Today’s NBA game is, in many ways, unrecognizable from it’s mid-90s self. One of the most prominent shifts has been the rise of the 3-pointer. It’s no coincidence that the 3’s rise has happened hand-in-hand with the move toward advanced analytics, maximizing offensive efficiency, and the use of sophisticated in-game camera technology. The trend has been most embraced and emblemized by the Houston Rockets, their shooting guard James Harden and GM Daryl Morey. Supporters say the 3 point threat has made the game more exciting by spacing the floor and spurring positionless basketball. Critics say it’s dulled the game by reducing shot diversity. I wanted to rub the stats between my fingers and see for myself how the 3-pointer has evolved. I was astonished at what I found.
The purpose of this analysis is to explore and visualize the evolution of the NBA 3-pointer. The dataset, downloaded from Kaggle and originally scraped from Basketball-Reference, includes season stats for every NBA player from 1950 to 2017. I focus on observations from 1980, the year the 3 pointer was introduced, to 2017 and on the variables Player, Position, 3PAr, 3P, 3PA, 3P%, Points, PER, and Team, all of which I’ve detailed below. I sought to answer 3 burning questions:
- How much have 3PAr, 3P, 3PA, and 3P% changed since the inception of the 3 pointer?
- How do 3 pointers look among teams and positions currently?
- Is there any correlation between PER, PTS and 3 point metrics?
My analysis includes transforming, cleaning, and subsetting dataframes, grouping data using groupby, setting and styling line, box, density, and multidimensional scatter plots and heatmaps using seaborn, and making exploratory statistical observations by year (time series), position, and team.
My code, explanatory notes, observations, and visualizations are below if you’d like to explore the data yourself.
We’re just scratching the surface. If you’d like to dig in further on this or any other topic, reach out to info@crawstat.com!
Stay tuned, future posts will include player analyses, shot charts, team stats, a SuperSonics deep dive, defensive stats, draft prospects, and more.
Next time, I’ll apply the concept of simulations from my previous post Danae & Shaelo on Strawberry Summit to bring to life the effects of non-representative sample and sample size.
For the swish,
Rish
