If you want to skip ahead and look at the final results of this process, click the links below.
The Problem In Current Shooting Metrics
Every statistic has its flaws.
Field Goal Percentage (FG%) is the most basic shooting metric in the books. It’s simply defined as Field Goal Makes / Field Goal Attempts. The issue here is that all field goal attempts are not created equal. Shots at the rim are converted at a far better rate than long jump shots. Unsurprisingly, the NBA leaderboard for career field goal percentage is completely dominated by big men who primarily reside in the paint.
Effective Field Goal Percentage (eFG%) was derived as a statistic that would adjust field goal percentage to account for the fact that three-point field goals are worth one more point than two-point field goals. The formula which accomplishes this is as follows:
(Two Point Makes + 1.5 * Three Point Makes) / Field Goal Attempts
That’s still not perfect. While eFG% is far better than FG% and is often used today, the NBA leaderboard for career eFG% still mostly consists of big men.
Certain shots still have a higher chance of being converted, even with two-pointers and three-pointers isolated. A contested long mid-range jumper is far harder and less efficient than an open layup, but eFG% cannot recognize the difference between the two…
Expected Effective Field Goal Percentage takes into account two additional factors in order to act as a better representation of shooting efficiency: Shot Location and Distance From Nearest Defender.
Shot Location is the distance between the shooter and the basket at the time of the shot. The data on exact distance in feet isn’t available, only certain shot ranges: two-pointers (<10 feet), two-pointer (>=10 feet), three-pointers. Through this qualifier, post players are successfully distinguished from jump shooters.
Distance From Nearest Defender represents the strength of the defensive contest. The four categories for this are 0-2 feet (very tight), 2-4 feet (tight), 4-6 feet (open), and 6+ feet (wide open). By including this variable, we can account for the fact that open shots are far more efficient than contested shots.
Every shot in the NBA since the 2013-14 season can be indexed and placed into different categories using both variables. The league average field goal percentage for each type of shot can be calculated. For example, this season the average NBA player is converting wide open shots within 10 feet at a clip of approximately 90%. We can account for the incredible efficiency of these shots, while not overshooting how much we reward players for hitting these shots.
By knowing the league average percentage for each type of shot, we can determine every player’s unique Expected Effective Field Goal Percentage (XeFG%). XeFG% is most simply defined as the eFG% a league average player/team would shoot if they took the same exact shots as the player/team in question. This value isn’t meant to be looked at alone — it can be compared with a player’s actual eFG% to see if they are performing above (positive differential) or below (negative differential) expectation. A look at the league’s current standings in XeFG% reveals that big men unsurprisingly have the highest XeFG%, while jump shooters have lower XeFG% values.
Of course, having a high XeFG% isn’t supposed to be considered poor. If a player has a high XeFG% but manages to match that with their actual eFG%, there’s nothing wrong with that. Shot selection (the ability to take smart, efficient shots) is an important skill in the NBA.
The main purpose of the statistic is to provide context to the type of shots a player is taking. The final goal may not always be to compare that to their actual shooting efficiency, but it’s an available possibility.
The same reasoning behind this is what led to the creation of Expected Effective Field Goal Percentage Allowed (DXeFG%). DXeFG% is essentially the same concept as XeFG%, except it is used to gauge defensive players to see if they’re forcing their opponents to overperform or underperform.
It reduces the inherent positional bias in the typical DFG% (Field Goal Percentage Allowed) metric. Centers typically allow scoring at a higher rate because they’re usually facing shots in the paint while guards are facing jump shots. DXeFG% looks to tackle this issue from a new angle. In summary, DXeFG% is defined as the eFG% a league average player/team would shoot against a certain player/team if they took the same exact shots as actual players/teams did in any given season.
While these methods of contextualization can be very useful, these new metrics are not an exception to the statement I opened with: every statistic has its flaws.
It is very important to understand the deficiencies of Expected Effective Field Goal Percentage before it may be used inappropriately.
One of the main flaws which stand out is the inability to distinguish between players who create their own shots off the dribble and players who catch and shoot. For instance, Kyle Korver has an XeFG% of 48.72% while James Harden has an XeFG% of 50.78%. Korver has made a career out of these moving catch and shoot threes while the vast majority of Harden’s points scored are unassisted. The distance between the nearest defender is similar on both types of shots because of the space Harden generates on his stepbacks, but shooting the ball off of the dribble is less efficient and more tiring.
Also, all three-pointers are treated equally. There is no way to distinguish between a corner three-pointer and one of Steph Curry’s patented bombs from just ahead of the logo. A corner three-pointer is about 22 feet from the basket while some guards let it rip from 30 feet out — there’s obviously going to be a difference in efficiency.
With the defensive version of this statistic, DXeFG%, the flaws are perhaps even more apparent. A reliable defensive metric does not exist and DXeFG% is no exception. Jrue Holiday, a premier two-way guard, allows opponents to shoot at an eFG% that is 0.24% above expectation. While this suggests that Holiday is a below-average defender, he clearly isn’t. The problem is that Holiday guards the best perimeter players on the opposing team. His defensive numbers can’t be compared to those of a defensive liability like Isaiah Thomas, whose team will try to hide on players who don’t pose an offensive threat.
Just like the statistics that came before and the statistics that will come after, improvements are needed.
For now, however, this metric is something new that allows the public to make their own inferences and interpret the data the way they see it.