NCAA Says Goodbye to the RPI

Let’s just get the overused joke over with. R.I.P. RPI. This week, the NCAA announced that they will be retiring the Rating Percentage Index (RPI), the primary tool used by the NCAA Tournament Selection Committee since the metric’s inception in 1981. Taking its place is the newly created NCAA Evaluation Tool (NET). Very clever, I know.

RPI was an outdated system which was pretty arbitrary to begin with. A combination of winning percentage and strength of schedule that focuses solely on wins and losses, RPI rankings gave a decent ballpark figure of a team’s ability, but was highly flawed, as I wrote back in March. 

In short, by being so heavily based upon strength of schedule and disregarding point differential, the system becomes more about scheduling and less about team performance, generally favoring power conferences and at times leading to puzzling evaluations of smaller schools.

The new NET system will factor in wins, losses, location, scoring margin, offensive and defensive efficiency, and quality of wins and losses. NET is a predictive metric, similar to the rating systems used by statisticians such as Ken Pomeroy and Jeff Sagarin. Last year’s quadrant system will return, now using NET rankings instead of RPI.

This new method of evaluating teams seems like a big improvement, although the exact formula is unknown and we’ll need to wait until the season to see it in action. I do have one problem with it, however.

NET’s margin of victory for a single game is capped at 10, meaning larger victories still factor in as 10-point wins. This doesn’t make any sense to me, as a 30-point victory is completely different from what could’ve been a back-and-forth game the whole way that ended as a 10-point win after a failed attempt at extending the game by fouling.

On the other hand, the NCAA made a good decision by labeling all overtime games as one-point victories, considering the teams were deadlocked after 40 minutes.

There were a number of other factors that could have been used for NET, but I think it’s fine to let the committee decide how to treat things like early season vs. late season results, travel, and injuries. Each team will have different situations, and I think it’s very difficult to put a number on the impact of these factors.

However, we do have to make sure the committee considers them, and takes a look at other predictive metrics besides the NET. Over the years, despite claiming to take a team’s entire resume into account, the committee has consistently favored RPI over everything else, even when it leads to conclusions most fans and experts alike disagree with. The committee has to use a holistic approach rather than a “nothing but NET” mentality.

Overall, it’s good to see the NCAA become more progressive recently in changing the way they evaluate teams. Undoubtedly, over time, the metrics at our disposal will become only more refined. To repeat what I said back in March, there is no perfect way to judge a basketball team, and there will never be. But if the goal is to create a fair system that results in the best 68 teams being selected for March Madness every year, NET is a step in the right direction and a win for the sport.