2023 REPORT 4: CURRENT NCAA RPI versus BALANCED RPI IN RELATION TO NCAA TOURNAMENT AT LARGE SELECTIONS

Historically, the poorest ranked team to get an NCAA Tournament at large selection is #57 using the current NCAA RPI.  Thus I consider the candidate pool of teams for at large selections to be those ranked #57 or better.

To illustrate the difference between how the current NCAA RPI ranks teams in relation to the Tournament and how the Balanced RPI ranks them, I prepared the following table showing my simulated end-of-season Top 57 under the current NCAA RPI and under the balanced RPI.  Given that the table is based on a pre-season simulation that has the limitations described in 2023 Reports 1 and 2, I do not see the exact teams inside and outside the Top 57 under the two systems as something to focus on.  On the other hand, the general nature of the teams inside and outside the Top 57 is significant. And, it is consistent with what I expected given that the current NCAA RPI discriminates against teams from stronger conferences and regions and the Balanced RPI does not.  (See RPI: Modified RPI? at the RPI for Division I Women’s Soccer website.)

As a note based on past history, for the teams in the Top 57 under the current NCAA RPI but not under the Balanced RPI (yellow highlighting in table), some of them are likely to be Automatic Qualifiers and most of the others likely would not get NCAA Tournament at large selections.  Thus their dropping out of the Top 57 under the Balanced RPI, by itself, probably would not affect Tournament at large selection outcomes.  However, those teams are occupying spaces within the Top 57 candidate group that instead would be occupied by teams that are in the Top 57 under the Balanced RPI but not under the current NCAA RPI (blue highlighting).  Based on past history, some of those teams likely would get at large selections.  Thus those are the teams that are potentially hurt by the defects in the current NCAA RPI.

In the table, the URPI 50 50 SoS Iteration 15 Rank is the Balanced RPI Rank.

The table does not show, however, the likelihood that the blue highlighted teams in fact would get at large selections.  To explore that likelihood, I turn to the factor I use that is the best predictor of teams the Committee will give at large selections.  That is the paired factor RPI Rank and Top 50 Results Rank.  This paired factor, if used as the only basis for making at large selections, matches past Committee selections at a rate, on average, of all but two selections per year.  Thus it is a very good predictor of Committee at large decisions.

The following table uses this paired factor to show which teams likely would be treated the same and which likely would be treated differently under the Balanced RPI as compared to the current NCAA RPI.  As with the previous table, I do not see the exact teams affected by the change as something to focus on.  On the other hand, the general nature of the teams affected is significant.