### By: Brian Suk

One prevalent topic in the modern NBA is the notable shift in the offensive style of today. Back in the 2000’s and earlier, offensive strategy was much slower and all about isolation. Giving the ball to your star player or throwing it to the post and letting them go to work for a long two-pointer or layup was the offensive strategy that appeared to make the most sense. Players like Kobe Bryant and Carmelo Anthony thrived during this era, and offense was much slower. However, with the modern shift towards analytics, spacing, and speed, a particularly heavy emphasis has been placed on the value of the three-point shot: teams are letting it fly from downtown more than ever. Now, instead of isolation with your star, offenses are more predicated on ball movement and more “efficient” three pointers. We can observe this trend in the data, as the league average for 3-pointers per game back in 2010 was 18.0, and currently it sits at 34.9. But has this shift in offensive strategy actually led to more tangible offensive success? In this article, I will compare teams that shoot a high volume from 3 and teams that were efficient from mid-range. The ultimate goal is to perform statistical analysis on teams’ offensive rating vs. playstyle to see which offensive style, if either, leads to greater offensive success.

First, let’s examine the statistics that we are working with and determine what they mean. Offensive rating is the number of points produced by a team per 100 total possessions. So it represents the extrapolation of how many points a team would score given a hypothetical number of 100 possessions, based on their current offensive statistics and points scored each game. It measures the offensive efficiency, and often, success rate of each team: a higher offensive rating correlates to a better offensive team, in most cases. I decided to use offensive rating instead of net rating for this experiment because, although net rating might be a better overall measure of a team’s success and winning, the defensive component of the net rating may act as a confounding variable and conflict with the objective of comparing offensive strategies. Therefore, although a team may have a high offensive rating, depending on their defense and net rating, they may or may not have greater team success, and this experiment is purely meant to test their offensive efficiency and capabilities. The statistic of three-point attempts per game represents the number of shots behind the 3-pt arc each team shot per game, and 2-pt percentage represents the percentage of shots that went in from inside the 3-pt arc, referred to as “mid-range.”

With that established, we can move towards the different statistics against which I plotted offensive rating to measure the success of each strategy. The first statistic I plotted was 3-pt attempts of each team per game, and this was meant to represent the strategy of high volume from 3-pt range. In addition to the scatterplot, I also performed a linear regression on the data, in order to create a line of best fit that sufficiently represents a general trend in the data, as modeled below.

As can be observed, there is a general increasing trend in the graph, with Offensive Rating increasing with 3-pt attempts per game. This seems to suggest at least somewhat of a positive, linear relationship: the correlation coefficient of the data is approximately .509. An interesting statistic from this data is that teams who shot 35 or fewer 3-pointers per game didn’t seem to have any particular correlation between their 3-point attempts and their offensive rating, with the data being indefinite. In other words, teams that shoot a middling amount of 3-pointers or less were not particularly affected in their offensive rating. However, we can clearly observe that teams who shot more than 35 3-pointers per game have a clear increasing trend in offensive rating, and all of these teams have relatively high offensive ratings in general. This seems to show that teams who shot an extreme amount of 3-pointers relative to their competition generally tend to indeed possess a higher offensive rating, but teams that are not as reliant on their 3-point volume from game to game have varying degrees of success. In other words, we can observe that teams who fully commit to launching from deep every game have consistent offensive success, while teams that shoot a modest number are more inconsistent in their offense. Analytically, this result makes sense, as 3-pointers are widely regarded as the most “efficient” shot in the game due to the general increase in ability of modern day NBA players to shoot the three. So as a team shoots more, they are essentially getting more points per possession if they have skilled players who can consistently knock them down. The league average 3-point percentage back in 2011 was 34.9%, and has increased by a full 2 percentage points (which is a lot in basketball terms) to 36.9% today.

The second set of data I plotted was offensive rating against 2-point percentage. This plot represents the strategy of efficiency from the mid-range, and again I performed a linear regression on this data below.

Now with this data, there is a much clearer trend throughout the whole data, as we can see a positive, linear relationship for the data set as a whole. This trend is supported through observation of the correlation coefficient of .713, which is higher than the value for 3-point attempts of .509. Teams who shoot a higher percentage from mid-range almost universally tend to have a higher offensive rating. If a higher percentage of attempts score, there will be a higher offensive rating. We can compare the two strategies through a confidence interval test of each of the slopes of each dataset. Since we have a small sample size of 30 teams, we can utilize a t-distribution. The estimate of the slope of 3-point attempts through linear regression is 0.4996, and with a critical t-value of 2.05 and a standard error of 0.1596, we calculate the 95% confidence interval to be [ 0.173 , 0.813 ]. Interpreting this, we can expect a slight positive, linear relationship overall in the data through the positive slope. Similarly, for the 2-point percentage data, the estimate of the slope is 1.612, and using a standard error of 0.2995, we calculate the 95% confidence interval to be [0.999, 2.226]. As supported by the correlation comparisons earlier as well, 2-point percentage seems to possess a slightly stronger positive, linear relationship with offensive rating. Both the estimate and confidence interval of the slope seem to be higher for 2-pt percentage as well. So does this mean efficiency from the mid-range is automatically the better strategy? It turns out that it is not quite so simple. I dove deeper into the data and plotted offensive rating against 2-point attempts as well. This will represent teams who tend to shoot a large volume from the mid-range.

This data set paints a different picture than the other two strategies, as there looks to be very little correlation overall, and, if anything, there is a negative relationship between offensive rating and 2-point attempts sloping regression line. The correlation is 0.245. This suggests that shooting more two’s leads to less offensive success if anything, and the data point in the top left corner representing the Houston Rockets shot the least amount of two’s in the league at 45.2, and consequently had the 2nd highest offensive rating in the league at 117.8. So from these results, we can clearly see that an offensive strategy built around volume from mid-range is much worse than a strategy emphasizing volume from three.

So which strategy is better, volume from three, or efficiency from mid-range? In basketball, the two strategies are not mutually exclusive. From our analysis of the three sets of data, it seems the optimal combined strategy for a team is to shoot as many 3’s as possible, while simultaneously shooting as few 2’s as possible at the highest efficiency. Shooting a high efficiency from mid-range does not necessarily mean utilizing an offensive strategy that emphasizes mid-range shots, as seen by the contrasting positive and negative trends in 2-pt efficiency vs 2-pt volume. Rather, it could mean an emphasis in the specific *types* of mid-range shots, i.e. shots at the rim vs. long two’s. Since layups and shots near the rim are much easier and consequently have a much higher success rate than long mid-range jump shots, shooting a larger number of these will most likely lead to a higher 2-pt efficiency. Thus, the optimal offensive strategy for a team in the modern NBA would be one revolved around 3-pointers and shots at the rim, while eliminating the long mid-range jump shots all together. This way, a team can have both the positives of 3-pt volume and mid-range efficiency, while eliminating the negative of 2-pt volume, as examined in the data. A pioneer of this shift towards analytics was the influence of Stephen Curry, who boasted a mix of incredible 3-pt efficiency and volume, revolutionizing the Warriors’ offense, and bringing significant team success along with it. Another prime example of a team in recent history that represents this philosophy to the extreme are the 2019-2020 Houston Rockets. Completely shifting towards the idea of “small ball”, they eliminated mid-range shots from their offense almost entirely, and almost solely launched from behind the arc, consequently possessing a staggering offensive rating of 117.8. So it seems that in conclusion, the shift in offensive strategy from the isolation-heavy, ball-dominant strategies of the 2000’s and beyond, to the 3-pt heavy, spacing-emphasized strategy of modern analytics has indeed led to more tangible offensive success, provided a team is still shooting efficiently at the rim as well. This is also supported by the data, as offensive rating league wide has increased from 103.0 in 2000 to 111.9 in 2020, indicating the offensive success resulting from the change in emphasis on offense strategy. Another important thing to keep in mind is that a high offensive rating does not necessarily correlate with *team* success, as that is ultimately also determined by the team’s defensive performance as well. Sometimes, 3-point volume can also lead to worse defensive performance as well, as long misses can lead to long rebounds for the opposition, leading to more fastbreak chances for opponents, leading to easier baskets. This was particularly an issue for the 2020 Portland Trail Blazers, as despite boasting an offensive rating of 113.7, good for 3rd in the league, they struggled with a defensive rating of 114.8, good for 28th in the league, leading to an overall negative net rating of -1.1. In conclusion, based on the data, the optimal offensive strategy is one centered around high 3-pt volume, high 2-pt efficiency, and low 2-pt volume, predicated on an offense emphasizing volume from 3 and at the rim.

*Sources: *__https://www.basketball-reference.com/leagues/NBA_2020.html__*, *__https://www.basketball-reference.com/leagues/NBA_stats_per_game.html__*, *__https://www.basketball-reference.com/about/ratings.html__

## Comments