Note: Making its debut on the site today is PASR (Predictive Analytics for Successful Recruiting), a model created by Jackson Fambrough using economic principles, that predicts where college basketball recruits will go to school. The predictions cover the top 150 recruits (as determined by Rivals) for future classes and will be updated approximately once a month. (The forecasts are linked from the player section of the team page for teams that are involve with at least one recruit. Or bookmark this link for future reference.) Jackson provides a description of the model below. You can contact him with your questions and comments at [email protected] or on twitter @JackFambrough

The theory for the model is based on an economic term called utility maximization. Utility, in an economic sense, means the satisfaction or happiness received from consuming a good or service. The model’s main purpose is to show which school will provide a recruit with the highest possible expected utility. Expected utility is divided into two different categories, short-term utility (the 1-4 years the recruit is in school) and long-term utility (the years after they leave the university).

There are several factors that play a part in generating short-term utility, such as winning percentage. A recruit is more likely to go to a school with a higher winning percentage over the past five years because they want to be with a more successful program than a less successful one.

In addition, any recruit coming into college wants to play right away and if there isn’t any immediate playing time available when they get to a school, they might not want to go that school. The more playing time available at a school, the more likely a recruit will want to commit to that school.

A school can also be more attractive to recruits in amenities they offer, like weight rooms and state of the art stadiums because recruits want to play in top-of-the-line facilities. One of the biggest dreams for any young basketball talent growing up is being able to play on television and becoming a star. As such, the more media coverage a team has, the more likely it is that their players will be shown on television, thus increasing the likelihood that recruits will choose that school.

The last short-term variable deals with distance between the recruit and the school. Recruits want to play close to home because the closer they play to home, the less money their family and friends are going to have to pay in order to travel to the recruit’s games. This means schools are more likely to pick up commitments from recruits closer to the school rather than recruits from different parts of the country.

In analyzing the long-term factors, we expect that both graduation probability as well as the academic ranking matter most to a recruit’s parents. Parents typically want their child to get an education, resulting in an eventual degree from a university. The higher the graduation probability of a school, as well as a higher academic ranking, the more attractive to a recruit’s parents, meaning a recruit is more likely to commit to that school. The ultimate dream for any talented kid playing basketball is to play in the NBA; thus, a recruit is more likely to commit to a school that has a reputation of sending players to the NBA.

The school having the best combination of the short-term and long-term factors detailed above will be the school generating the most expected utility or satisfaction for the recruit, which should result in the recruit choosing that school.

In order to model the theory, a probit model is used. A probit model predicts the probability of something happening based upon certain variables. This model takes variables dealing with three categories (recruit characteristics, school characteristics, and the relationship between the school and recruit) and uses them to predict where a recruit will go to school.

Recruit characteristics involve recruit rankings and what position they play. School characteristics include items such as a school’s success athletically and academically as well as age/capacity of their stadiums. The relationship between the recruit and the school involves the geographical location of the school, if the recruit took an official visit to the school, and the relationship between the recruit and the school’s academics.

Essentially, PASR is capable of predicting the chances any school has with any recruit. All it needs is the data for the characteristics mentioned above. For every recruit the probability listed is the probability the recruit will commit to that school, factoring in all schools that have offered the recruit a scholarship.

Now the question is, how accurate is PASR?

Below is a chart breaking down the forecasts made by PASR for the 2006-2015 recruiting classes into probability ranges. For example, when PASR forecasts a school with an 80-90% chance to land a recruit, the recruit commits to that school 80.71% of the time. As you can see the observed success rates are close to being in line with the probability ranges.

Something that might be noticed is some recruits chose schools with a less than 5% chance. While this may seem odd or may even lead to questions concerning the model’s validity, several things can help explain it:
1.) The model assumes every recruit acts rationally in their decision making, but sometimes that isn’t always the case.
2.) The model doesn’t account for everything because of lack of information (i.e. playing styles of recruits and high schools fitting into a prospective college).
3.) Recruits occasionally have extremely unique circumstances causing them to commit to a certain school (i.e Bill Walker and his high school eligibility).

PASR can be used as a real world application by providing fans with unique insight into the recruiting process. PASR, using numbers and data rather than feelings and leaks, is a tool that provides the public with another method to predict where a recruit might attend school.

A few thoughts on the initial set of forecasts:
– Looking at one of the top players in the 2015 class, LSU was the both the predicted and actual choice for Ben Simmons, even though Florida was the in-state option. The unofficial visit to LSU is what caused the model to swing in favor of LSU. It was the only unofficial visit scheduled by Simmons to any school that had at that point offered him a scholarship and it shows he was already leaning towards LSU. Simmons’ case shows the importance of a school trying to get a recruit to come visit campus as soon as possible.

– PASR shows Kentucky’s influence as a recruiting juggernaut. If a recruit has a Kentucky offer, that is where the model predicts them to go and gives them a sizable advantage in almost every case dating back to the year after Calipari was hired.

– Recruiting in-state is a key to being successful because having a recruit in the same state as the prospective school is a huge boost for obvious reasons. The most recent example is PJ Dozier who committed to South Carolina over Louisville. While his commitment to a school who hasn’t had a winning season since 2009 over a school who won a national championship two years ago might be shocking, the explanation as to why is pretty simple.

PJ Dozier is from South Carolina, he has family members who have played/are playing a variety of sports there, it would mean his family doesn’t have to travel that far to every single one of his games (one of the largest costs to a basketball player’s family). All of these benefits can be seen in the forecast from PASR, where being an instate recruit is a significant factor, in the end giving South Carolina an almost 10 point advantage.