Recently I read an article written by Harlan Schreiber over at Hoops Analyst. Harlan noticed that some big name players have been shooting poorly this year (which he defined as having a FG% worse than 37%), and he wondered if there were more players shooting terribly this year than in previous years. Using both a game (30 games) and a minute (15 MPG) cutoff, he found that there were two seasons that stood out — this season and 98-99 — which are both lockout years.
Well, I’ve looked at the data, and I disagree. I think there are some problems with Schreiber’s methodology. First of all, I don’t see the point of even bothering to look at FG% when eFG% is a much better way to evaluate field goal shooting. Second of all, rather than using both a game and a minute requirement, I’d rather just use a minute requirement. This is because I only care that a player has played enough minutes for their stats to be meaningful. It also avoids running into a problem that Schreiber ran into: how do you correct for a shorter season? Schreiber decided that the game cutoff for the lockout years would be 20 games instead of 30, but not only was this arbritrary, it also introduced a difference between lockout seasons and non-lockout seasons that may account for his results.
So how would I do it? I’d look at eFG% instead of FG%, because eFG% corrects for the added value of three point shots. I’d also like to look at TS% (which is like eFG%, except that it also corrects for the added value of free throws) and FT%, just in case there are some similar trends in those areas as well. The actual percentages I ended up using (TS% =<0.45, eFG% =<0.40, and FT% =<0.50) aren’t important; what really matters is picking a benchmark that is hard to beat and examining the trend. And my playing time cutoff would be 400 minutes, which is the minimum amount of minutes played during a season that I feel comfortable using. Using these parameters, I looked at seasons dating back to the 79-80 season, which was the first season to use the three-point shot.
Under these parameters there are some fairly obvious trends. There were more bad shooters in the NBA from 98-99 to 03-04 — almost twice as many — than there have been during the current season so far, and this is true whether we use eFG% (the red line) or TS% (the blue line) to determine who a bad shooter is. Also, there is an upward trend with regards to the number of bad free throw shooters (the green line) in the league. From 79-80 to 93-94, there were only 2-7 bad free-throw shooters each season. But from 94-95 onward, the number of bad free-throw shooters has varied between 9-18.
Going even further, we can even see the effect (or lack of effect) that two rule changes have had on the number of bad shooters. In the graph above, the area highlighted by the thicker green rectangle represents the three seasons where the three-point line was closer to the basket; oddly, it doesn’t appear that the closer three-point line had any effect on the number of bad shooters. But the thinner green rectangle highlights the first year of the NBA’s new hand-checking rules, and that appears to have had an effect. Ever since those rules were enacted, the number of bad shooters has been much lower.
But that isn’t to say that the lockout hasn’t affected the number of bad shooters. In the graph above, the red rectangles highlight the two lockout seasons. The 98-99 season did have a small spike in the number of bad shooters as determined by TS%, but not eFG% or FT%. And the current lockout season seems to have increases in all three areas. So the lockout does appear to have had a small effect on the number bad shooters.
But before we tout these results, there’s something else we should probably control for. A lot has changed since the 79-80 season; there are now 30 teams in the NBA instead of the 22 that played that season. Accordingly, there has been a rather large increase in the number of players meeting the 400 MP cutoff (from 237 in 79-80 to a peak of 360 during the 10-11 season). We should make sure that the increase in the number of bad shooters is not simply due to the increase in the number of players meeting the cutoff. Doing that makes the graph look like this:
The graph has flattened somewhat, but the story is basically the same. There were more bad shooters in the league during the 89-90, 91-92, and from the 97-98 to 03-04 than there have been during the current season so far. Bad free-throw shooting is still trending upwards — it seems that coaches and GMs have become more tolerant of players who can’t make their free throws – but the peak seasons for bad free-throw shooters were the 97-98 and 98-99 seasons.