|L|east-Squares |E|rror |B|ased |R|egressor for |O|ffensive and defensive |N|umbers (LEBRON)

jc-gellidon-361849.jpg
 
 

Summary

I’m an avid fantasy basketball player and NBA stat-head, so when the opportunity to do a linear regression project came up, it seemed like the perfect application given my domain knowledge and interest.

I developed an AR-3 model that also used a number of other natural and engineered features (e.g., projected playing time, team pace) to predict 9-cat fantasy basketball statistics for players with at least three years of NBA experience.

At its most complex, LEBRON used 81 predictor variables for the predictive models for each of the 12 stats needed to predict 9-cat fantasy basketball.

As LEBRON got more complex, I was happy to see the predictive accuracy increase, although, there were diminishing returns with increasing sophistication.

Blog post to come.