Plotting 2 EPA means against each other is.. unsatisfying. In this code-through, I try to do a little better. I use depth medians to measure the central tendency of the (Off_EPA, Def_EPA) duple, then build a plot that uses depth medians plus deviation estimates to express the confidence interval of the estimate.

Is there some advantage for NFL teams to pass when they "should" run, and to run when they "should" pass? Let's take a look!