The Terrors of the RPI

Today I watched Jim Colton's slideshow on the Flaws of the RPI. I highly recommend it for anyone who is interested on the accuracy of the RPI. It is rather lengthy (Part 1, just on the RPI, is 21:30), but informative at the same time. Futhermore, in Part 2, explaining the Colton Index, I am completely sold.

To give a quick overview of the Colton Index:

Playing bottom feeders can kill your RPI. A team playing 1, 2, and 330 at home has the same SOS as someone playing 110, 111, 112 on the road. A tournament quality team would go 1-2 against the first schedule, and 3-0 against the 2nd schedule. Therefore, the 2nd team is rewarded for not playing weak opponents by appearing better than the 1st team.

The Colton Index fixes this. He ranks the teams by W-L record, and then uses probability to establish how many games a tournament quality team should win. Teams that win more than that number are rewarded, teams that win fewer are penalized. Since this changes all the rankings, the process is repeated until the numbers stabilize.

Obviously I haven't done as good of a job as the presentation I linked to. But I now believe that the RPI is flawed, and I will not be using it as much this year.

The Colton Index will also now be linked on the side.