Eliminating the NCAA Selection Committee

Every Wildcats fan should know Jon Scott, founder of The Kentucky Basketball Statistics Project.

The amount of UK basketball information there is astounding, not to mention the perfect encyclopedia for every bit of UK history one can find.

We have the honor of having Jon on for a guest post on possibly eliminating the NCAA Selection Committee. Enjoy!

***

Introduction

Over the past month, UK Athletic Director Mitch Barnhart has been named a member of a future NCAA selection committee, while John Calipari will be on a steering committee.  These appointments have led to a surge in interest in the NCAA selection process, and how it might be improved.

Calipari himself has been busy suggesting possible changes; some brilliant, some ridiculous.  They range from moving the SEC tournament back a day (to Saturday) to moving the tournament to November [when absolutely no one (other than UK-fans) in the football-crazed SEC will pay attention].

Cal has also talked about the need for greater transparency, the need for better guidelines for what is required to make the tournament and ultimately a fairer process, which are all great goals.

While discussion of ways to improve the process are helpful, most of them miss the biggest and most obvious issue which is the cause of most of the problems.  The main issue is the existence of the NCAA selection committee itself.

The Problem

The committee is composed primarily of career administrators, many with little to no actual basketball experience, and each with their own interests and biases, who come together for a short time to assemble and seed a field of teams.  This is done in secret, using a set of suggested guidelines which they may or may not choose to follow.

For Kentucky fans, who have many seasons seen their team's abused by the process, it's a great victory that they will finally have some representation on the committee in Mitch Barnhart (even though he must excuse himself from the room when UK is discussed).  But the mere fact that having a representative on the committee is almost a necessary requirement to receive a fair shake underscores just how dysfunctional the system has been.

One concern with respect to Kentucky and the SEC is the fact that in recent years, the committee has steadfastly ignored the results of the SEC tournament championship game, a game which is played on Sunday afternoon, finishing about 4 hours prior to the NCAA selection show.

For example, this past season, the committee kept Texas A&M as a #3 seed, with Kentucky receiving the #4 seed even though UK had just beaten the Aggies for the SEC tournament title and the teams essentially tied during the SEC regular season, with both being named regular season co-champion.

It would have been straightforward for the committee to lay out contingency plans, where the winner of the SEC tournament received the #3 seed, but the committee couldn't even be bothered to do that.

Instead, they ignored the result altogether.

In response, many have advocated moving the tournament final to Saturday.  But doing so doesn't necessarily solve the problem, and potentially creates a whole lot of new issues.

The committee has already demonstrated that it doesn't do its job by not considering all the games and accounting for all the results, but that's not the only type of mistake they've made.  Giving them additional time before the brackets are announced may or may not address the issue.  It's not a given that the results will be any better. This is especially true considering that the same politics and biases, and shifting criteria are still at play with a committee.

There are a number of new problems created in conjunction with moving the game to Saturday.

These include:

1.) by ceding Sunday afternoon, the conference loses a prime national TV audience with few distractions for college basketball fans.  Moving to Saturday it's not a given that a major network even has a time slot available, and if they do find or create one, it will be competing with a slew of other conferences who already hold their finals on Saturday; and

2.) moving the tournament up a day means one more weekday is used for the early rounds, making it less likely that fans will be able to take off work to attend the tournament.  Similarly, any students (including the players themselves) would need to miss an additional day of school.

Proponents of the move to Saturday mention that doing so would give the teams who play in the championship game an extra day of rest to prepare for the NCAA tournament.  That is certainly true, but what is not considered is whether that extra day of rest is really needed, or even beneficial.

Since the tournament expanded to 64 teams in 1985, Kentucky has played in the 1st round of the NCAA tournament on only 3 full days of rest (i.e. play in SEC championship on Sunday and play their 1st round NCAA match on Thursday) 11 times.  Their record in those games is 11-0!

During that same time, Kentucky has only lost twice in the 1st round of the NCAA tournament, once (vs. Ohio State in 1987) as a #8 seed on six full days or rest and the other (vs. Marquette in 2008) as an #11 seed on four full days of rest.

The truth of the matter is that a team who makes it to the SEC championship final is likely good enough that they don't require more than three days of rest to win a 1st-round NCAA game, against a low seeded opponent that is likely overmatched to begin with.

In other words, giving an additional day of rest prior to the NCAA tournament is not a problem that needs solving.  If anything the teams that need the most help are the ones who aren't strong enough to go far in the conference tournament, and because they exit early they get plenty of rest.

A Potential Solution

So what is the answer?  I propose that rather than try to accept the mistakes that the committee has been known to make time and time again and to make efforts to anticipate and circumvent these mistakes, to go directly to the heart of the matter:

Abolish the selection committee altogether!

In its place, I suggest adopting a computer model which uses as its inputs a wide range of factors, from power ranking (as determined by both computer models and human polls), to efficiency ratings to assessments of strength of schedule, etc.

The model should be developed (preferably by a reputable mathematician, not yet another NCAA committee) prior to the season and be sophisticated enough that it's not even worth trying to subvert the model, unlike the NCAA-created RPI which is hopelessly simplistic and easy to game.

It's clear that even taking a relatively simplistic compilation of five or six sources (for example Sagarin rating, Pomeroy rating, AP poll, Coaches poll, RPI, etc.) that one can combine the results, re-rank the teams according to their combined score and get a pretty clear ranking of the best teams from top to bottom.

From there, it's a simple step to assign seed lines and to find out where the bubble is and which teams would make the final cut and which ones don't.

Doing that, the results actually aren't much different than what the NCAA selection committee already produces. There are cases where the committee might have mis-seeded a team, but it's generally not more than 1 or at most 2 seed lines away from where the consensus would place them, and there are some cases where one or two teams make the cut which prove controversial.

Using a computer model instead will help to avoid some of those obvious mistakes, but overall will be largely similar to what the NCAA committee would do, so there wouldn't be a dramatic difference seen by the untrained eye.

But where a computer model would show a definite improvement is in terms of trying to keep the brackets balanced, by actually trying to follow the S-curve (something the committee is supposed to do but too often ignores.) And of course, since a model can be run in minutes, it makes it possible to truly account for all the games that are played, something that past committees regularly struggled with.

There are a number of other benefits of using a computer model:

1.) the model can be validated against previous season fields to see exactly what the results would be as compared to what the various selection committees chose.  These results can be analyzed to determine if any biases or unexpected results are evident. Because of this, the model can be adjusted accordingly to address any issues brought up, prior to use for the current season.

In other words, any proposed model can be thoroughly vetted and tested against previous seasons before it is actually put into use.

2.) As part of the model development, the factors which are utilized will be made known, so there is no mystery as to what is required to be rated high enough to make the tournament, or earn a high seed.  The system is completely transparent, which again is an area that the revolving committee (and their shifting points of emphasis) struggles with year after year.

3.) Because the computer model requires no human committee, it would be easy to run periodically during the course of the season (say once a week beginning part-way during the season once enough games have been played to have a statistically meaningful inputs).  This would give teams a running and realistic update of where they stand if the season were to end that day.

4.) As part of process of choosing sites for teams, the model would be bound by a number of constraints (for example avoiding two teams from the same conference from meeting prior to the Sweet Sixteen, ensuring BYU isn't scheduled to play on Sundays etc.) all while attempting to follow the S-curve and keeping the brackets balanced, while also attempting to place schools as close to their natural geographic sites as possible.

In effect, it turns into a complex multi-variate optimization problem, which computers are particularly adept at solving.

But there is room for the computer to take into account preferences provided by the schools themselves. For example, perhaps a team would rather participate in a particular 1st round location over being placed in a particular regional site during later rounds.  Perhaps a team would rather choose a location which is more convenient for their fans to attend via airplane or a bigger alumni base, than one which may be closer geographically but more difficult to reach. Perhaps a team would rather play on Friday-Sunday than on Thursday-Saturday.

While it wouldn't be possible to accommodate every school, where there are cases where there's no strong reason to place a team in a particular spot, their preference could be taken into account in many cases,  with the higher rated teams getting higher consideration for their preference.

Seeing mock results during the course of the season would help schools determine what scenarios they prefer, so it provides them an actual say in their placement, rather than having a committee assume what they might want.

Expected Criticism

The very idea of abolishing the selection committee is one that is such a dramatic departure from current custom, that there will sure to be plenty of resistance, even though it addresses nearly every problem experienced with the current system.

Some will say that the NCAA likes to control things and would not willingly give up the power to essentially design the tournament field.  And that may be true. But what really does the NCAA accomplish by holding onto this? They pull any number of administrators out of their day job at great expense to transport, house and feed them, taking them away from their families for multiple weekends in what at the end of the day is a thankless job which they are sure to be criticized for.

If the argument against going to a computer is that it robs the committee of the ability to create interesting made-for-TV matchups, maybe that's a prime example for why the committee needs to get out of the bracket business to begin with?

Television ratings are important, but if it involves placing a team where it shouldn't be, simply for the sake of boosting television ratings, then the system is patently unfair.

Besides, even with a blind draw, there are sure to be plenty of matchups which the TV networks will find intriguing.  The networks will do what they always do: search for the most interesting matchups available and then do their best to get them scheduled during a time which maximizes viewership.

Conclusion

Current suggestions for improving the NCAA tournament field and selection process, including such drastic actions as moving conferences tournaments to less convenient dates, all revolve around trying to correct and accommodate the numerous flaws of the NCAA selection committee.

The problem is that none of these ideas fundamentally address the real issue, and none guarantee that the selection process will actually improve.

Rather than dance around the edges and hope that the NCAA selection is both willing and capable of doing a respectable job, it's well past time to simply abolish the committee altogether in favor of a more accurate and ultimately fairer method of picking and seeding the NCAA tournament.

Adopting a well-designed computer model which utilizes multiple factors, whose formula is known and can be validated against previous season's tournament fields, will go a long way towards improving the transparency and fairness of the process for all.

The fact that it can run on a computer without the need for a human committee means that the time, effort and cost required to determine an accurate field is reduced dramatically, which removes the perceived need for conferences to finish sooner just so that their full resume will be recognized by the committee.  A computer also allows for new benefits, such as the ability to provide periodic updates during the course of the season, among other things.

I'd like to see this idea gain momentum among college basketball circles, and hopefully, persons with enough pull with the NCAA will take up this cause.  Even though it may sound extreme at first, it makes too much sense not to pursue this change.

And even if the NCAA ultimately vetoes it, just the public discussion of potentially abolishing the committee may by itself be enough to force them to do a better job and achieve a balanced and fair tournament, which ultimately is the goal anyway.