How the NBA Playoffs Differ From The Regular Season

Regular season and playoff basketball in the NBA are seen as completely different animals. There are no load-management games in the playoffs, there are no possessions off, and there are real stakes for every game. As Bob Myers stated at the 2019 Sloan Analytics Conference, the playoffs and regular season are "two completely different sports." With the 2020-2021 playoffs approaching, I wanted to look at the difference between the playoffs and regular season from an analytical perspective and see if there were significant findings in the numbers. 

For this investigation, I used NBA tracking data from the stats.nba.com API. I scraped team-level data from the past 6 seasons (2014-15 to 2019-2020) for each nba.com tracking statistic in both the regular season and playoffs. Further, I manually added an indicator for each season's NBA Champion.  After looking at the data, it became evident that there is a significant difference in offensive schemes between season segments, as offenses in the playoffs are much more individual and isolation-driven than in the regular season.  Now, let's walk through my analysis and how I was able to understand these differences

To begin the analysis, I ran t-tests on each tracking statistic between the playoffs and regular season, and sorted them by p-value (significance). This helped tell me which statistics were the most different between the playoffs and the regular season. The most significant variable in the entire analysis was Average Seconds Per Player Touch. Further, Assist Points and Average Dribbles Per Touch were also in the top five of significant differences between all of the regular season and postseason metrics. Below, I show the six most significant offensive metrics as a result of the t-tests. 

These are all statistically significant differences and made it clear that there is a significant contrast in offensive tracking metrics between season segments. Visualizing these offensive metrics for each year, helped highlight the differences in offensive style between the regular season and playoffs. The playoff numbers are in red, while the regular season numbers are in green.

Note: on mobile phones, tilt your phone sideways to see the entire graphic.

In these figures, Seconds Per Touch and Dribbles Per Touch are significantly higher in the playoffs than in the regular season. This illustrates a higher share of individual offense in the postseason each year, as individual players keep the ball for longer and take more dribbles to initiate their own offense. Creating your own offense, and having players that are able to do that efficiently is essential to playoff success. 

Looking at Assist Points Created Per Game, it has an upward trend in both the playoffs and regular season, which is a consequence of increased pace of play in today's NBA. More points are scored, so more assist points are created per game. Even with this trend, the playoffs feature fewer Assist Points Created than the regular season each year, indicating a higher share of unassisted points in the playoffs.

Further, we see that Passes Received in the Paint and Passes Received Near the FT Line decrease in the playoffs, and Pull-Up 3-PT attempts increase. While all three statistics  reinforce the year-to-year trends of the evolving NBA and its exodus from two-points shots, the differences between season segments remain each year, and expose the playoffs as a more individual-driven game. There is less passing to different scoring areas of the court, and more individual creation behind the three-point line in the playoffs. One thing that is important to note is that over this time frame the regular season points per game average is ~1.6 points higher than the playoffs and there are ~1.8 more possessions per game in the regular season. Yet, these overarching differences are not significant enough to drive the differences between season segments that we see above.

The t-tests and visualization that I have outlined compared all 30 regular season teams with the 16 playoff teams for each season. I did not account for the possibility that the 14 non-playoff teams could have highly variable play-styles that might help drive differences in the regular season statistics. I wanted to ensure that these 14 non-playoff teams were not driving the significant results between season segments. To remove this potential bias, I looked at the regular season statistics for the 16 playoff teams and their subsequent postseason metrics, to understand if these teams adapted significantly when in the postseason. Thus, I ignored the regular season statistics of the 14 non-playoff teams for the following analysis. The visualization below shows the regular season distribution and playoff distribution of the 16 playoff teams each season for three of the most significant metrics that reflect isolation basketball.

Note: on mobile phones, tilt your phone sideways to see the entire graphic.

Looking at the visualization, it is clear that the same trends are present when focusing only on playoff teams. Assist Points Created decrease in the playoffs, while Seconds and Dribbles Per Touch significantly increase. This suggests that non-playoff teams do not drive these differences in play-style, but rather that playoff teams do, in fact, adjust their play-style in postseason competition. This finding was very significant, but left me wondering how NBA Champions fit into this narrative. Do champions follow these trends of isolation basketball to win titles, or do they employ their own unique strategy for success?

Examining the box-plots above (specifically the blue dots), champions generally follow the other playoff teams' trends from the regular season to the playoffs. Champions remain at similar position in the distribution for each statistic, for both season segments, but their statistical marker follows the same trend as other playoff teams. Although this is a small sample size, this suggests that champions might not adapt differently than other playoff teams. While champions, such as the some of the elite Warriors teams, might be outliers in some of these metrics, they still similarly alter their style in the postseason relative to the other postseason teams.

Due to potential factors that include extended preparation, heightened defensive intensity, and greater individual talent, the data show that playoff teams generally adapt to a more individual-driven offensive scheme. This is a large driving force of why franchises should prioritize elite shot-makers with high isolation efficiency, and why the player archetype of a "creator" has become so valuable in the modern era. Just think about it: who was the last NBA champion that did not have an all-NBA level player that you could just give the ball to and he could go create a bucket for himself? The 2013-2014 Spurs? Maybe? It is an interesting thought experiment, and also why many argue that NBA Championships cannot be won without a "top-5" player which fits this creator archetype. We are just scratching the surface with this level of tracking data, and as we get more and more years of data, it will be interesting to see how these trends progress. Yet, for these playoffs, sit back and enjoy the increased share of isolation offense and individual creation from the NBA's best players.