What Schools Recruit the Best, the Worst and Perhaps a Bit too Well?

For the final entry in our college basketball recruiting series we have taken a look at how well different schools recruited for the period from 2002 to 2011.This is the culmination of our other analyses that looked at factors that are expected to affect recruiting such as a team’s fan base support and ability to convert recruiting hauls into draft picks.  In this last entry we take a look at how schools recruit versus how we would expect schools to recruit.

What do we mean by “expect schools to recruit?”  Basically, our premise is that recruits are interested in playing for teams that have supportive fan bases, play in high profile conferences, are successful on the court, have significant financial resources, produce NBA players and have storied histories.  Our analysis begins with a model that predicts recruiting results (we use Rivals recruiting points as the dependent variable) as a function of these factors (revenues, last season winning rates, previous NCAA tourney appearances, previous final fours, recruit conversion into draft picks, conference, etc…).

We then compare a school’s actual recruiting results with the model’s prediction for each year in the data.  We then look at the ten year average of the difference between the actual and the predicted results (the residuals) to classify schools as over and underachievers. Because our results have the potential to stir up emotions, before we get into specific results we should make a couple of points clear.  First, the meanings of over achieving and under achieving recruiting results can be interpreted in multiple ways.  One interpretation is that schools (and coaches) that “over” achieve do a great job in attracting recruits.  However, given that the model controls for factors such as winning rates, being on the list of over achievers can also imply that the school underachieves on the court with the given talent.  Likewise, at the bottom of the list, “under” achieving can be interpreted as either lousy recruiting or an ability to get the most out of recruits.

The Top 10 list for the high majors is led by Texas at number one (I can almost hear Texas fans saying that this proves that Rick Barnes is a poor game coach), UCONN at 2, Florida at 3, Villanova at 4 and Memphis at 5.  Duke was number 10.

At the very bottom of the list of high majors we have Boston College, Houston and Arkansas.  In the cases of Boston College and Arkansas, these are fascinating results.  These schools regularly make the tournament and win games.  They just don’t seem to be able to draw elite recruits.  If I am a college AD looking for a new coach, I would take a close look at the coaches at these schools.  Perhaps these are coaches that if surrounded by super start recruiters could build elite programs.

While we aren’t going to spend much time on the mid majors in this analysis, our analysis did yield one very interesting finding for this group.  The school at the very bottom of the list is Butler.  Again, this is a result that can be spun in either direction.  Perhaps Brad Stevens is truly a basketball savant who can succeed with any players.  Alternatively, maybe schools like Illinois and UCLA dodged a bullet because Stevens would not have been able to recruit at the high major level.

Finally, maybe the most interesting element of our analysis is that we are able to identify recruiting results that are statistically unlikely.  If we agree that our model captures the key drivers of recruiting (expenditures, revenues, past success, current success, conference affiliation, conversion of recruits to NBA picks, etc…) then exceptional recruiting hauls should be a bit troubling.  These unusual results mean that either a given coach or program have a “specialness” not included in the model.  We will let readers speculate as to what this “specialness” might be. Our list includes three programs: Kentucky, Texas and Villanova.

The Kentucky results are especially dramatic.  Our calculations (which are a bit of the back of the envelope variety) suggest that the probability of Kentucky’s results occurring by chance is just 1%.  But again, we do acknowledge that there may be something special about this program that our model doesn’t capture.  However, we should also note that we do not find a similar “specialness” for schools such as North Carolina, Kansas, Duke and UCLA.  And to take things just a step farther if we just look at John Calipari’s results across Memphis and Kentucky our estimated probability of his recruiting results is less than .1%.  As before we acknowledge that we may be omitted a variable or two that captures coach Calipari’s recruiting gifts, but our model doesn’t identify other high powered recruiters such as Thad Matta, Bill Self or Coach K as outliers.

Mike Lewis & Manish Tripathi, Emory University 2013

College Basketball Recruiting Series

One of my (Lewis) favorite things in sports is college basketball recruiting.  Given the growth of the recruiting guru industry, it’s safe to say that I’m not alone in my fascinations.  For example in the case of the University of Illinois, If you took a look at the message boards you might think there is as much interest and speculation about the recruitment of Cliff Alexander (the number 3 ranked player in the 2014 class) as there is in the this year’s team.

Over the next couple of weeks, our plan is to take an in-depth, data-based look at the world of college basketball recruiting.  Our emphasis will be on judging how well teams really recruit and whether players make rational decisions about where to play ball.  As always, the key to these analyses will be that we will use statistics and data to go beyond the conventional wisdom and drill down to the fundamental issues.

As a starting point for our series, we are re-running an earlier analysis that looked at fan support across teams. This study is important for two reasons.  First, intuitively we expect that players will be more attracted to programs that have strong support.  This is a rational criterion because support likely translates to plentiful resources and television exposure.  Second, this study highlights the nature of our approach to these studies.  Rather than rely on simple metrics such as attendance, that are a function of team performance we examine fan support after controlling for short-term fluctuations in team performance.  In other words, we control for the fact that it is easy to be a Duke or Kansas fan, while it takes real character to support a team that may struggle on the court (e.g. Maryland & Illinois).

We have four analyses planned.  As noted the first one focuses on the “fan equity” enjoyed by each teams.  These rankings provide a sense of the customer or brand equity of each team.  The second analysis will take a look at each school’s ability to produce NBA draft picks as a function of their recruiting rankings.  This is something that recruits should definitely consider.  The third analysis will examine draft pick production as a measure of team success.  This analysis really gets at the value of choosing a high profile, blue blood program.

The fourth analysis is probably the one that we are most enthusiastic about.  In the fourth study, we examine recruiting success after controlling for a myriad of factors such as current winning percentage, markers of historical success and financial investment.  As we will discuss later this analysis as some significant implications for how we should evaluated coaches and may even provide some evidence that some teams recruit “too” well.

Mike Lewis & Manish Tripathi, Emory University 2013.