All the data referenced below comes from basketball-reference.com. The data was gathered and visualizations were made using Python (the pandas and seaborn packages, specifically).
I also referenced Syed Sadat Nazrul’s Medium post on towardsdatascience.com for scraping tabular data from the basketball-reference.com team page for each franchise.
BB1: A Little League History (1947 – 2020)
We need not mention that the style of play in the NBA has changed dramatically over the league’s seventy-three-year history.
In the earliest days (1947 – 1970), the game was stylistically and aesthetically the antithesis of today’s “modern” play. Seven-foot tall ambidextrous bigs were the bell of the ball – George Mikan in the ’50s, Russell vs. Wilt in the ’60s; extending to Lou Alcindor’s (later Kareem Abdul-Jabbar) dominate rookie season. Alcindor was the NBA’s best talent over the next fifteen seasons (1970-1985); overlapping the introduction of the 3-point shot in the 1979-1980 season and the birth of the league’s greatest rivals, Magic Johnson and Larry Bird.
However, since 1985, only four franchises – the 1994-1995 Houston Rockets, the 2000-2002 Los Angeles Lakers, the 2008 Boston Celtics, and the 1999/2000s/2014 San Antonio Spurs – have secured an NBA championship with a seven-footer as the premier talent.
Just eleven championships in thirty-five opportunities, roughly 33%, have been awarded to a team with an explicit inside presence. To put it modestly, the game has changed remarkably. So, if the NBA’s preferred modus operandi for the majority of its lifespan is no longer “the way to win”, what is?
BB2: NBA “Decade Representation” Teams
To attempt to answer this question and discover what caused this revolution, I have selected the most successful team from each decade to serve as an over-simplified representation of their era. In making this comparison we will use play-by-play game data that serves as a decent, albeit incomplete measure of each franchise’s style and success. Given that we’re attempting to discover what impact the 3-point shot has had on success (i.e. winning games), I am forced to exclude teams from the 1950s, 1960s, and the 1970s. The teams I have selected as proxies for the rest of the league history are the Los Angeles Lakers of the 1980s, Chicago Bulls of the 1990s, San Antonio Spurs of the 2000s, and Golden State Warriors of the 2010s.
Let’s first take a look at the total wins for each team for all the seasons in our sample – forty in total. (Note: each season contained eighty-two games except for lockout-shortened seasons in 1998-1999 and 2011-12. For both seasons I’ve extrapolated the Bulls’ and Warriors’ season winning percentage against an 82-game benchmark.)
Wow, that’s a lot of winning. The red line intersecting the plot is equivalent to a .500 team (one that loses as many games as it wins). Just four seasons – the 1999 Bulls and 2010-2012 Warriors, were below that benchmark. Now that we’ve established each of the four franchises as being objectively successful, let’s take a look at how each weaponized the outside shot.
BB3: Winning Teams and Shooting Accuracy
The below figures are interesting in their regard. But, it should be noted that these are incredibly small sample sizes, and these data should not be substituted as proxies for the NBA as a whole in each respective decade.
First, here is the correlation between winning and 3-point shooting for the Magic Johnson Lakers:
Largely because the Lakers had a lower overall volume of attempts (BB4), their winning correlated .56 with their outside shooting. Of the four teams, the Lakers had some of the worst shooting seasons – from 1980-1983 they shot 20%, 18%, 14%, and 10% as a team, respectively.
The Michael Jordan Bulls shot the ball much more consistently and better:
Aside from the 1999 season, the year after Jordan, Scottie Pippen, Dennis Rodman, and Phil Jackson departed Chicago, the Bulls shot well and won. In their 1996 season, they made over 40% of their attempts as a team, which had a strong influence on the correlation in these data.
And here are the “more modern” Spurs and Warriors:
Of course, the Gregg Popovich Spurs are bucking the trend in our sample.
But if you look closer, the Spurs’ plot is an archetypal example of a deceiving visualization portending a dubious reality. Look again at the Lakers’ plot and you’ll notice the x-axis (“3PT %”) ranges from 10% to 38%, while the Spurs’ x-axis ranges from 35% to 41%. Without careful examination, one could conclude that the Spurs’ shooting was more seasonal than the Lakers, or in other words, it had a higher standard deviation. Rather, it was the Spurs who were a model of consistency, both in terms of games won and 3-point attempts made. In this case, the Spurs’ consistency impacted the correlation.
And the Warriors shot the ball well and won a lot of games. At least from the 2012-13 season on.
BB4: Winning Teams & Shot Volume Changes
The correlation between shooting accuracy and winning for the sampled teams is interesting, but it’s also rather obvious. Yes, usually teams that shoot better also win more games, particularly if the made shot is worth an extra point. But, maybe we can find some interesting trends in 3
-point volumes (as a percentage of possessions, which we’ll call 3-point attempt percentage or 3PA%) from our selected decade proxy teams.
Let’s take another look at our four teams. This time, comparing the correlation between 3-point attempt percentage and winning.
For the Lakers, the team-wide 3-point attempt percentage was much lower than the rest of our sample (1% to 8% of total possessions), and their winning happened to coincide with an increase in overall volume relative to the beginning of the sample – the introduction of the 3-point shot. In other words, once Magic Johnson entered his prime and the Lakers really started winning, the Lakers were shooting more 3-pointers simply because at the start of the decade the 3-pointer was a brand new weapon.
The Jordan Bulls’ numbers are worth digging into further. While the team was relatively successful throughout the entire decade, they certainly peaked in the 1995-96 season when they set the then single-season wins record and won the championship. And again in 1996-97, the Bulls won more than 80% of their games and a championship. These are also the two seasons the Bulls’ attempts the largest number of outside attempts in their sample, 1349 and 1403, respectively. But, the number of possessions per game also increased during these seasons and continued to increase into 1997-98 and 1998-99, when Chicago had their worst record in the sample. Indeed, the correlation is hurt by an outlier.
Also, it’s interesting to see the progression of the Bulls’ style of play. In the second “three-peat” 1996-98 seasons, Chicago’s best players were attempting a much larger number of outside shots than they were at the start of the decade. Michael Jordan increased his volume from 93 attempts in 1990-91, the Bulls’ first championship season, to 297 in 1996-97. Scottie Pippen used the outside shot as a weapon more than any other Bull by the end of the decade – taking 424 attempts in 1996-97 compared to 68 total in 1990-91. Jordan’s, and especially Pippen’s, increase in 3-point volume also coincided with their winningest seasons. If we remove the 1998-99 season from the sample, it’s likely we find a fairly significant correlation.
The Tim Duncan Spurs were less reactive to changes in 3-point attempt percentage. The Spurs, along with the NBA as a whole, sequentially shot more threes in each season from 2000 – 2009. San Antonio attempted 882 3-pointers in 19990-00, a figure that was only dwarfed by the Bulls in the 1995-1998 seasons, compared to 1,620 attempts by the 2008-09 season, a ~100% increase in shot volume over the decade. The number of possessions also slightly decreased by 2009; resulting in the Spurs having their highest 3PA% figures in the 2007-08 and 2008-09 seasons, their second and third-lowest win percentage figures in their sample (only above the 1999-00 championship season).
Because the Spurs were the most consistent winner in the group – winning between 64.6 percent and 76.8 percent of their games in these ten seasons – the correlation with both their 3-point accuracy and attempt percentage isn’t as strong. In other words, they won no matter what.
The Steph Curry Warriors also followed a similar trend to the Spurs. Each season they shot more 3-pointers as a percentage of total possessions than the preceding season, with the exception of a slight dip in 2017-18. The majority of their winning came after Steve Kerr was hired as head coach in the 2014-15 season, in which they went on to win between 69 and 81 percent of their games for the remainder of the sample. However, because their winning increased so dramatically from the beginning of the decade, 26 wins compared to 73 wins in 2015-16 (the new single-season record), Golden State’s correlation is stronger. Similar to the Spurs in the 2000s, the Warriors gradually increased their outside shot attempts, along with the rest of the league, as the ’10s progressed.
Mostly, these results show the impact of consistency on the presentation of data. The Spurs never showed a sizeable shift in their winning percentage, so their outside shooting figures were not highly correlated with success (i.e. winning was always high, and 3-point accuracy and volumes climbed steadily). It also shows how the change can be amplified with the base – whether it be 3-point accuracy or 3-point attempt percentage – starts relatively smaller. This is a lot of the story for the Lakers. Even though the Warriors used the outside shot in a manner unlike any other in our sample, the change in volume from 1700 to 2800 attempts (a 65% increase) is not as dramatic as the change in volume for the Lakers – 90 to 450 attempts (a 400% increase).
BB5: Other Interesting Use Case Teams
For additional context, here are figures for some of the other salient exemplars of 3-point shooting in league history. Some are not considered quintessential “winners” today.
First, we have the 1980s Doug Moe Denver Nuggets, one of the NBA’s historically lesser-known examples of the “run-and-gun” philosophy. The strategy, which Mike D’Antoni’s Suns later perfected, involves increasing the number of possessions in a game in the hope that you can outscore your opponent. Denver also had the advantage of having extreme altitude on their side; more possessions means more running up and down the court with less oxygen in the air. Moe was the coach for the entire sample, and the Nuggets qualified for the playoffs in all ten seasons.
Mike D’Antoni’s “seven-seconds-or-less” 2000s Phoenix Suns are rightfully cited as NBA pioneers of the modern game empowering floor spacing and player/ball movement. Like the Warriors sample above, the Suns benefit from winning the most towards the end of the decade, when league-wide attempts were also rising (BB6). Regardless, 2,000+ attempts in both the 2004-05 and 2005-06 seasons showed the Suns were a harbinger of a change to come.
The 2010s Miami Heat followed a different trend; their winning happened before the league-wide spike in attempts around 2015-16 (BB6). When LeBron James, Dwyane Wade, and Chris Bosh were in town, the Heat won between 66 and 80 percent of their games. Towards the end of the decade, the Heat’s attempt volume increased but the organization toiled in mediocrity (in both the 2014-15 and 2018-19 seasons they finished below .500).
The 2010s Milwaukee Bucks data told a similar story to that of the Warriors and Suns. They won their most games in 2018-19 when the NBA’s league-wide 3-point volume had exploded. It’s also worth remembering how much the Bucks won in the 2018-19 season (and were winning in 2019-20, before the season was cut short).
And finally, the 2010s Daryl Morey Houston Rockets. They have become the poster child for the offensive movement built around spacing and high-efficiency shots. The Rockets’ data also benefits from the franchise winning 79% of their games in the 2017-18 season. And in the 2018-19 season, 3-point attempts accounted for 44% of the Rockets’ possessions, as they amassed 3,700 attempts overall.
BB6: Baselining with Rest of the League
Now, let’s put those figures in context. This plot helps us visualize the league-wide 3-point attempt percentage for each season in our sample.
There are a few interesting insights that immediately jump out. The first is the 1995-97 seasons cluster, in which the league average 3-point attempt rate explodes and then drops back to gradual levels for the remainder of the 1990s. The reason for the sharp increase was the shortening of the 3-point line for the 1994-95 season from 23′ 9” (with 22′ corners) to a uniform 22′ all around the arc. The league’s impetus for the shortened shot? An attempt to decrease the decline in average game scores that had become noticeable since the “hay day” of the 1980s. Another interesting insight is that the introduction of the 3-point shot in 1979-80 seemed to have encouraged creativity and curiosity across the league. It would take until 1984-85 for the increase in attempts to match the initial levels experimented with in 1979-80 (around 2.6% of possessions).
It’s also important to distinguish the divide between the seasons prior to 2003-04 and those after when the league changed their stance on what is known as “hand checking”. Following the rule change, players on defense were forbidden from putting their hands on an offensive player unless that player is located near the basket with their back to the basket. Essentially, offenses after 2004 have this advantage over those prior to the rule change.
If you’re rather numerate you might notice that some of our case study teams were leaders in ushering the use of the 3-pointer.
Now, a worthy caveat for the graph above is that only four of the seven teams called out in the plot finished as champions. And the outliers with the largest difference between their 3-point attempt rate and the NBA average rate are not champions. The 2005-06 Phoenix Suns finished 25% of their possessions with a 3-point attempt, a roughly 8% increase over the league of 16.8 percent. The 2015-16 Golden State Warriors had the second-largest delta, a 6% increase over the 24 percent league average.
And, as the league is well aware, winning a majority of the games on the regular-season schedule is hollow if there is no championship hardware to show for it. So how does the champion’s 3-point attempt rate compare with the league average rate in each of the last forty seasons?
Less than half of the champions in the sample attempted 3-pointers at an equal or greater rate than the average team that season. Even more insightful, following the rule changes in 2003-04, just four champions have had a higher 3-point attempt rate than the league average that season.
So, maybe, attempting 3-pointers is not the historical recipe for an NBA championship, even in the era of the 3-point eruption. It’s tough to know using correlation and averages alone, but these data give us encouragement to keep digging.
Photo Credit: Russ Isabella, USA TODAY Sports