This is a review of the NBA research conducted by Philip Maymin, applying Euclidean Distances to measure pace and acceleration.
Basketball coaches illustrate plays using x’s and o’s on a white board along with verbal instructions regarding the pacing of the play. This creates a barrier to creating a method to objectively measure execution and the contribution said execution has to winning a game. In order to combat this issue a dynamic algorithmic approach is created.
Optical tracking tracking for 233 regular and post season games from the 2011-2012 NBA are analysed for half-court situations which begin when the last player crosses the half court line and ends when the offense no longer has possession of the ball. This limitation is set in order to filter out improvised breaks and include only those possessions where the player’s movements are the result of intentional practice. The resulting data set included 30,950 plays, each lasting an average of 7 seconds. The set includes the location of all 10 players as well as the basketball.
Acceleration is calculated as the second difference of the Euclidean distances between sequential moving average positions of a player on the court, divided by the standard of gravity.
Analysis shows that acceleration is rare and events of greater acceleration occur even less often. The frequency acceleration by position is charted with positions organized by average height of players in the position. Examining the chart shows that centers accelerate more than guards do. Bigger players tend to have more episodes of extreme acceleration. Players seem to accelerate based on who they are and not based on whom they are guarding. The spatial distribution of all players and the spatial distribution of accelerating players are quite different. Acceleration typically takes place in one of three areas: the paint, the top of the key, and the combined area of the extended elbows and wings. With each area on the right and left hand side of the court that comes to six locations where extreme acceleration typically occurs.
Proportions of co-accelerating players are illustrated in pie charts, with each chart based on a different acceleration rate. The lowest acceleration rate has the greatest proportion of all five players accelerating together, indicating that mild acceleration is the norm for most players the majority of the time. As acceleration increases, the likelihood of co-acceleration drops dramatically.
Analyzing the acceleration of each offensive and defensive player helps analysts define basketball plays, perhaps resulting in a playbook of plays for each team. Past data regarding acceleration could be used to predict future performances, helping scouts analyse plays performed by future opponents in order for coaches to create and practice the best strategical plays possible. Analysts could look for any possible relationships between acceleration and co-acceleration and field goal percentage, or any other statistic. Those relationships could then be used in training players how to best implement acceleration and co-acceleration into their plays to make them even more effective.
Analytics Used: Euclidean Distances, Dynamic Algorithm
Find out how Sports Analytics Expert Victor Holman can give your team the competitive advantage.
How mature is your team’s analytics program? Take the Sports Analytics Maturity Assessment.
Discover the Groundbreaking Sports Analytics Application and Framework coaches and sports analysts are talking about!
Learn all about sports analytics in Victor Holman’s Sports Analytics Blog.