As promised in my last article, here is an explanation of the new NET Rankings. First the first off, NET the stands for NCAA Evaluation Tool, and it is just that, the primary sorting tool for the selection committee for the Men’s DI basketball tournament. It replaces the ratings percentage index, or RPI. The RPI was controversial because it was limited by the data used to calculate it.
As a reminder, the RPI, which was created in 1981, only took into account game results and had a weighting factor of opponent’s winning percentage and opponent’s opponent’s winning percentage in all DI games. Games against non-DI teams were ignored. There was one positive factor of the RPI – there was no pre-conceived strength of schedule calculated. The strength of schedule matured as the season went on. Consequently RPI ratings early in the season did not correlate with the eye test, but by the time conference tournaments rolled around the RPI took relative shape up. The major issue with the RPI was good teams in bad conferences were penalized because of the weight of opponent’s winning percentage and opponent’s opponent’s winning percentage. There was enough outcry from the college basketball community about the RPI that the NCAA sat down with top analytics experts and devised a new system.
The result, the NCAA Evaluation Tool, is a combination of results-based data, like the RPI and Sagarin Ratings, and efficiency-based data, like KenPom Ratings. While the actual algorithm is proprietary, the new tool takes into account several factors. They include the team value index, net efficiency, adjusted winning percentage, and scoring margin. The team value index is based on game results against DI opponents only and weighted by opponent and location. Net efficiency is a calculation based on offensive points per possession minus defensive points for possession. Adjusted winning percentage is a calculation in that it values road wins by a factor of 1.4 and home a wins by a factor of 0.6. How those factors were established is not clear. Scoring margin is also used in the calculation but is capped at 10 points and all overtime games are capped at 1 point.
The NET Ranking will also take the place of the RPI in the BracketBluster calculations. For those of you new to BracketBluster, here’s a quick primer on how our rankings are produced. We look at several data sets, including both results-based data and the efficiency-based data. We take the data from last season and create an algorithm that when applied to the data outputs last year’s bracket (or as close to it as possible – we couldn’t find an algorithm that kept Saint Mary’s out of the tournament or had Loyola below a 7 seed – the Ramblers were an 11 seed) then we apply that algorithm to this year’s data, then applying the NCAA rules of bracketology, we produce a bracket. In theory if the committee uses the same logic from year to year our bracket will be very close to the actual, but if they vary their thought process, our results will be off. The challenge will be how much weight the committee gives the new NET Rankings as opposed to the RPI.
From now until selection Sunday we will produce a weekly seed list. We hope you enjoy the First and Fan’s BracketBluster and participate in the conversation. View our first seed list here.