College Football Recruiting: Analyzing commitment dates

The timing of recruiting commitments is often undervalued. The earlier big name players commit – and assuming they stay committed – the longer they have to help their future college coaches build elite recruiting classes. If you’re a head coach or recruiting coordinator, there is nothing better than starting a class off with an elite recruit or two early on in the recruiting cycle to help sway other blue chippers.

A few weeks ago, I ran some initial tests on the 2014 five-star recruits’ commitment patterns, but over the last week I’ve put together a larger sample of recruiting data from 247’s composite rankings, with the hope of producing more generalizable results.

So, I created a stratified random sample of 90 five- and four-star recruits from the 247 composite’s top 247 players between 2012 and 2014, with each year being a different stratum (i.e. 30 recruits per year). I then used a random number generator to pick the 30 recruits to be sampled in each recruiting class. The stratified random sample helps ensure a more representative, and hopefully generalizable, sample that accounts for any variations between recruiting classes.

After gathering my sample of recruits, I calculated the average commitment date for each class and compared each recruit’s commit date with the class’s average to create a +/- score. For instance, Deshaun Watson became Clemson’s first commitment for 2014 way back in February 2012, committed 525 days before the average commitment day for the 2014 sample recruits.

This time around I was able to test the effects of more independent variables, including recruit position, conference, being in-state, and 247 composite recruiting ranking, to see which affect the timing of a recruit’s commitment. Briefly, I wanted to see if certain positions were more likely to commit earlier than others, whether recruits were more likely to commit to certain conferences before others, whether in-state commits give their verbal pledges earlier or later than others, and whether higher-ranking recruits commit after lower-ranked ones (as this sample includes both four- and five-star recruits).

The results were interesting, if not highly predictive. The model overall had an r squared of just .04 and only the in-state variable was close to statistical significance at the 95% confidence level (with a P-value of .06). This means that, for the sample and holding the other variables constant, recruits committing to an in-state school commit an average of 59 days before the average commitment date. Assuming this is relevant for the whole population of recruits, this theoretically works both ways: in-state recruits commit to their schools earlier than others, and/or schools like to secure commitments from in-state kids first – the "securing the borders" recruiting strategy. Other than that, on average, neither a recruit’s position, conference, nor his ranking affected the timing of his commitment in a statistically significant way.

However, there were some monthly patterns of commitments.

The spike in February represents those recruits announcing on or near signing day, while the relatively high number of January commits might be from the high school all-American games. The peak in the beginning of summer might be commitments at school-specific camps and all-star camps like Nike’s The Opening; some of them might simply follow unofficial visits after the end of the school year as well.