I’ve noticed that someone with a lot of nerdy friends who like volleyball has linked my original Attacking Theory post on FB, so I figured I’d give you folks a little extra look on some data from this season. It’s just a snapshot, but it’s a pretty clean look at how these numbers get computed. This is 2018 data this time – X5 is a Go, X6 is a Red, X7 is a 3 to the MB.
So while Illinois did in fact play Stanford, this is not the specific data from that match in isolation. Like before, we must have an input and an output for our hypothetical attacker. In this case, we use the Attack Code as the input (Go, Red, etc). For the output, we look at what follows the attack (kill, error, perfect dig, block to cover by offense, etc).
In the columns, you can see the four columns as described in the first Attacking Theory.
The Stat_Eff_IN column is: when Illinois sets the X5 (Go) against all teams, what is their attack efficiency?
The Versus_Eff_IN: when Stanford defends against the X5 against all teams, what the attacker’s efficiency?
The Stat_Eff_OUT: when Illinois attacks into a (for example) perfect dig (D#), what is Illinois’ efficiency while defending on that possession (often negative since your opponent is attacking now)
The Versus_Eff_OUT: when opponents in general hit into a perfect dig by Stanford specifically, what is the opponent’s efficiency while defending against Stanford in transition (again, often negative).
From these four, we average the two input columns and the two output columns. Finally, we find the difference between Output_Eff and Input_Eff. If Output is greater, the player did a good job putting her team in a better position than what she was given. If the Output is less than Input, the team is in a worse position because of the attack.
Let’s just look at the row with X5 as the input and B+ as the OutputTouch. Illinois hits .230 on the X5. All opponents against Stanford hit about .210. Therefore we expect Illinois hitting an X5 against Stanford to result in a .220 efficiency.
The B+ is a good touch on the block for Stanford. For Illinois facing opponents with a good touch on the block, they win at a -.160 rate (they lose at .160). When all opponents are facing Stanford after a good touch on the block, opponents lose at .223. Therefore we expect Illinois to lose at .192 (or win at -.192 in the data).
Finally, we do some match. Output minus input. -0.192 – 0.220. This gives us -0.412. This means that the attacker, by taking an X5 swing and getting soft-blocked by Stanford, actually loses Illinois about 40% of a point. This is because the X5 is a good situation for Illinois to be in – and the outcome, Stanford in a good situation in transition, is bad for Illinois. This swing hurts Illinois and lowers their ability to win the point – therefore, the attacker should be held accountable for this change.