The old adage “Defense Wins Championships” is used in all sports, especially when the playoffs are around the corner. Coaches and players around the NBA are no exception, as it is always thrown around in post game interviews and pre game speeches. However, does defense actually win championships? Is defense actually more important than offense when it comes to making a deep run in the playoffs? By looking at the stats, we can see if this adage is true or just a myth.
Method 1: Wins Over Expected
I tried to separate the offensive and defensive contributions to a longer than expected playoff run to first quantify whether defense actually wins championships. This method basically tried to see whether a significant improvement in defense or offense causes a long playoff run, or if their impact is the same. In order to create a baseline for expected playoff wins, I used an exponential regression, using league rank as the input.
The model fits very well, as seen by the curve intersecting the points with accuracy and a high r-squared of 0.92. I created the model by taking the natural log of both the regular season rank and average playoff wins, then creating a linear regression, then transforming again to get the coefficients. (The final formula was 14.5(0.86)^X, where X is the league rank of the team.)
Then, to incorporate offensive and defensive contributions, I used the season and playoff offensive and defensive ratings for each team. Rather than just subtracting to find an improvement, I decided to use z-scores, with the sample of the regular season ratings being only the teams the made the playoffs. This way, the sample remains the same for both the regular season and playoff offensive/defensive ratings. Afterwards, I took the difference between the playoff z score and regular season z-score for the offense and defense, so that a value above 0 would mark an improvement, while a value below 0 would account for a worse performance. Additionally, it accounts for pace and era by using per 100 possession stats and z-scores, respectively.
Lastly, to combine these two methods, I regressed offensive and defensive improvement against wins above expected, which was found by taking playoff wins minus the expected playoff wins. The results for each is shown below.
The results look very similar for both offense and defense. The defensive effect is slightly greater, with a correlation of 0.312, greater than the offensive correlation of 0.301. Since the results were so similar, I explored further to greater see the effects of offense and defense on playoff runs.
Method 2: Separating by Round
The second method I used to see if defense actually wins championships was to explore the effects of playoff and regular season offense/defense, as well as its improvement, for each separate round of the playoffs. Basically, what I did was compare the regular season offensive/defensive z-scores for each team based on how far they advanced in the playoffs. For example, the first table looked like this.
On the y-axis is the average regular season z-score for the rating that corresponds with the title. On the right side, the teams are grouped into five sections, one for each round in the playoffs. Based on the above graph, NBA champions have a slightly better regular season defense than offense, though they are unlikely to be significant with such a low sample size (only 16 championship teams included). The largest difference comes for the teams that made the Finals but lost. Those teams have historically been much better on offense in the regular season. However, if the “defense wins championships” saying applies to getting better on defense, there is a better stat to use than regular season stats.
The two graphs above show the average playoff improvement and playoff stats for defense and offense by round. The graph showing the playoff offensive and defensive z-score do not show many differences between the two, though it actually favors offense. Historically, championship teams have been better offensively than defensively in the playoffs.
However, the graph which gives the most information is the one showing improvement for offense and defense in the playoffs. Focusing on the last two columns, it is seen that the teams that make the finals and lose often improved greatly on defense, while the teams that won the championships actually improved offensively more than defensively. Intuitively, by combing the two sections into one section of teams that made the finals, the numbers would look similar for offense and defense.
Conclusion
Depending on your interpretation of the “defense wins championships” saying, the stats proving it right or wrong can vary. If you think it means a defensive improvement contributes to a longer run in the playoffs, then the stats support you slightly. If your interpretation means that championship teams are better in the regular season defensively, then the stats do support you slightly again, but if you think it means championship teams are better in the playoff defensively, the stats do not a specific advantage. Lastly, if you believe that a significant improvement on defense contributes to a championship, the stats actually disprove you, showing that an offensive improvement contributes to championships. While the “defense wins championships” saying may partially be true, offense is equally or even more important to making a deep playoff run and winning an NBA championship.
Best Ball Stats 