I spent the past weekend using basketball data to develop my knowledge of a couple statistical software tools: R, for analysis, and Tableau, for visualizations. In the process, I discovered a relationship I found interesting and thought I’d write about it. The idea was to look at NBA players’ percentage of field goals assisted (%AST) vs. their true shooting percentage (TS%). It stood to reason that if a player was making many field goals of assists from others, that he was not having to create his own shot and thus getting better looks at open shots. That in turn would lead to a higher shooting percentage, so logically one might expect to see a positive correlation between %AST and TS%.
Before I delve into the results, I’d like to provide a quick explanation of TS% and how it deviates from traditional FG% (Field Goals Made/Field Goals Attempted). TS% takes into account that a 3-pointer is worth more than a 2-pointer and also incorporates free throws. If interested, more information can be found here. One problem with this analysis actually has to do with TS% including free throws, but I’ll come back to that later.
For the data sample, I looked at the 90 players who play at least 30 minutes a night, from Luol Deng down to Tyreke Evans. The minutes played filter served as a proxy for players who shoot enough for their %AST’s and TS%’s to stabilize at a certain level. The reason for the shortcut was simply to make data collection less time-consuming.
Now I could get to the question at heart- what’s the relationship between %AST and TS%. Put simply, there isn’t one. In this sample, there exists only a weak positive correlation (r ≃ 0.15) between the two variables, and one that’s not statistically significant. Below is a scatterplot of the data along with the best-fit line (yay Tableau!). It can also be filtered by position.