Duke. Kentucky. North Carolina. Michigan State. Kansas. Villanova. Virginia. What do all these teams have in common other than having been exceptional for the past few seasons? The answer is that they all have great coaches that are able to guide their teams to success year after year despite the roster turnover they face. Prominent one and done schools such as Duke and Kentucky are able to contend every year even though their best players leave each year. Coaches like Jay Wright and Tony Bennett for Villanova and Virginia, respectively, have led their teams into an unprecedented period of glory. Behind every dynasty and blue blood in college basketball is a successful head coach who is able to develop their players and implement systems that improve the team. But who is the best coach of college basketball?
While ranking the coaches, I needed to find a way to measure how well a coach did during a season. Some believe that good coaches manage their talent well, while others believe that good college basketball coaches are able to keep their teams in contention every year despite their roster. I think that the latter is better for ranking coaches as the best coaches usually have the best teams. Therefore, the coaches who are constantly able to avoid regression and stay at the top are the best coaches.
First, I started by creating a linear regression predicting the SRS for a given season (Simple Rating System, which is calculated by adding margin of victory and the average opponent margin of victory) using the previous year’s SRS as an input. The output (predicted SRS) was subtracted from the coach’s team’s SRS, which equals the residual of the output. The residual represents the actual value minus the expected value, meaning that a positive residual indicates a good season while a negative residual indicates a bad season (based on expectations). Therefore, a positive residual means that the team did not regress as much as expected and a negative residual means the team regressed more than expected.
Then, I found the average, standard deviation, and the count of the residuals. Using those statistics, I was able to conduct a statistical test (a one sample, one sided t-test) comparing the mean of the residuals to 0, which is the average value. This method means that a high average residual, a low standard deviation, and a larger count of seasons gives the lowest p-value, which is the chance that such a result happens by random chance alone assuming that the coach is only average. By comparing the p-values of the residuals of the coaches, the best college basketball coaches of all time can be ranked. A lower p-value means that a coach is better whereas a higher p-value means that a coach is worse. (Note: This method is best for ranking coaches’ careers as a whole, instead of ranking coaches for just one season)
Since I couldn’t find the stats and produce the results for all of the coaches in college basketball history, I selected thirty coaches to rank. The coaches were chosen by taking the top thirty coaches with the most career wins and coached in a time where SRS was recorded. After gathering the stats and finding the p-values of the residuals for each coach, I constructed a table ranking the top coaches.
Using this method of ranking, Rick Pitino had the lowest p-value with a value of 0.000039. Meanwhile, coaches Mike Krzyzewski, Bill Self, Jim Boeheim, and Lon Kruger lie right behind Pitino as the best coaches of all time. One interesting part of the ranking was that the coach with the best residual average did not necessarily always have the lowest p-value. For example, Jim Boeheim had a low p-value because he coached for so many years. Additionally, Lon Kruger did not have an abnormally large average or number of seasons, but had a low standard deviation of the residuals which lowered his p-value. However, one thing to notice is that all of the coaches had a p-value below 0.10 except for Cliff Ellis, meaning that they were all very good coaches.
Further Coaching Comparisons
Rick Pitino Career Outlook
Since Rick Pitino came out as the best coach of all time, I decided to take a deeper look into his career as a coach.
Out of Pitino’s 31 years of a head coach, only 6 resulted in a negative residual, meaning that he had only six seasons where he disappointed based on his expectations. In contrast, Pitino had 18 of his 31 seasons with a SRS over expected of 5. Therefore, Pitino always was able to improve his teams and avoid regression towards the mean. Meanwhile, only 2 of Pitino’s 31 seasons as a coach had an SRS over expected below -5, which is a very bad season.
Many great coaches over the years of college basketball have had a rival coach that marks an important game in the college basketball calendar. The two rivalries that I will be exploring are Mike Krzyzewski versus Roy Williams and John Calipari versus Rick Pitino. The Roy Williams versus Mike Krzyzewski rivalry is embedded into the deeper Duke versus North Carolina rivalry. Coach K of Duke has coached the Blue Devils for 40 years, whereas Roy Williams has led North Carolina since 2004. In their years at their respective schools, both coaches have been evenly matched.
Roy Williams has outperformed Mike Krzyzewski in 9 of the 17 years that they have been at their current schools. However, Coach K has a better average residual than Roy Williams does (3.21 vs 2.71). Coach K and Roy Williams have been evenly matched for the years at Duke and UNC, shown by their win-loss record against each other (Coach K with an edge as Duke is 22-16 versus UNC since 2003) and the amount of championships won during this time frame (Roy Williams has the edge with 3 championships to Coach K’s 2).
The second coaching prominent coaching rivalry between two of the greatest coaches in the history of the sport is John Calipari versus Rick Pitino in their time at Kentucky and Louisville, respectively. They were rivals from 2009, when John Calipari was hired at Kentucky, to 2017, when Rick Pitino was fired due to scandals at Louisville.
John Calipari has outperformed Rick Pitino in 5 of the 8 seasons that they were rivals at their schools. Additionally, John Calipari had an average residual of 4.15, greater than Rick Pitino’s average residual of 2.94. It is interesting that even though Pitino ranked as the best coach using the p-values, he was outdone by John Calipari while both coaches were a part of the Louisville-Kentucky rivalry. Calipari has also had an 8-3 record against Pitino in games played against each other in the Kentucky-Louisville rivalry, but both won one championship in the years of 2009-2017. John Calipari had was significantly better than Rick Pitino in many aspects while they were at Kentucky and Louisville, respectively. It is safe to say that John Calipari won this rivalry, but the Coach K versus Roy Williams rivalry is much closer.
NCAA Tournament Results
While looking at the SRS residuals for each coach is helpful in determining who was best, one major part of college basketball was left out. Using that method, the results of NCAA tournament appearances (or lack of them) are very telling of how good a coach is. Since the p-value method is able to narrow the amount of the best coaches. I wanted to really determine which coach was the bets, so I looked at the post season results of the four coaches with the lowest p-values. These four coaches were Rick Pitino, Mike Krzyzewski, Bill Self, and Jim Boeheim.
As seen by the graph above, Mike Krzyzewski had the highest percentage of seasons advancing at least as far as the final four (just over 30%). Additionally, he has the most championship wins among the four coaches with 5 titles, compared to one for Boeheim, one for Self, and two for Pitino. Since championships are the ultimate goal of playing the sport, looking at these results is arguably more important than looking at how well a coach outperformed expectations.
After looking at the coaches with the lowest p-values of their residuals and their postseason performances, I think that the two best coaches can be narrowed down to Rick Pitino and Mike Krzyzewski. Even though the selection of Pitino as a top two coach may be controversial, he constantly did better than his expectations based off the previous year and was able to avoid the pull to the average almost every year. Since Pitino and Coach K were the two best coaches using p-values, looking at their NCAA tournament results and championship wins is a way to better choose the best coach. Since Coach K had 5 championships compared to Pitino’s 2, and he also had a higher percentage of seasons advancing to the Sweet 16, I believe that he is the greatest college basketball coach of all time. Since he has avoiding regression to the mean year after year and also won many championships, Mike Krzyzewski is the best college basketball coach of all time using statistics.