3 - 83.4
3.Big East - 83.1 - 83.1
4.ACC - 84.5 - 82.5
5.Big 10 - 82.2 - 82.2
6.SEC - 82.1 - 81.2
7.Missouri Valley (MVC) - 80.6 - 80.4
8.Atlantic 10 - 79.7 - 79.7
9.MWC - 77.8 - 77.8
10.Horizon - 77.6 - 77.6
11.Conference USA - 76.7 - 76.4

What you can see (besides that Memphis' conference is a total joke) is that our Pac-10 is way above the pack this year when the Beavers are thrown out. The average difference between the adjusted rankings (not counting the difference between 1 & 2) is .78, meaning that the difference between the first and second conferences is 4.4 times larger than the average distance between 2-11. Said another way, the difference between the Pac 10 and the Big 12 (#1 & #2) is just about the same as the difference between the Big 12 and the Missouri Valley Conference. Quick, who would you rather the Bruins play in the tourney: Kansas & Texas (top two Big 12) or Drake & Illinois St (top two MVC)?

As you can see, most conferences did not even have a true outlier as the original power ranking and adjusted ranking are the same. Of some interest, both Duke and NC were thrown out for the ACC adjustment which is the reason that one fell from being ranked 2nd, to ranked 4th. I actually expected Conference USA to fall off the map but their lowest team, Rice, got tossed as well as Memphis.

So what does this all mean? Well, UCLA won the regular season and conference tournament in by far the most demanding and stacked conference in the Country. That's about it. Does it mean they'll win it all? We don't know because my calculator deals with statistics which means that it will never guarantee an outcome and it'll be 30 years before I even know if it works properly, which kinda sucks.

We know that our Warriors are ready and prepared but we love this game, our Bruins and this tournament because no one can ever predict the outcome of game with 100% accuracy. That's why tonight, we need to focus our cumulative energies on trouncing by far the worst team (statistically) in this year's tournament!

++ An outlier for the purposes of this post is defined as 1.5 x (IQR) outside the 25th or 75th percentile with the IQR (interquartile range) defined as the difference between the 75th and the 25th percentile. EX: If the 25th percentile of a data set is 20 and the 75th percentile is 30 then anything below 5 or above 45 would be an outlier (IQR = 30 - 20 = 10. 10 x 1.5 = 15. 20-15=5. 30+15=45).

SB Nation User Blog: Tinman17

Conference Rankings Revisited