Count ball screens defended, not minutes played

Minutes are a poor measure of defensive workload because two players on the same floor can face wildly different demands.

Thirty-four minutes. That is what the box score says about a guard's night, and it is one of the least informative numbers we still print.

Two guards play thirty-four minutes. The first is hidden on a non-shooting forward, chases him through nothing, and defends four ball screens all evening. The second is the point of attack for a team that runs a pick and roll on almost every half-court possession, which means he fought through screens something like fifty times, changed direction on each one, and did it while being the player the opponent was specifically trying to tire out. Same minutes. Not the same job, not by a factor of anything you would call small.

Load conversations still run on minutes because minutes are easy and available going back decades. Teams internally have better inputs, and the one I would put first is a simple count of screens taken on at the point of attack, split by whether the player went over or under. Going under is a rest. Going over is a collision followed by an acceleration, and it is the acceleration that costs.

There is a tactical consequence, not just a medical one. If you know that your opponent's best defender has been dragged into forty screens by halftime, the correct thing to do in the fourth quarter is not to attack him with your best scorer. It is to attack him again with the same action, because the failure will not come from a mismatch, it will come from a leg that will not push off the floor for the fiftieth time.

Teams that hunt this way are accused of being boring. They run the same thing repeatedly and analysts complain about a lack of imagination. The imagination was in the decision to be repetitive.

The counterargument is that screen counts are noisy, since a screen defended by going under and a screen defended by chasing over the top are wildly different events. Fair. Then weight them. Any weighting at all beats the current position, which is that a defender's workload is described by a clock that does not know what he was doing while it ran.

If a player is worth resting, rest him from the screens, not from the game.