Shot selection is everything, right? On offense and defense, it feels like where opponents shoot from and what types of shots they take should have a huge impact on how good a team is on either end. To look into this more, I used PBP Stats package, which is a great website for some more advanced NBA stats which you can download into a CSV. I used the shot quality numbers available on the site, which calculates expected eFG% based on shot location and play context. The graphs generated from last season's stats are below:
Now, these are really interesting (if you are a basketball nerd). Offensive rating seems to be barely affected by expected shot quality. However, defense does seem to have some correlation when it comes to shot quality. For a defense, lower opponent shot quality seems to have some correlation to lower defensive rating (which is good).
I have a couple of guesses as to why offense doesn't seem to have much correlation. First, teams who have the best offenses tend to have players that take tougher shots. In general, superstars take more difficult shots. That is not only because they can make more, but also because they can create their own looks. When this happens, their shot quality compared to an average player may be lower, but the shot quality for themselves may be higher. An example of this would be that Damian Lillard shoots 40.4% on pull-up threes, whereas the league average may be way lower. Another theory I have is that shot quality doesn't take into account how often a team gets to the line or how often they get offensive rebounds. Getting an offensive rebound gets you another crack at the same possession. Getting to the free throw line is one of the most efficient places to shoot from.
My guess as to why there seems to be more of a correlation for defense is that teams can scheme for where they want their opponents to shoot from. A team like the Bucks can scheme around not letting the other team get to the rim, which is the most efficient spot on the floor. Obviously, there are still the same confounding variables as listed above. However, we can see that the way a defense schemes for shot selection can have an impact on how good their defense is.
Something that is interesting is to look at the same charts, but from the sample of this season:
So far this season, there seems to be very little correlation between shot quality and offensive or defensive rating. Besides all the confounding variable already talked about, I have a couple of guesses as to why this is. First of all, the sample size is simply smaller. I am going to look at this chart again at the end of the season to see how it has changed. My other guess is the craziness factor that has gone into this season. With COVID, there have been a lot of strange results. I also think that teams have had a lot less time to put in their own defensive schemes that could determine opponent shot quality. Whatever the reason, this will be something interesting to monitor in the future.
The code for this project can be found here:
Update:
I decided after I posted this article that sample size may have factored in to why the results from this season and last season were so different. To counter this, I looked at the relationship between shot quality and offensive/defensive rating from the 2004-2005 season to this season. I found the results interesting enough to add to the end of this article.
As you can see, now both defensive and offensive rating how some correlation. While the correlation still isn't incredibly strong, there seems to be a relationship between shot quality and offensive/defensive rating. This obviously makes perfect logical sense, as forcing opponents into tougher shots (pull-ups, mid-range, contested) would make it harder for them to score, just as shooting easier shots makes it easier to score. However, the correlation still isn't as strong as I thought it would be coming into this. This is still an interesting topic that could use more investigation as to why the numbers aren't as strongly correlated.
Comments