RPI Madness

In the next couple of months you will be hearing a lot about RPI Rankings. RPI college basketball formulas are similar to the BCS college football rankings in how they are calculated. However, both methods are severely flawed.

To begin with, RPI rankings are all over the place. Anyone can come up with them, based on a series of formulas and statistics. We have even tried to come up with our own RPI (ratings percentage index) the past couple of years with no measured success.

Basically, you take a number of statistics (W-L against top 25, top-50, top 100 teams; W-L records at home, away, and at neutral sites; strength of team offense and defense; strength of schedule; the possibilities are endless) and apply a weight to each one. Everyone applies a different weight, or importance, to each stat. Each RPI is different, and, unfortunately, no matter what someone may tell you, no RPI is an accurate representation of any team.

Let's use the example of the top-5 teams in this week's Coaches poll. See below listed each team's updated RPI rankings in 4 different RPI polls:

1. Florida (32, 33, 16, 5)
2. UCLA (1, 1, 110, 6)
3. Wisconsin (7, 10, 23, 39)
4. UNC (5, 4, 12, 14)
5. Kansas (21, 21, 19, 32)

As you see, there is absolutely no consistancy, especially when you compare RPIs to rankings. The problem, of course, is the amount of variables and the importance you place on each variable. College basketball is not a controlled experiment in which you can take several variables and have a controlled situation. The amount of available variables is just too great. You could even break it down to how well a certain team does when the humidity inside the arena they are playing in is above or below 50%.

For this reason, the RPI means nothing. N-O-T-H-I-N-G. We proved this last year when we took 5 RPI rankings, averaged them, and used those averages to try and predict winners in each NCAA Tournament game. We did no better, and even worse, than the folks who based their predictions of wins on likings of colors of team uniforms.

So, laugh in the face of anyone who cites an RPI ranking for their decision to pick one team over another.