Nylon Calculus: The AAU to college player pipeline

DURHAM, NC - FEBRUARY 09: (L-R) Head coach Mike Krzyzewski of the Duke Blue Devils and head coach Roy Williams of the North Carolina Tar Heels shake hands prior to their gaem at Cameron Indoor Stadium on February 9, 2017 in Durham, North Carolina. (Photo by Streeter Lecka/Getty Images)
DURHAM, NC - FEBRUARY 09: (L-R) Head coach Mike Krzyzewski of the Duke Blue Devils and head coach Roy Williams of the North Carolina Tar Heels shake hands prior to their gaem at Cameron Indoor Stadium on February 9, 2017 in Durham, North Carolina. (Photo by Streeter Lecka/Getty Images) /
facebooktwitterreddit

“It’s not what you know, it’s who you know.”

Recent allegations of corruption aside, this idea is extremely relevant in college basketball. This article is not explicitly about the corruption of college basketball but it is an attempt to track the ways in with personal relationships and familiarity play a role in college recruiting.

Players jump AAU teams and even circuits quite often. Unfortunately, a detailed list of which players committed when thanks to which influences, conversations, and additional factors does not, will not, and likely should not exist. Also, contact with these coaches and relationships are built even while players jump teams. The slightly more complex version is that some of these players, especially the top ones, are free agents for all intents and purposes. The shoe brands, colleges, and AAU teams collectively and independently bid for these players. There is no list for that either. Still, while we have very little hard data on the process, using social network analysis on the results is at least somewhat illuminating.

Social network analysis requires a few stages regardless of whether you are using UCINET, R, Python, or literally any other tool. The crux of proper network analysis starts with appropriately structuring the data. Network analysis primarily comprises itself of lists of the nodes and matrices.

The beautiful mess of a network above is a visualization of all colleges with their various ties to AAU teams. A minimum of one player is required for a tie and the size of the nodes and text represent their centrality to the network. The circles represent schools and boxes represent AAU teams. Even on first glance, Maryland is a great example of how a school’s centrality may change from a player playing for multiple AAU teams. Ricky Lindo played for several AAU teams before he ultimately arrived in College Park. Some may argue a player such as Ricky Lindo may make Maryland appear more central to the network than in actuality. In fairness, they also yielded what many considered a top 10 recruiting class, welcoming six new freshmen to the roster.

Kansas, Kentucky, and LSU are schools that pass the initial eye test as central players in the network. Schools like Incarnate Word, Central Connecticut, and Sacred Heart may raise some eyebrows that require some deep diving as the analysis progresses. (Spoiler, two of those three coaches are fresh on the scene). One last high-level glance finds Harvard and Western Kentucky as a sensical outsider with some influence largely due to the likes of a player like the Hilltopper’s Charles Bassey and notoriously connected coach Rick Stansbury.

The mid-majors mean quite a bit to college basketball and are responsible for busting many brackets each year. However, the network of all college teams is massive. The intention to reduce as much noise as possible while still fairly representing the network requires dissecting each tier, low-major, mid-major, and high-major, independently.

I found the monetary details for as many college-shoe brand sponsorship deals as possible but the results showed little correlation between a college’s annual money and eigenvector centrality in the network. Not all schools disclose their agreements. For all the reasons mentioned above and the sake of a complete college-shoe brand partnership list, we will only consider the Power 5 schools, Big East, and American, termed ‘Power 7’, in all analysis going forward.

Above is the network of colleges in these seven conferences who share an AAU tie, they both have players who played for the same AAU team. One of the obvious takeaways I never truly considered starting this project was the regional nature of recruiting. West Coast, Midwest, and Northeast schools each are relatively tied to their region. A strong example of this is the Indiana, Iowa, and Marquette strand on the right side. Another, slightly more hidden one is the UCLA node on the left, which is tied to Arizona, Utah, and Oregon.

Analyzing the eigenvector centrality of this network, Memphis’ influence juts out at an appalling level. No close second exists. For any network, that measure of centrality eye-popping. Digging deeper, Penny Hardaway is the new coach at Memphis. Penny Hardaway happened to own an AAU team up until his recent hire. Since we’re only analyzing a single year of data, the network tends to favor newer coaches who must scramble for players. Still, Memphis is a bizarre outlier.

The rest of the top 25 appear to make some logical sense, with Villanova as an honorable mention, ranked No. 26. For those surprised by Duke and Kentucky’s lower ranking, the players who now play Duke and Kentucky never really jumped across AAU teams like many of the other players. Right, wrong or indifferent Marvin Bagley’s father was awarded an AAU team on the EYBL circuit before attending Duke last season. Obviously, Marvin Bagley is not going to play for another AAU team. This is one slightly more extreme instance but allows one to start to understand why some of the top players really stay put more than some of the less elite players.

Before going any further, it is important to also consider which AAU teams were most influential in the network. Looking at AAU teams’ eigenvector centrality, take a wild guess which team’s influence blows the socks off of every other team? Team Penny. E1T1 with a more human figure is still ranked second. It placed players at Texas Tech, LSU, Oregon, Memphis, and East Carolina. This fits pretty nicely considering the influential schools on the list above and explains how a school like East Carolina becomes so central.

Further down, we see long-standing AAU programs who routinely place players, such as Boo Williams and Team Final. The most recent players from each program include Justin Anderson from Boo Williams and Mikal Bridges and Donte DiVincenzo from Team Final. An AAU team can help promote a young player (outside the top 100) to find a school. Also, AAU coaches may have just as much, if not more, determination on where a player commits than the actual college coach.

The shoe brands’ strategy primarily targets elite youth basketball players and entices them with all the brand has to offer, with the hope of funneling them into a college they also sponsor. After the elite player’s brief college experience, the shoe brand hopes to leverage the previous two years of brand affiliation and trust into an endorsement deal for the now NBA player.

The question is whether these grassroots efforts work. Out of 245 college roster spots considered Adidas, Nike, and Under Armour represented 32, 167, and 46 of those slots, respectively. 20 players,12 ranked in the ESPN100, played for multiple circuits. Including these duplicates, there were 265 instances, split 87 Adidas, 129 Nike, and 49 Under Armour across the three circuits. Out of 82 schools, 57 of the available sponsorship agreements belong to Nike. 15 Under Armour and only 10 Adidas partnerships constitutes the remains of the market. On the other end, players stemmed from 121 different AAU programs 48 of which were Adidas, 47 Nike, and 26 teams were on the Under Armour circuit.

There are three reasons this is fascinating.

1. Nike is generating far more talent and commands the college market but with a smaller or equivalent pool of players compared to Adidas.
2. Adidas is clearly not a dominating brand in the Power 7 conferences but is producing more talent than its colleges can accommodate.
3. Under Armour clearly has a shortage of basketball talent even though their reputation in the football landscape has allowed them to represent more colleges than Adidas.

To perform a slightly deeper level of analysis, implementing QAP Regression analysis compares the two matrices to find the correlation between shoe brand and tie in this case. It is the social network version of multiple regression. As it turns out, 71 percent of the time a player started college representing the same brand he did in AAU. A legitimate argument exists to say Nike owns so much of the market that the results are skewed. This is true to an extent. However, the random probability of any of these players attending any school is still less than a Nike player attending a Nike school. Undeniably, Nike forces the baseline up but they retain the players in their system with more success than the other brands relative to opportunity.

The top 100 players change very little from the time they are 15 until their draft. These are the real power players everyone is pushing for so let’s take a dive. Memphis and Team Penny are nowhere to be seen in the top 10 most influential if we limit our dataset to just those players. long-standing AAU programs such as E1T1, Team Final, and California Supreme continue to produce talent and become influential players even in a one-year network. More importantly, eight of the top 10 teams are Nike sponsored and the other two are on Under Armour’s Circuit. In fairness, the 11th and 12th-ranked teams are Adidas sponsored and that could be a product of the current heat Adidas is under.

According to ESPN’s 2018 top-100, Duke, Oregon, and Kentucky were the only schools to land five top-100 players. That being said, most of top-10 most central athletic programs above are storied, though they are not all ranked with the top recruiting classes or the current top 25. It might make sense to put old money and new money schools in two separate classes of influence in this network. This is not implying any of these schools are cheating or anything even close to it, just a clear metaphor. Old money are your SEC and football schools, like LSU and legendary basketball schools, like Duke. New money are the power programs, like Oregon, and schools building on recent success, like Boston College, that are in the Power 5.

Performing Girvan-Newman cluster analysis, eight different clusters emerge. Three of the five smaller subgroups show explicitly clear regional ties. The Tech subgroup is an interesting one that all fit into the Bible Belt where Virginia Tech and Georgia Tech share the ACC and Texas Tech and Virginia Tech could be considered in a fairly similar category of success, ranked No. 8 and No. 9 on kenpom.com right now. A regional aspect is present in almost all the smaller subgroups with ties. The largest subgroup sees the elite tier of programs — Duke, Oregon, and Kansas — operate independently of conference or even region. The rest of the cluster shows clear elements of conference influence partnered with regional influence. The ties to Duke are a prime example of this as we see Florida and Kentucky, both top members in the SEC, while North Carolina St. shares a conference and state with Duke.

Nike still dominated Adidas and Under Armour in carrying the brand among the top players. Adidas placed five of 11 of their top-100 players at Adidas-sponsored colleges. However, three of those were to Kansas. Under Armour placed only four of 16 athletes at similarly represented colleges. Meanwhile, a mammoth 44 of 53 players repped the swoosh throughout the circuit and into college. Clearly, Nike is maintaining a much higher success rate than its competitors.

For these top 100 players, the logo of their AAU team determines which logo they will rep in college about 63 percent of the time. This is lower than the 71 percent mainly because the demand for these players is so much higher. Likewise, the brands are fiercely competing for these players as they provide the opportunity to represent the brand at the highest level. Also, the figure is significant but Nike’s success rate is so great that it skews the other figures.

So after all this, do these AAU to college pipelines really work for these elite prospects? Not really… unless you are Nike.

*Data from Kenpom.com and AAUStats.com partnered with a whole lot of fuzzy matching laid the crucial groundwork for this analysis. If you ever consider playing with AAU data causally, I kindly advise you to avoid it at all costs. At best, the name and statistical (in)accuracy leaves you baffled and in awe all at the same time. With red warning tape now clear as day on the wall, the data is not perfect but I personally combed through it hundreds of times to the point it is pretty darn close.

A few final obligatory notes: the data contains every AAU team a player ever played for not just their last team or any other specific filter. To clarify, it only includes 17U with some 16U teams that played on the EYBL (Nike), Adidas Uprising (Adidas), or Under Armour Association (Under Armour) circuit and it only takes into account new freshman to college basketball rosters for the 2018-19 season who have played AAU basketball. Transfers are important but not the first landing spot for a player and represent a slightly different recruiting process and set of circumstances.