Freelance Friday: Recruiting Risk In a One-and-Done World
By Guest Post
Freelance Friday is a regular series at Nylon Calculus featuring guest posts from around the web. If you have an idea or submission for a future Freelance Friday post, get in touch on twitter, Facebook or email TheNylonCalculus at gmail dot com. This week’s post comes to us from Will Schreefer (with an assist to his buddy Mike Young) and it concerns the modern NCAA recruiting environment, balancing the elite short-term contributions of a top-top recruit with the longer career of a slightly more modest three or four-star prospect. Will also brought us dispatches from this year’s NESSIS conference earlier in the week. Follow him on twitter @ReferSadness
In the 2005 collective bargaining agreement, the NBA’s players and owners agreed to restrict access to the annual draft by both age and time since a player’s class finished high school. This effective ban on “preps-to-pros” players, continued in the 2011 CBA, has mainly done the intended job for both sides. Billion-dollar ownership groups have offloaded risk by forcing potential NBA prospects to ‘intern’ at their own risk, on their own dime & timeline. Meanwhile, a few veteran players each year squeeze out one more big payday with their potential replacements waiting in the wings that one extra year. Depending on one’s perspective[1. To whit, that of a fan who cares not for the financial, labor, and moral aspects of the issue.] the most important thing is that college basketball got more talent! One year of the best and most talented prospects in the NCAA system is better than none from that standpoint.
Unsurprisingly the entirely predictable “one-and-done” phenomenon which ensued fundamentally changed the recruiting landscape for top-tier college basketball coaches. The expanded pool of elite recruits attending school increased the possibility of ‘stacking’ top twenty or even top ten-rated prospects on a single team and creating single-season dynasties built on the talents of these shooting stars. John Calipari[2. Ed: Hall of Famer John Calipari, to you.], one of the earliest and most notorious beneficiaries of such a strategy, has fashioned a yearly contender at Kentucky by consistently pulling in top-top-tier talent, watching many of them leave for the NBA after a successful season. He then reloads for the next year with a new wave of elite recruits. Rinse. Repeat. The success of this approach has forced some of the other mainstays of the NCAA to recalibrate their recruiting approaches. Notably (beyond the success of coaches like Calipari and Arizona’s Sean Miller), Mike Krzyzewski’s Duke captured the 2014-2015 NCAA Championship behind a freshman class that included four top 25 (by RSCI) recruits – three of whom were drafted in the first round of the 2015 NBA draft.
This ‘new’[3. College teams have been using one-and-doners to succeed for years – there are just more of them to go around today.] trend is a relatively obvious means of maximizing the talent on a team’s roster in a given year – recruit the best guys possible and then play them. And it’s clearly been successful, both recently and in the past. But there are other ways to succeed as a program at the college level. Look no further than Duke’s opponent in last year’s title game. The Wisconsin Badgers, while hardly slouches in the yearly recruiting game, relied almost exclusively on a roster comprised of upperclassmen, including Player of the Year (and former 3-star recruit) Frank Kaminsky, to reach the title game.
Recruiting is ultimately a crapshoot – even if ranking was considered a 100% accurate proxy for talent[4. All too often, it’s not.], factors like fit, team construction, and team mentality can ultimately figure into the success of a given recruit far more than his ranking coming out of high school. But as we approach the 2015-2016 season, with shiny new recruits and reconfigured teams, are there any clues as to the right[5. Ed. Strictly in terms of winning college games. Bringing ‘right’ into a discussion of the NCAA is a whole can of worms that Will didn’t himself open.] approach? There is no single prescription, as both most methods have plusses and minuses.
To better understand the risks and rewards, I took a look at the historical returns on recruits, as ordered by rank on 247 sports, over the last twelve NCAA seasons, to try and shed some light on the subject.
Most of the analysis was performed by aggregating win shares, win shares/40, or box plus-minus (while accounting for percentage of team minutes played). While all one-number metrics tend to have their problems, it’s difficult to envision a better way of comparing players across school, year, and position in a ‘league’ with 350+ teams, rather than 30. RealGM’s databases were used to supply the calculated win shares as well as necessary rate statistics for the calculation of raw and team/adjusted BPM. KenPom was used for adjusted team efficiencies. 247 sports was utilized as the basis for the recruit ‘rankings,’ as the historical records of their aggregate rankings stretched back further (in time and player number) than any other publicly available source.
Only players who played games at the college level were included in the analysis. A number of recruits on each yearly top 250, for a variety of reasons, never played basketball at the NCAA level. The reasons for this were varied and seemed impossible to predict or model, so these players were simply ignored.
On to the data.
On a ‘per-year’ basis, top 20 recruits significantly outperform everyone else. Some of this is certainly talent level. Highly ranked players coming out of high school are, as a very general rule, quite good.[7. Ed. For a fun exercise, look at how well the McDonald’s All-American Selections predict eventually making the NBA going back as far as you want to go. The recruiting game might be shady, but folks who making their living spotting the best talent get pretty good at it.]. It doesn’t hurt that recruits ranked in the top 10 to 20 almost always get sufficient minutes to contribute. Among top 10 recruits who played in their first year with the program they signed with out of high school[2. Thus removing redshirts, and season-long injuries.] only 5.1% played less than 500 minutes[2. Around 15 minutes per game.] in that freshman season. Even in the 11-20 bracket, just under a quarter of the recruits failed to play at least 500 minutes as first year players. Every other recruiting “bucket” graphed above saw at least 40% of players play less than 15 minutes per game. This probably illustrates some of the pressure coaches feel to give super-elite recruits more playing time[4. Which necessarily leads to bigger gaps in counting stats and, typically, one-number metric scores.]. It’s possible some lower-ranked recruits could have accrued stats could be closer to those of top recruits, but for the most part, these less talented freshman did not deserve roles quite so large.
The drop off seen by Top 20 seniors in their fourth year makes sense viewed through the lens of talent attrition. The most talented players in those groups will typically take off for the NBA, leaving the less successful (but still quite good!) leftovers to drag the average down some. For most other groups, a natural progression of improvement on a year-by-year basis (for the players who continued to play) can easily be seen – though the ‘degree’ of improvement is likely muted, in much the same fashion as the top 20 with the best players from each tier leaving the college ranks for better opportunities.
On a per season basis, it should not be a surprise that the top recruits shine brightest. But what about over the course of a college career?
Top 10 recruits are often outperformed by their lower ranked counterparts over the course of their college careers[1. For this analysis, only lower-tier recruits (36+) that began playing before the 2013-2014 season were included as the careers of many of the three and four star recruits that began playing in and after that year have not concluded yet, with their best years likely still to come, possibly skewing the results.]. Judging by the medians of each data set, top 10 recruits contribute about as many win shares during their college careers as those between 76 and 100. Of course, the average performance of top 10 recruits is significantly higher! More interestingly, recruits ranked 21 through 75 have done ‘more,’ NCAA career-wise, than those in the top 10 – and about as much as those in the 11-20 range.
The reason for this isn’t difficult to parse – those lower-tier top 100 recruits are around quite a bit longer than the guys in the top 20. Over half of the top 10 recruits from the last 12 years only played 1 year at the NCAA level, and 75% didn’t make it past the second year. Though the numbers aren’t quite as drastic, less than half of the players 11-20 made it to a third season of NCAA play. The average career length for each of the bins is listed above – the general average after the top 20 is about a 3 year career. The chart below indicates the historical prevalence of players leaving school as underclassmen after a given season based on their initial recruiting rank and minutes played in that season.
Obviously, the longer a player is around, the more counting stats and win shares they can amass. But it raises an argument of value for coaches and recruiting strategies in general – should more effort be expended chasing the more volatile (but potentially transcendent) 1-2 year contributions of a 5-star, top-20 caliber recruit, or are there benefits to securing the more dependable – and potentially greater in the aggregate – long-term career of a top 100 recruit outside the top 20? In the new normal of college basketball, that choice faces every coach not named Calipari nearly every season.