A midfield 3 of Grimes, Torp, and Rudoni is clearly the way to go, although this method of evaluating players is pretty nonsense really.
Sheaf has somehow jumped further across the graph since last time, despite not playing. Rudoni making apparently only 0.1 chances (what a silly variable) per game more than him, yet having about 8 more assists. Torp has a much higher level of vision, and ability to pull off a smart forward move, yet has a pretty low goal assist contribution.
Data like this can be useful, but as I wait for a chorus of laughing emojis, is far too one dimensional to use in isolation. What that paper says paints some irregularities compared to what happens in reality on the pitch overall.