In over 80% of NBA games the team with the higher field goal percentage (FG%) wins the game. This is by far the most important statistics when it comes to winning basketball games. What are the factors influencing the FG%? There are many variables but I would say mainly the shooting player's quality and their shot selection. How can we tell what is a good shot? With the new SportVU data, introduced at the 2013-14 season, we have information on the closest defender distance from the shooter, the shot distance, the number of dribbles before the shot and a few other shooting parameters. In this trilogy of posts I will explore the affect of the shot distance, closest defender distance and dribbling on the FG%.
To get some idea of how the defender's distance from the shooter affects the FG% I used all the NBA shooting data since 2013-14 until today and mapped it into a 2D map where the y axis is the closest defender distance and the x axis is the shot distance and the number (and color) of each pixel represent the FG%:
How do we read this map? For example, top left are wide open layups, bottom left are highly contestant layups, top right are wide open threes and bottom right are contestant threes. Obviously the closer the defender and the further the shot distance the lower the FG%.
Let's choose a few representing distances (see illustration below) - 0-2 feet, 8-10 feet, 16-18 feet and 24-26 feet:
Now let's plot the FG% as a function of the closest defender distance for those shot distances:
The first thing that caught my eye when looking at this data is that the FG% never reaches a 100%. Wide open layups are scored at about 95%. This makes me feel much better about the wide open layup, where no defender was at sight, I missed in high school. When shooting right under the basket (0 feet), the defender's distance dependence on the FG% is linear. Roughly, for every 1 foot of distance the FG% changes by 6%-7%. The FG% does not change anymore when the defender's distance is above 8-10 feet which means that a wide open layup is when the closest defender is at most 8 feet away.
For threes (24 feet), the effect on the FG% is not as linear. The FG% changes from ~25% at 1 feet to 40% at 9 feet. About 2% FG% change for each foot. The FG% stops changing when the defender's distance is about 8 feet away.
The FG% does not change drastically when comparing shoots from 8 feet to shots from 16 feet.
We can plot the FG% as a function of shoot distance for different defender's distance (0-2 feet, 2-4 feet, 4-6 feet, 6-8 feet):
The data can be divided to two different regions - shooting distance below 8 feet, where the FG% increases drastically as the shooter gets closer to the basket, shooting distance above 8 feet, where the FG% drops slowly with increasing shooting distance. We can again see that for close range shots (< 8feet) the defenders distance to the shooter has a much larger influence on the FG%.
Finally, we can also take into account shooting off the dribble vs. catch and shoot:
For almost all shooting scenarios, shooting off the dribble results in lower FG%. The biggest difference is when shooting close to the basket (black curve) where the FG% can be almost 10% lower when shooting off the dribble. Also, when shooting 3's when the defender is very close, there is a very big difference in FG%. Other than those two cases, shooting off the dribble has only a minor effect on the FG%.
Conclusions:
When shooting close to the basket the closest defender's distance has a large effect on the FG% (6%-7% for each foot away from the shooter).
Shoots further from the basket are not effected as drastically from the defender's distance (2%-2.5% per foot).
Shooting off the dribble adds to the complexity of the shoot mostly when shooting close to the basket or long contestant 3's.
The NBA defines a wide open shoot when the defender is 6 feet away or more. While for some shooting distances this seems to be correct, if I had to choose one number the data suggests that 8 feet is a better choice.
What's next? Now that we have a better sense of how the defender's distance, shoot distance and dribbling change FG% we can look at individual players and their shoot selections.
Do you enjoy this blog? If you enjoy this blog please let me know. Please comment if you have any thoughts. If there is enough interest in this blog, I can convert it into a website with some of the statistics updating on a regular basis.
Follow new posts on twitter: @eyal_shafran
Technical information:
In the heatmap (1st figure) some of the results are noisy. This highly depends on the number of shots taken from each spot. There are scenarios which do not happen very often. For example, when the defender is 12 feet away and a shooting distance of 6-10 feet. In the next post I will explore the most common shooting scenarios in the NBA.
The SportVU data does not always agree with other NBA data. When trying to compare two different shot distance sources from the NBA API I noticed that sometimes there is a large discrepancy between them. In this post I used the data from http://stats.nba.com/stats/playerdashptshotlog
For all figure (except the 1st one), I only included data points where at least 50 shots were taken to insure sufficient statistics.