README.md

A fun little summer project to predict how good NBA players can become based on their rookie stats. 🏀

NBAPredict.io is now longer operational. Feel free to peruse through the code and check out some of my more recent NBA related work.

Some popular searches

KEY: 0+ (>0% probability of stardom), 1+ (>10%), 2+ (>20%), 3+ (>45%), 4+ (>75%), 5+ (>90%).

Player Name	Prediction
Ben Simmons	5.00
Lauri Markkanen	3.75
Donovan Mitchell	5.0
Jayson Tatum	3.75
De'Aaron Fox	2.5
Malik Monk	1.25

Some (of the gazillion) numbers we track

Metric	Purpose
per	overall efficiency
bpm	box plus/minus
vorp	value over replacement
usage	usage %!
tov%	mistake rate
ts%	scoring skill
war	record orientation
WS	win shares (!)

How it works

We look at a broad sample set of (current and former) NBA players and their rookie statlines. Once we've done that we begin to track certain "between-the-lines" metrics that can give the best insight as to what rookie tendencies and on-court habits resulted in the development of stars. We then apply that model on n number of players' rookie statlines and then can see whether or not they may become stars. A good example of this would be Jimmy Butler. As a rookie, Jimmy Butler averaged 2.6 points and other absymal on-the-surface numbers, but our model was able to indicate that he had a solid chance at stardom.

All predictions and ratings are not promised to be representative of exactly how good each respective player will ACTUALLY become (but we shall try our best). NBA-Predict is just a fun summer project which in no way is intended to make certain players look worse than others. All predictions and ratings are based off factual data. Our prediction model is oriented around favoring numerical, statistical prowess. Things like importance of games and presence of the "killer instinct" are never accounted for.

https://www.youtube.com/watch?v=AZUKEqkyJw0