≡ Menu

The simplest way to measure individual offensive efficiency


Click for a full-sized, interactive version

UPDATE: I received a request to change points attempted per game to points attempted per 36 minutes. That made a lot of sense, so I did. The biggest difference resulting from this change is that I think this chart now very concisely captures the criticism levied against Russell Westbrook. He is 2nd in the league in points attempted per 36 minutes with a whopping 53, just barely behind his 1st place teammate Kevin Durant, who has 53.61. The difference, of course, is that Durant converts on  55.83% of his attempts whereas Westbrook converts on 48.43% of his attempts, or right at the league average. Even infamous chucker Carmelo Anthony attempts to score less, and scores more efficiently, than Westbrook, and he doesn’t have Kevin Durant to pass to. 

Sometimes, probably most of the time, the simplest solution is the best. Admittedly I haven’t always been good at keeping things simple, but I’m going to rectify the situation with this post.

There are quite a few indicators available to measure offensive efficiency. Some combination of usage, effective field goal %, true shooting %, and points per possession is usually thrown around in the name of efficiency. But it doesn’t have to be that complicated. Efficiency is just about two components–number of times something is attempted and number of times those attempts are successful.

In the case of basketball, we’re talking about trying to score and actually scoring. The above chart plots the number of points players tried to score per game against the percentage of their attempts that were converted (points scored / points attempted). For the record, the chart considers all attempted and scored 3pt-fgs, 2pt-fgs, and free throws. Only players who played at least 45 games and at least 22 minutes per game were included.

Some of you might be asking, isn’t points efficiency (on the x-axis) the same thing as true shooting percentage? The answer is no, because a little known fact is that true shooting percentage isn’t calculated the way people think it is. For 3pt and 2pt fgs, it’s as one would expect. But instead of counting the number of free throws a player takes and including it in the formula, true shooting percentage instead uses a constant for all players. The constant represents what portion of a person’s scoring attempts comes from free throws. Worse, the constant was calculated using data from several years ago. So not only is it not player specific (since some players shoot more free throws relative to field goals than others), but it’s not even year specific.

Basically, true shooting percentage is going to be less and less accurate the more an individual player strays from that calculated average constant. It just so happens that the Houston Rockets have two players that fit that mold–James Harden and Dwight Howard. True shooting percentage underestimates Harden’s efficiency because he takes a lot more free throws than average and makes them, but overestimates Dwight Howard’s efficiency because he takes a lot more free throws than average but misses them.

Enough digression. Some things worth pointing out from the chart are:

1) James Harden is 4th in the league at points attempted per game at 48.62, behind only Kevin Durant, Carmelo Anthony, and Kevin Love. Harden is also above average in converting those attempted points at 52.2%.

2) Big men are still more efficient than guards and wings, even though guards and wings are more proficient in shooting threes.

3) You can see how Chandler Parsons’s oddly poor free throw shooting affects his overall scoring efficiency. Despite shooting 50% in FGs and 40% on threes, he’s slightly below average in points efficiency. Whereas Harden, who is worse in FG% and 3ptFG%, makes up for it in a big way by having and converting so many free throw opportunities.

4) Omer Asik attempts to score the least out of all players in the sample, at barely over 10 pts per game. Compare that to Durant, who attempts to score over 57 points per game.

View this discussion from the forum.

About the author: Richard Li is an independent researcher and consultant. He likes numbers and pictures.

in essays