This is a review of the player tracking data research conducted by Dan Cervone, Alexander D’Amour, Luke Bornn, and Kirk Goldsberry.
The key moment of an offensive play in a basketball game may not be when the points are scored at the end of the possession. It may have occurred earlier in the possession with a pass or a move to elude a defending player. The points could not have been generated without the offensive strategy that put the ball in the shooter’s hands. However, traditional basketball statistics do not take this into account. Stats tend to look at quantifiable data like points, rebounds and turnovers. These evaluate the skill of the shooter but fail to take into account the skills of the players whose actions lead up to the shot.
In Pointwise, a framework is built using player tracking data to develop a quantitative representation of the entire possession as a series of summaries for each moment of the possession in terms of the number of points the offense is expected to score – expected possession value or EPV. The model is able to determine how the ball handlers make decisions based on where the players are positioned on the court. Every possible option a player has is given a point value in order to evaluate how each move added to the possession. It is also possible to look at alternative moves the player could have made to determine if they made the best choice.
This possession model using player tracking data gives coaches the ability to estimate the probability that the player will make a particular choice in a particular situation as well the resulting EPV of the resulting possession. Player’s options include single moves such as passing or shooting as well as longer moves such as moving to the left or right.
EPV is calculated continuously through the possession, increasing and decreasing as the expected value changes based on the actions of the players on the court and whether these actions are more, or less, likely to lead to points being put on the scoreboard.
Analysts can evaluate a player’s value in certain situations by replacing him with another player and observing how the EPV changes. Plays can be analyzed to determine if a player passing the ball creates a higher EPV than taking a shot. This allows analysis of whether the player is making good choices for the team or selfish choices to pad his point total. Coaches can look at the plays their team typically employ during a game to decide if those plays are generating a high enough EPV or if they should be modified, using player tracking data.
How mature is your team’s analytics program? Take the Sports Analytics Maturity Assessment.
Learn about the Groundbreaking Sports Analytics Model coaches and sports analysts are talking about!
Learn all about Sports Analytics here.