Big Data in Sports: Predictive Models for Basketball Player's Performance

Dae-Jin Lee; Garritt L. Page

doi:10.33774/miir-2021-h4x62

Sports

Search within Sports

Big Data in Sports: Predictive Models for Basketball Player's Performance

13 June 2021, Version 1

Working Paper

Show author details

Abstract

Aryuna is a platform that allows to perform advanced data analytics of men's professional basketball statistics of the last 16 seasons in more than 25 professional leagues and 71 FIBA tournaments. The complete database consists of more than 37,000 games and upwards of 20,000 players. Based on a historical database, the report aims to: characterize the performance curve, peak and optimal age in professional men's basketball using performance ratings of players in top basketball leagues; determine a rating correction factor for different basketball leagues, which accounts for intra-league and cross-league variability as well as for player characteristics (position, age, player ratings, etc.); determine which are the most important factors for predicting future outcomes of a basketball player.

Content

Comments

Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here .

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Great read! Especially liked the comparison between Argentina and EuroLeagues and narrowing down to the 13 variables that are most likely to predict a players success. I'm wondering if there's some redundancy between the variables and the same 'variablex100posessions', as they are highly correlated? Would also like to see this study enhanced with more deep analytics metrics that have come on the scene recently, especially with the rise of unicorns like Jokic who are strong in these metrics, but even stronger in similar metrics. I've seen some interesting data from a group of researchers at <a href="https://leans.ai">LeansAI</a> who studied very similar metrics in the team network context to see how variables of interest such as assistsx100posessions plays into the team strength beyond a single player. Thanks for this work and keep it up! Cheers!