More fun with Stats
We are in an exciting period for basketball analytics. The SportsVU cameras are giving us gobs of data and some great minds over at Harvard are using that data in a new and exciting way. Kirk Goldsberry of the amazing site Grantland has detailed an exciting new model to look and predict a possessions value on a second-by-second basis (Check that out here). That new model takes into a just about every possible action that occurs during a possession and spits out the Expected Possession Value(EPV).
This new stat is pretty exciting for a number of reasons but what excites me most is that this model can actual put a value on decision making in real time. By calculating a players tendencies to pass or shoot at any given position on the court we can, using this model, actually weigh a players decision to pass or shoot accurately. The implications of this stat are huge. EPV gives us our most accurate look at decision making to date. Eventually, coaches could identify the exact moments during an offensive set or action that would produce the most points for that possession with the given personnel. With this data it could be a possible to create new offensive and defensive philosophies that the NBA has't seen yet. This data will make clever coaches even more clever.
Coaches aren't the only group of people that would benefit from this new wealth of information. This will change the way we value decision makers on the court. It is generally understood who the best decision makers are in the league. The article from Kirk Goldsberry points out that Chris Paul holds the most "points added" crown for the 2012-2013 season and that would seem to makes sense. In his case this stat only affirms what we see with our eyes. But for the fringe players and D-Leaguers these stats could be the difference between a 10 day contract or a trip overseas.
You might have noticed the copious use of 'could' and 'would' throughout this article. This is largely because all of this probably wont be happening for at least few years at the very earliest. Goldsberry notes that the amount of data that would need to be collected is just too huge for most NBA teams to analyze. The biggest deterrent for owners will probably be the cost of implementing this data into their organization but as technology advances that cost should be mitigated.
If anything, the people over at Harvard have only demonstrated that we still have tons of potential new uses for the data the SportsVU cameras gives us and that's pretty exciting in and of itself.
This new stat is pretty exciting for a number of reasons but what excites me most is that this model can actual put a value on decision making in real time. By calculating a players tendencies to pass or shoot at any given position on the court we can, using this model, actually weigh a players decision to pass or shoot accurately. The implications of this stat are huge. EPV gives us our most accurate look at decision making to date. Eventually, coaches could identify the exact moments during an offensive set or action that would produce the most points for that possession with the given personnel. With this data it could be a possible to create new offensive and defensive philosophies that the NBA has't seen yet. This data will make clever coaches even more clever.
Coaches aren't the only group of people that would benefit from this new wealth of information. This will change the way we value decision makers on the court. It is generally understood who the best decision makers are in the league. The article from Kirk Goldsberry points out that Chris Paul holds the most "points added" crown for the 2012-2013 season and that would seem to makes sense. In his case this stat only affirms what we see with our eyes. But for the fringe players and D-Leaguers these stats could be the difference between a 10 day contract or a trip overseas.
You might have noticed the copious use of 'could' and 'would' throughout this article. This is largely because all of this probably wont be happening for at least few years at the very earliest. Goldsberry notes that the amount of data that would need to be collected is just too huge for most NBA teams to analyze. The biggest deterrent for owners will probably be the cost of implementing this data into their organization but as technology advances that cost should be mitigated.
If anything, the people over at Harvard have only demonstrated that we still have tons of potential new uses for the data the SportsVU cameras gives us and that's pretty exciting in and of itself.
Comments
Post a Comment