Context

I was having an everyday conversation with two of my friends here about how much programming knowledge we need for our college classes. One of my friends is extremely knowledgeable on basketball statistics; he can recall seemingly randomly stats for almost any college player. When we learned about his process for writing articles and making conclusions based on data, we realized that using machine learning would expedite his process almost immediately. So the first step would be to compile all the data we need.

Content

Some of the statistics are obvious such as points, blocks, etc. However, some advanced statistics employ complicated equations, such as offensive rating and PORPAG. These statistics all need to be taken with a grain of salt, since some can be misleading. Specifically, plus/minus may seem to be an effective statistic for ranking how much of an impact players have on their team, but this can be heavily impacted by rotations.

For instance, on my favorite NBA team, the Golden State Warriors, plus/minus is almost irrelevant, since any player that is on the court with Stephen Curry almost always has a much better plus/minus than players who are forced to play without his presence on the court.

However, with the sheer bulk of stats present, I'm hoping there will be clear patterns that emerge with further digging into the data.

Acknowledgements

Avinash Chauhan and Logan Norman, who helped inspire this idea.