Thursday, April 17, 2014

Dan Stafford Final Project: Basketball Japan League Player Efficiency Ranking Tool




Executive Summary

The Basketball Japan League (BJL) is professional league and the purpose is to provide entertaining basketball for fans, and earn money for the teams. It was founded in 2005 with 6 teams; the league has expanded rapidly, and today there are 21 teams. Due to rapid growth the game rules, league administration, and head coaches are constantly changing. Following the end of the 2010-2011 season 13 of the then 16 head coaches lost their jobs, including the coach of the championship team.

What does it take to be successful as a coach? The NBA, which is the world authority on professional basketball is moving increasingly toward analytics—the statistical analysis of success drivers—in order to maximize winning. NBA Coaches that are successful pay attention to analytics to guide decision making. NBA.com and ESPN.com both list efficiency ratings (EFF) and performance evaluation ratings (PER), in addition to clutch ratings and other important measures. The BJL website contains nothing beyond basic player stats. Coaches in this league are blind to analytics regarding their own team as well as their opponents.

What I have built is an EFF ranking system for the BJL. This can be used by a coach to see the most effective players on their own team as well other teams. This data is useful in making decisions about playing time, line-up dynamics, and as a starting point for studying other trends that go beyond data such as strategy formulation.


The EFF ranking system pulls cumulative season statistical data for every player in the league, and then runs that data through a formula that adds positive contributions and subtracts negative ones to produce a raw contribution number or EFF. The EFF is then divided by minutes played to give the efficiency per minute (EFF/MIN) rate, this rate can be used to compare players across the board. The final step is to rank the players in order from highest EFF/MIN to lowest. 

No comments:

Post a Comment