I Built a Model to Predict the NBA's Next Stars

I got tired of arguing about young NBA players with absolutely nothing to back me up, so naturally I spent an unreasonable amount of time building a model to predict which players are most likely to break out next. For this video, I went through years of NBA data, Most Improved Player voting, walk-forward validation, regression models, and even used semantic analysis to figure out which players the internet already thinks are going to be stars. The goal was simple: find the guys who might be next before everyone agrees they're next. Somehow this channel also just hit 10,000 subscribers, which is genuinely insane to me. The support on the last video was way beyond anything I expected, and seeing so many people enjoy these weird, overly complicated basketball experiments has been incredibly motivating. Seriously, thank you for watching, subscribing, commenting, sharing the videos, or just giving them a chance. I have a lot more fun ideas I want to do and play around with Enjoy the video :) — THEHQE DATA / METHODOLOGY Player statistics and historical data were collected from publicly available NBA/statistical sources. Historical Most Improved Player voting was used to help define what a “breakout” looked like rather than manually choosing successful players after the fact. As always, this is an experiment, not basketball prophecy. If all of these players are terrible in six months, please pretend this video never existed. BUSINESS INQUIRIES inquiries@thehqe.com Data sources: • NBA / NBA.com, they also have a python API which sometimes works. • Basketball Reference • Additional publicly available basketball data referenced throughout the video