While Kansas marched towards a title last season, Bill Self subtly adjusted his rotation. The Jayhawks head coach liberally doled out minutes as they rolled past Texas Southern, but just six players played all but 12 of the available minutes in their epic, title-clinching comeback against North Carolina.
This performance is a microcosm of how talented basketball teams manage their rotations throughout the season. My analysis of rotations shows that coaches play larger rotations in games against weaker opponents. In addition, they play looser rotations during non-conference and in the earlier rounds of the NCAA Tournament.
To be sure, coaching styles also play a big role, as does personnel.
Rotations can be hard to quantify. Every coach has a different definition of what the word means in their program. Colloquially though it appears to mean the number of players that are given a significant opportunity on the court. The statistics to quantify this are also limited. Ken Pomeroy, for instance, lists “Bench Minutes” on his team pages (Kansas ranked near the bottom at 301st last season), but that statistic ignores certain players entirely.
I analyzed two slight variations in hopes of finding additional insight into how coaches are managing playing time. One was the number of players that played at least 20% of the minutes available at a position in a game, or eight minutes in a regulation game. The second is the percentage of minutes played by the top five players during each game. Both provide a window into how coaches manage playing time.
It’s worth noting that neither of the metrics include anything about which players started the game. This is for two reasons, one practical and one analytical. The practical reason is that most stat platforms don’t capture the starters. The second is that starters can create false impressions. Just because a player is on the court for the tip-off doesn’t necessarily mean they’ll play a significant role—though it does have some signal. Looking at total minutes played and the players who played the most minutes avoid the complications caused by both of these issues. [A big thank you to Bart Torvik, whose invaluable college basketball site provided all of the data used for these analyses.]
Rotation Sizes
How big is a rotation? The most common number of players playing at least 20% of minutes is eight, which occurred in 37% of the games. The next most common are (not surprisingly) seven and nine, each with just under a quarter of all games.
The top five players get a median of 74% of a team’s minutes during the game, but that can vary widely, from as low as 41% to as high as 100%.
Rotation sizes have stayed incredibly consistent over the past six seasons. Coaches have played an average of 8.1 players 20% of team minutes each of the past five seasons. (It was 8.2 during the 2017 season.) Coaches also play their top five guys right around 74% on an incredibly consistent basis.
Outliers
An analysis of six seasons worth of player minutes data found a number of interesting outliers. There have been five games in the past six seasons where a team played 14 players for at least eight minutes. They are:
All of these games had large final margins. The 2018-19 New Mexico St. team rolled through the WAC, finishing with a 15-1 conference record. Northwestern St. played 14 players in a loss because it was the second night of a back-to-back against Gonzaga. Wichita St. extended its bench in 2017 because the comfortable victory occurred during the first night of Arch Madness, where preserving key players is of the utmost importance.
On the other hand, 134 times over the past six seasons coaches have played only five players at least 20% of available minutes. Not surprisingly, the two schools that have done it the most are St. Bonaventure (nine times) and Syracuse (eight times). Six of those nine instances for the Bonnies occurred during the 2022 season, as Mark Schmidt rode an experienced starting five to the semifinals of the NIT.
The biggest rotations over the course of a season include the 2019 New Mexico St. and 2021 Northwestern St. teams that appear above. It also includes 2020 Saint Peter’s, 2022 Cornell, and 2021 Winthrop. The smallest rotations over the course of a season include 2022 Saint Bonaventure, Western Kentucky, Oakland, and Tarleton State, along with 2018 Syracuse.
Rotations by Game Prediction
Coaches are often more willing to play larger rotations when their teams are projected to win comfortably. The graph below shows the relationship between T-Rank’s projection for a team and the average size of the rotation.
This is also influenced by the results. Teams average between 7.9 and 8 players in the rotation in all games with an ultimate margin between one and 11 points. But the rotation sizes start to steadily climb as the margin balloons.
The expected outcome of the game does seem to have some influence on the rotation size even after accounting for margin. Rotations are tighter during games that are potential toss-ups where both teams have about a 50% chance of winning when compared to games where the outcome is less in doubt.
Rotations by Part of Season
The relationship between score expectations and rotations carries over into the different phases of the college basketball season. Coaches play larger rotations during the non-conference then whittle them down in conference play. Rotations shrink even further come conference and NCAA Tournament time when the stakes are the highest.
The most extreme example of the shrinking rotation from non-conference to conference play was the 2020 Georgetown Hoyas. Patrick Ewing played an almost nine man rotation during a pretty easy non-conference schedule where the Hoyas went 10-3, but once they entered Big East play he tightened the rotation down to just 6.5 players. It didn’t help. Georgetown went just 5-13 in conference play that season.
Rotations continue to tighten as teams advance further in the NCAA Tournament with the exception of the First Four, which makes sense given that the opening round should pit teams of (seemingly) equal quality.
Because of this phenomenon, teams that advance further end up playing smaller rotations overall.
Tempo and Rotations
There is a slight relationship between the tempo a coach has his team play at and their rotation size. Every 10 possessions adds about one-third of a player to the rotation.
As you can see in the scatter plot above though there are definitely teams that defy the trend. One particularly interesting team is the 2021 Coppin State. The Eagles were the fastest team in the country at 76.8 possessions per game adjusted tempo, but their average rotation size was just 7.3 players.
On the other side is the 2017 Virginia team. Tony Bennett always has one of the slowest teams in the country and this often lets him play a shortened rotation, but he changed it up during the 2016-17 season. The only player who played significant minutes was senior point guard London Perrantes. Otherwise Bennett played nearly nine players eight minutes per game. Those Cavaliers were 46th in KenPom bench minutes percentage when Bennett’s teams consistently rank near the bottom of the country in most seasons. Since that season, which ended in a second round NCAA Tournament loss, Bennett has tightened his rotations up significantly.
Coaches and Rotations
Team quality and coaching changes are the reasons rotations sizes change the most dramatically. For instance, Arkansas underwent a dramatic transformation in rotation size when the Razorbacks changed head coaches from Mike Anderson to Eric Musselman. During the 2017-19 seasons Anderson played 8.9 players per game. Musselman took over after that and has shrunk rotations dramatically, despite continuing to play an up-tempo style of basketball. His rotations at Arkansas have included an average of just 7.3 players per game.
It turns out though that this is completely in character for both coaches. Musselman played the shortest average rotations of any coach in the past six seasons while he was at Nevada from 2017-19. He’s ranked seventh while at Arkansas. Anderson, on the other hand, played the 16th largest rotations during that time while at Arkansas. Since moving to St. John’s he’s lowered his average rotation a bit, but to 8.6.
The team that Musselman left, Nevada, has also increased its rotation size under new head coach Steve Alford. Alford plays about 8.2 players per game. Alford played a slightly smaller rotation, 7.6 players, during his previous stint at UCLA.
Here are the coaches who have played the largest rotations during the past six seasons (minimum three seasons at a school):
Some other interesting names near the top include: King Rice at Monmouth (6th), Bob Huggins at West Virginia (12th) and Shaheen Holloway at Saint Peter’s (14th).
And here are the coaches who have played the shortest rotations during the past six seasons (minimum three seasons at a school):
Some other interesting names near the top for the shortest rotations include: Jay Wright at Villanova (6th), Musselman at Arkansas (7th), John Gallagher at Hartford (8th), Jamion Christian at George Washington (9th), Joe Mihalich at Hofstra (10th), Josh Pastner at Georgia Tech (13th), Rick Stansbury at Western Kentucky (14th) and Mike Brey at Notre Dame (15th).
Below is a Google sheet with the 488 coach-school combinations with at least two seasons at a school (I expanded the universe to allow you to have more to dig into).
No comments:
Post a Comment