Advanced Analytics in Basketball

Written by Iván Fernández and Edgar Paz

Iván and Edgar are basketball analysts and founders of Pick StatsGal

Statistics has always been closely related with basketball. There has always been a global interest to know who was the player with most scoring points, how many rebounds a team could grab... but all these basic figures were sometimes difficult to translate to a player’s impact on a team or to measure how good a team was. Is it really better the offense of a team scoring 90 points per game than a team scoring 80 points per game?

Then, advanced statistics were introduced to basketball: the concept of basketball as a turn-taking game and metrics like pace to measure the number of possessions in a game brought a new vision to stats. Mathematical regression models to see the real impact of a player in a team and statistics based on different player combinations allowed staff to evaluate performance in a much more accurate way.

The power of stats has been so strong that it has expanded into the most important leagues worldwide, bringing new technology to record as much information as possible for later analysis. This expansion has proven so successful that it has changed the way teams design plays—focusing on open 3-point shots in catch-and-shoot situations or using pick-and-roll for advantages.

Teams that are winning championships dedicate more effort to these analyses each year because the competitive advantage they deliver helps teams achieve their goals and titles.

It’s in the NBA where advanced analytics have had the greatest impact, and it’s where some of the most innovative metrics have been developed, which we’ve been able to apply to FIBA ​​basketball as well, such as APM (Adjusted Plus Minus) and RAPM (Regularized Adjusted Plus Minus). APM estimates a player’s influence on a team’s point margin through mathematical regression, taking the point margin of each combination of players who participated in each possession and applying mathematical regression methods.

An improved method for APM is RAPM, which attempts to correct the overfitting inherent in the regression used in APM, which is particularly prevalent in players with small sample sizes and introduces bias into the final results. For more details on APM and RAPM, see Adjusted Plus-Minus: An Idea Whose Time Has Come by Steve Ilardi and Measuring How NBA Players Help Their Teams Win by Dan T. Rosenbaum.

It is important to mention that the volume of data for this type of regression in FIBA ​​basketball is usually much smaller than the volume of data available in the NBA—82 games per season; 48 min per game—, where data from several consecutive seasons are often used to eliminate the bias of small samples, so the relative value of this type of metrics applied to a limited number of games must be given.

Another highly interesting metric we’ve been able to apply to FIBA ​​basketball is Box Plus/Minus (BPM), developed by Daniel Myers for Basketball Reference. BPM, unlike APM and RAPM, is based on both individual and team Boxscore statistics, which are used to calculate a player’s contribution. Different coefficients are applied to these Boxscore statistics, which are weighted and combined until the final value is obtained. The use of coefficients and the way in which the Boxscore statistics are combined were not designed subjectively or randomly, but were calculated so that the final results of the metric were as close as possible to the players’ RAPM, calculated over a sample of 14 seasons.

We’ve also applied other metrics designed to analyze specific aspects of the game, such as BoxCreation and Offensive Load, created by Ben Taylor for Backpicks. These metrics attempt to measure a player’s generation capacity and usage in a way that’s different from the traditional USG% by also including his generation capacity data.

The use of these and other metrics, but especially visual analysis, a variety of external tools, and the design of reports tailored to the needs of the coaching staff have become essential for any basketball team to achieve its goals.