Boyd's World-> Breadcrumbs Back to Omaha-> Schedulability Revisited About the author, Boyd Nation

Schedulability Revisited

Publication Date: September 17, 2002

That Time of Year

I'm going to spare you all a rant this week, even though I'm scheduled for one. Baseball America just released their annual release of the official NCAA RPI, but, frankly, I just can't work up the venom to rave about the inaccuracies in them. Go read an old one if you need to; a few of the team names have changed, but the picture hasn't. Part of it is that you don't really want to hear it again; part of it is that the pseudo-RPI's are much closer to the real ones than they used to be (although not identical, oddly enough; I'm going to have to do some more work to see if they're slightly changing the formula each year).

There was an interesting note in a BA cover article that I won't bother linking to because it's in the subscriber area. It describes a recent meeting between Jim Wright, the NCAA's Director of Statistics, and the Mountain West coaches, including Tony Gwynn and Vance Law. Basically, Wright was encouraging them to improve their schedules in order to improve their RPI. In principle, that's fine. In practice, though, the MWC schools already play decent if not great schedules (BYU's ranked 68th last year, for example, by my measures). What they don't play are schedules that work well with the RPI's. In other words, Wright is indirectly encouraging them to do the things that I talked about explicitly in my piece last year on gaming the RPI rather than actually improve their schedules, and that leads into what I want to look at this week -- whether those schedulability factors I described worked well as a predictor for who the RPI would overrate or underrate.

Doomed before They Began

In this column, I listed the teams most likely to be overrated by the RPI based on the schedulability factors of their opponents, based on 2001 results. The main question was whether 2001 results would relate well enough to 2002 results for those predictions to be accurate. Here are the ten teams most likely to be overrated based on their schedules, with their final rank in the pseudo-RPI and ISR:

Team                     RPI    ISR

Delaware                  44    100
Temple                   201    219
Quinnipiac               274    278
Albany                   209    238
Massachusetts            165    209
Boston College            73    137
Hartford                 217    244
Arkansas-Pine Bluff      253    259
St. Peter's              264    269
Northeastern             163    207

Most of these show that being overrated isn't enough when you're also bad, but it's striking that every single one of them ranked higher in the RPI's than in the ISR's. Other significant teams in the top 20 were Rutgers (#45 RPI, #91 ISR), Seton Hall (#106 RPI, #161 ISR), and Elon (#49 RPI, #76 ISR).

How about the other end? Here are the ten I listed as most likely to be underrated:

Team                     RPI    ISR

Gonzaga                  170     94
Pepperdine               159     85
Portland                 197    110
Santa Clara              191     97
Oregon State              88     42
San Diego State           89     43
St. Mary's               204    119
Cal Poly                  91     58
San Diego                 87     39
Oral Roberts             133     70

You know, even I didn't expect it to be this striking -- again, every single one of these teams, which this method predicted as soon as I had their full schedule without taking into account any of this year's results to be underrated by the RPI's, were underrated by a significant amount. If you've gotten lost in the math and the links (or in that last sentence), the short version of this is that you can predict that a team will be substantially underrated just from looking at their schedule and their opponents' RPI numbers from the previous year, knowledge that could have been quite useful to the two San Diego teams last year, for example.

I'm not sure how much of a breakthrough it is to know that the RPI is flawed in predictable ways. I still stand by the statement that I made in the original gaming column that I don't want to see anyone intentionally using this information; I think if enough teams do it, it could do substantial damage to the game as a whole as some good teams have trouble scheduling good opposition. Nonetheless, I'm increasingly convinced that the RPI's got to go, and pointing these things out is one of the only ways I know of to work toward that. I'm also in the process of formulating a suggested replacement; more on that in future months.

This Year's Schedulability Factors

Here are this year's ten most schedulable teams:

Rank   SF     WP     OWP    ISR   Team

  1  0.0137  0.804  0.441  100.2  Southern
  2  0.0131  0.745  0.489  103.2  Marist
  3  0.0128  0.803  0.519  113.6  Richmond
  4  0.0128  0.679  0.488   99.6  Maine
  5  0.0128  0.678  0.362   93.1  Delaware State
  6  0.0127  0.698  0.367   96.0  St. Bonaventure
  7  0.0127  0.596  0.414   88.8  Central Connecticut State
  8  0.0127  0.588  0.451   89.8  Monmouth
  9  0.0126  0.627  0.436   93.5  Manhattan
 10  0.0125  0.686  0.430  100.5  Pittsburgh

And here is the whole list.

If you're interested in reprinting this or any other Boyd's World material for your publication or Web site, please read the reprint policy and contact me


Boyd's World-> Breadcrumbs Back to Omaha-> Schedulability Revisited About the author, Boyd Nation