An "Advanced Basketball Statistic" That Makes SenseIt should be obvious that neither field goal percentage nor an "advanced basketball statistic" like true shooting percentage provide a complete description of a player's shooting ability and/or scoring prowess; a player who posts gaudy numbers in one or both of those categories may be a limited offensive contributor who can only score effectively in a very specific role (i.e., a big man who can only convert from point blank range or a slow-footed perimeter player who can only make spot up jumpers). A player's offensive efficiency can only be accurately determined by examining his ability to create shots for himself and his teammates, his ability to score from a variety of areas on the court and his ability to draw double teams that can break down the opposing defense. Such evaluations can only be made by an informed observer who watches the sport with an objective eye. While baseball is a game that consists of a series of discrete actions that can be isolated and analyzed individually, basketball is a much more complex game in which every action by one player affects and is affected by the actions of several other players.
A player who shoots .600 from the field but cannot make a shot outside of the paint and is easily defended one on one is not nearly as valuable as a player who shoots .450 from the field but can score from anywhere on the court and must be double-teamed. Is there a way to reasonably compare two players who have such divergent skill sets? Kirk Goldsberry's solution to this problem is a new statistic that he and fellow Michigan State professor Ashton Shortridge devised: ShotScore ranks every NBA player's shooting prowess based on the relative difficulty of each shot that he took; Goldsberry and Shortridge determined the average NBA field goal percentage from every spot on the floor, compared the average percentages to each player's percentages from those spots on the floor and then expressed the results in terms of actual points scored versus expected points. For example, based on the location of LeBron James' field goal attempts last season he would have scored 1397 points if he had shot an average percentage but he scored 1628 points, giving him a league-best +231 ShotScore. Kevin Durant ranked second (+204) and Stephen Curry placed third (+164).
There is little doubt that ShotScore is a more precise measurement of a player's shooting efficiency than scoring average and/or field goal percentage. The usefulness of ShotScore and the measured tone of Goldsberry's writing are a welcome contribution to basketball theory--and a marked contrast to the way that far too many "stat gurus" make arrogant and bold declarations that are unsupported by facts/objective observations. However, there are some potential drawbacks to ShotScore: (1) it relies heavily on the completeness and accuracy of play by play data regarding shot locations and (2) a person must have access to a tremendous amount of data in order to calculate a player's ShotScore. ShotScore could be very useful for general managers and coaches but it will be difficult for it to become a mainstream statistic unless/until the accuracy of the play by play data can be objectively proven and unless/until it becomes easier to compute each player's ShotScore; during a game, an informed observer can note the areas in which a player is effective and can instantly calculate his field goal percentage to get a "quick and dirty" estimate of his overall efficiency but there is no way that such an observer can instantly compute a ShotScore unless/until the NBA provides such data in real time (which could perhaps happen at some point).
posted by David Friedman @ 4:52 PM