A New Way to Evaluate the Next Great Men’s Basketball Coach - AETD (Actual - Expected Transfer Differential)
In today’s rapidly evolving world of collegiate athletics, the transfer portal has emerged as a key driver of team success, providing programs with opportunities to gain an edge through player acquisitions, essentially free agency. This structure of collegiate athletics has not only has captured widespread public attention—just look at this viral Twitter graphic displaying the Sweet Sixteen starting lineups and their original schools, which amassed 9.5 million views—but has also sparked a growing industry around college athletics and recruiting analytics, like these wonderful dashboards made by Evan Miyakawa. Clearly, there’s a growing interest in understanding how the transfer portal shapes outcomes on the court.
But while countless metrics exist to evaluate individual player performance, far fewer attempt to quantify the impact of coaching staffs. That gap sparked a central question that drives this article: How can the transfer portal be used to evaluate coaching performance in men’s college basketball?
To explore this question, I developed a metric that evaluates coaching impact by analyzing transfer trends over the past three seasons (2021–2023). Specifically, the metric compares the actual number of transfers from a program to the expected number of transfers, producing a value I call AETD (Actual Minus Expected Transfer Differential). I figured that good coaches at the mid-major level will keep their players around more than an average coach AND will win a majority of their games, as players want to play for them at a winning program. In the sections that follow, I’ll explain how AETD works and highlight a few coaches who stand out according to this metric.
What is AETD?
AETD is short for the actual minus expected transfer differential. This metric contains two components, which I will explore now.
Actual Transfers
This component is pretty straightforward. It is the number of transfers in each year under a head coach at a college. For coaches that have more than one season at a program, we take this as an average.
Expected Transfers
This component is more complex and is based on a model I created that assigns a probability that a player transfers from a team that season (think advanced logistic regression). The model is proprietary for now, but it is based on a number of factors such as usage rate, minutes per game, distance from home, conference, etc. It uses publicly available data from cbbd. If you haven’t checked this database out and you are looking to work with free and clean data, I highly recommend doing so! Next, I get the expected number of transfers for a coach at a program by summing the probabilities for each of the listed players (there are some exceptions listed below). Again, for coaches that have more than one season at a program, I took this as an average.
Exceptions
This does not include players that declared for the draft and players that did not play. I also did not include seasons for a coach where they left the program. Also, I did not include coaches from this year’s transfer window as it hasn’t finished at the time of writing.
A Brief Example
Since I am trying to uncover and evaluate coaches that have potential to make the next step, I will use one of the most talked about rising coaches in Will Wade and his McNeese team during the ‘23-24 season. Wade led his team to a 30-4 record and an NCAA tournament appearance and despite this increased notoriety, he was able to keep players like Javohn Garcia, Christian Shumate, and DJ Richards Jr. around. Each of these returners played a major role in McNeese's success, helping the team not only return to the tournament but also notch a signature win over Syracuse.
Despite their success in the ‘23-24 season, Wade managed to keep around several members of the core to make another magical run this season. Despite this, several unexpected members of the team transferred, including Nasir Mann and Mike Saunders Jr. As a result, Wade’s AETD for the season landed just below neutral at -0.08, indicating that slightly fewer players transferred than expected. While not a dramatic overperformance, it still reflects a level of player retention consistent with a stable and successful program.
How Should You Use AETD with Existing Measures of Evaluating Coaches?
Like any metric used to evaluate coaches or programs, context is essential. In this shorter article, I’m using win percentage as the primary contextual layer to complement the AETD metric. However, there’s much more depth to explore, particularly around the reasons players choose to transfer (low minutes played, high production at a mid/low-major level etc). While some of this nuance is addressed by narrowing the focus to coaches at the mid-major and low-major levels, diving deeper into the “why” behind each transfer would offer some direction for future analysis.
Though for me, win % is the necessary context that should be given to AETD. If you have a coach at the low/mid-major level who is frequently winning games and keeping players around who typically would’ve transferred for a higher competition level, that gives us two key observation points about a coach:
They can win. For this, I’d say finding a coach with a win % at or above 60-65% is preferred. Some of this depends on the sample size, if a coach is at a program for just 1 or 2 seasons, you probably want that number to be greater whereas if they’ve been their long-term, you’re probably alright with that win % dipping slightly.
Players want to play for them. Anecdotally (I know I’m supposed to be a statistics guy), this skill is more translatable across programs. We have seen time and time again that winning at one program does not translate to winning at another. Now I don’t have the current sample size to test AETD as a predictive value, but I think this speaks to the more personal side of coaching as opposed to the on-court side. For this, I will choose coaches with a AETD less than 0, meaning that less players are transferring than expected.
This plot provides us with an overall picture of the college basketball coaching scene over the last few years. The size of each point represents the number of games by a coach at that program with the color representing the competition level of that program. Ideally you want to find a coach that is as low and as far to the right as possible. We see long time head coaches Bill Self and John Calipari having very high career win %’s yet they don’t outperform their AETD like Tom Izzo, signaling that Izzo is able to keep his players around a bit longer. We see up and coming coaches like Dusty May (former FAU) and Darian DeVries (former Drake) in that bottom left quadrant, displaying their ability to not only win but keep quality players around.
Who are the Next Great Men’s Basketball Coaches?
Note: Before we get into the analysis, I made the decision to remove well-established head coaches (had over 5 seasons at a program) such as Kelvin Sampson, Mark Few, Randy Bennett, John Becker, and Ben Jacobson.
These are the list of 19 coaches who make the cut. As you can see over half (11) of these coaches have changed programs to a higher conference either following the 2023-24 season (red) or following the end of this season (yellow).
Adjusting for the coaches who recently left, we are left with just 8 coaches who fit all of the criteria: Speedy Claxton (Hofstra), Jeff Boals (Ohio), Darrin Horn (Western Kentucky), Chris Victor (Seattle), Tom Pecora (Quinnipiac), Takayo Siddle (UNC Wilmington), Casey Alexander (Belmont), and Andy Kennedy (UAB).
These coaches have a range of different experience in college and therefore the list provides a good range of options, depending on the wants of a program. If you are looking for a coach that could be the next big thing (think higher mid-major) but don’t have a lot of experience, look at guys like Speedy Claxton or Chris Victor. The coaches slightly above this tier who have some experience at a higher level would be Takayo Siddle or Jeff Boals (think fridge Power-5). And then the coaches that are fairly well-established within the collegiate coaching scene, but are perhaps finding their stride with their current program now would be Casey Alexander, Darrin Horn, Tom Pecora, and Andy Kennedy. These coaches would be a lower-risk option to jump from the Mid-Major to Power 5 level.
Honorable Mentions:
Drew Valentine (Loyola Chicago), Bob Richey (Furman), John Groce (Akron), and Joe Pasternack (UC Santa Barbara). For the most part, these coaches had less than 275 games at their current program and also had an AETD less than 0.
Use Cases
I think there are two distinct use cases for this model:
Athletic directors in a coaching search. These individuals could use this criteria as a reference to find and evaluate coaches that broadly speaking, consistently create a winning and healthy environment that players want to play in. These qualities are two key aspects of coaching and we see the best coaches accomplish both of these things.
With the new world of NIL, college agents could set themselves apart by using this approach as a broad check to evaluate the stability of a program before sending their clients (athletes) there. The best use case of this model may be to filter out the programs that consistently lose games and have a poor culture, where players are often leaving when they wouldn’t be expected to; however, I’m not here to bash anyone publicly :).
Conclusion and Limitations
Thank you all for reading, it was really fun to get some work out in the analytics space! In this new world where college sports are becoming professional and the transfer portal is turning into free agency for college athletes, I believe there are countless ways analytics can help us better understand coaching value. This project represents just one small step in that direction.
There are several paths I could explore next: analyzing how transfers from specific programs perform elsewhere, or measuring the net impact of a coach by comparing a program’s performance before and after their arrival. This methodology also has potential across other sports, particularly in areas where the head coach plays a central role, including the women’s game.
While I believe AETD is heading in the right direction to evaluate off-court coaching, the limitations are down to the broad assumption it makes that a player will choose to transfer because of the head coach or the team environment/culture. It ignores possible edge cases of a single incident impacting a transfer such as an issue within the family of a player and these incidents with low variance can impact a coach’s AETD if a coach is has only been at a program for a short period of time. This is why these values require careful interpretation and context must be provided to correctly understand the AETD for each coach.
That’s all for this article, I’d love to hear your thoughts and ideas! Feel free to reach out on LinkedIn, Twitter, or the new Bluesky that I created. I’m always happy to connect and talk more! Thank you and I’ll see you next time!
