Does Defense Win Championships? - An Analysis of the Relative Importance of Defense vs. Offense in the NBA

1. Introduction and Context:

“Defense wins championships” is a popular saying that has been echoed for decades by high school coaches, NBA analysts, and basketball fans. Ever since Michael Jordan famously repeated the phrase himself after lifting the 1991 NBA trophy, it has been treated like gospel by some in the basketball world. After 1991, Michael Jordan and the Chicago Bulls proceeded to win five more championships within the next seven years, seemingly further validating his claim.

Following Jordan’s retirement, the league promptly banned hand-checking on defense, allowed zone coverages, and instated a three-second violation rule on both sides of the court. In the late 2010s and early 2020s, the league then saw a surge in three-point shooting that drove NBA offenses to unprecedented heights. All of this begs the question, how much truth is there now in the saying that ‘defense wins championships’?

While the importance of defense is undeniable, the idea that defense matters more than offense for an NBA title is not beyond doubt. In this project, we will examine data of the past championship teams and playoff teams to dissect each half of the game’s impact on winning a championship. Our analysis will aim to provide insights into both the explanatory and predictive power of defenses and offenses on their respective chances of winning a title. Finally, it will help us broadly answer the questions: can offense-skewed teams win titles; does defense matter more than offense; and finally, does defense win championships?

2. Data and Methodology

2.1 Dataset

To evaluate the relative importance of offense and defense, we utilized team performance data from the 2000–01 season through the 2024–25 season. As zone defense and the three-second violation rule changes were made in the 2001-02 season, and hand-checking was banned in the 2003-04 season, we decided not to include older seasons past 2000. Those older seasons saw a structurally different style of basketball, and including data from that era may lower the overall strength and cohesiveness of the dataset. 

2.2 Key Metrics

The primary metrics used for this analysis are offensive rating (ORtg) and defensive rating (DRtg). Offensive rating measures how many points a team scored per 100 possessions, while defensive rating measures how many points each team allowed per 100 possessions. We chose to mainly use regular season data since they are based on a bigger sample size of 82 games. Relying on regular season data also provides conceptual consistency with our framework, in which playoff performance is treated as the outcome to be explained by defense vs. offense.

While both stats are great, the NBA’s scoring environment has changed drastically over the last 25 years. For example, the league average ORtg in 2004-05 was 106.1, while the average ORtg in 2024-25 was 114.5. Because of this, raw ratings are insufficient for historical comparison. To correct for this, we calculated relative ratings, which measure each team’s ratings compared to the league average in each season. A positive relative ORtg indicates an offense better than the league average, while a negative relative DRtg indicates a defense better than the league average.       

We utilized graphs to visualize metrics of past champions, an Ordinary Least Squares (OLS) regression to determine the correlation between these relative ratings and playoff wins, and kernel density estimation to visualize the efficiency drop-off between the regular season and the playoffs. 

3. Empirical Results

3.1 The Profile of a Champion (2001–2025)

We first gathered the regular season relative DRtg and relative ORtg data of each NBA champion over the last quarter-century to see if they collectively skewed heavily towards defense as the saying suggests. Figure 1 plots the relative ORtg vs. relative DRtg for champions.

Figure 1

The initial visual evidence suggests that the phrase "defense wins championships" is, at best, a simplification. The visual shows no pattern of defensive-skewness, which would have put more teams to top of the graph and to the left of the y-axis. Instead, all but one team are in the top right quadrant with both above average defenses and offenses. There were no champions that were below league average on defense. There was only one team that was able to win a title with a below average offense, which was the historical anomaly 2004 Detroit Pistons.

One notable pattern here connects defense with time. We observe that all four of the most offensively-skewed champions came within the last 12 years, while five of the seven most defensively skewed teams played in the 2000s. 

Overall, this distribution implies that while elite defense is common among champions, so is elite offense. There is little evidence supporting that defense is the sole driver of a title. Most champions needed to be significantly better than the league average on offense as well. However, there could be a trend in earlier teams winning titles with massive defensive tilts, and more recent teams shifting to a more offensive oriented approach.

3.2 Historical Trend:

Given the NBA’s three-point revolution and evolving coaching philosophies, and the pattern observed in the earlier graph, we examine whether these shifts have influenced the profile of championship teams. Figure 2 shows each champion’s “defensive skewness,” calculated as the absolute value of relative DRtg minus relative ORtg, where higher positive values indicate stronger defense than offense relative to league average. 

Figure 2

The graph reveals a distinct structural break in how championships are won across time.

The Defensive Era (2003–2012): The early part of the millennium is dominated by high positive peaks. The 2004 Detroit Pistons represent the historical apex of this trend (skew > +8), confirming that in the early 2000s, it was possible to win a title almost exclusively through defensive dominance while possessing an ordinary offense. During this span, no championship team had an offense that was better than their defense by more than 1 point, relatively speaking.

The Three-Point Era (2013–2018): When the Lebron James Miami Heat went 5-out in 2013, a new coaching philosophy was in the making. Followed shortly by Steph Curry’s Golden State Warriors dynasty and their historical jumpshooting, the three-point era fully blossomed. Teams went from taking an average of 18.4 threes a game in the 2011-2012 season, to 37.6 in the 2024-2025 season. Almost every champion in this era was better on offense than on defense. 

The Current Era (2019-Present): In 2019, the dominance of offensive-driven teams ended. The graph shows almost no pattern favoring either side of the game. Teams like the 2019 Raptors and 2020 Lakers cemented their legacy with defense, while the 2024 Celtics won with an all-time best regular season offense. 

Figure 2 confirms that the "defense wins championships" phrase might have had some statistical evidence two decades ago. Today, however, the same phrase may not be so concrete. Recent champions have found success by focusing on both sides of the court. 

3.3 Regression Analysis: Quantifying the Impact

To win a championship, a team must first win three rounds of playoffs in their respective conferences. In this section, we aim to quantify the impact of defense and offense on winning playoff games in the current era. We use data of playoff teams in the last five years, and run a regression of playoff wins on each team’s regular season relative ORtg and DRtg.

Figure 3

Interpretation of Regression:

The results provide a direct challenge to the traditional wisdom. Holding defense constant, a 1-point increase in relative ORtg is associated with an increase of 0.94 playoff wins. Conversely, holding offense constant, a 1-point improvement (decrease) in relative DRtg is associated with an increase of 0.84 playoff wins. Both variables are statistically significant (p-value < 0.001), confirming that both are critical predictors of playoff success. 

Following these results, we also ran a robustness check on the previous regression with two more control variables: injury and seeding. A team with a key player injury in the playoffs will underperform their regular season metrics. Injuries are hard to quantify and define, since it is almost inevitable that a team will have at least one key rotational player out for some time in the playoffs. We added an injury dummy variable that is equal to one for playoff teams that had an all-star miss a game, to capture the effect of star players getting injured. We added seeding variables that ranged from 1 to 8. While seeding is mechanically related to regular-season performance and therefore potentially endogenous, we include it as a robustness check to account for home-court advantage and playoff path difficulty.

Figure 4

The results show that even when injuries, opponent strength, and home-court advantage are controlled for, relative ORtg still has a larger and more significant effect on winning playoff games than relative DRtg. While the difference in their respective magnitudes is small, we can confidently say that in the modern NBA, having a great defense does not lead to winning playoff games significantly more than offense. 

Note that the R-squared in the original regression sits at 0.3, meaning that about 30% of variation in playoff wins amongst teams are explained by relative regular season defense and offensive ratings. This is to a small but decent chunk of the data, which is to be expected. There are countless other unpredictable factors that impact the outcome of playoff games

3.4 The Playoff Grind: Efficiency Drop-off

If offense has a slightly higher correlation with wins, why does the "defense wins championships" myth persist even in today’s NBA? The answer may lie in how the game changes from the regular season to the playoffs, when the games are more intense and officials allow more physicality on defense. We compared the change in efficiency ratings when teams transition to the postseason in the past 5 years.

Figure 5

Figure 5 reveals an inherent asymmetry in the transition from the regular season to the playoffs. We observed that offensive ratings dropped by an average of 3.96 points, whereas defensive ratings worsened by only 1.23 points. This indicates that offense is elastic and highly sensitive to the increased intensity, physical defense, and game-planning of the playoffs. On the other hand, defense is "inelastic" and transfers to the playoffs more reliably. 

This volatility partially explains why the saying persists, but may also provide an interesting explanation of our regression results from the previous section. The playoff environment naturally suppresses scoring. A team entering the postseason with only an average offense may see their efficiency dip below the threshold required to win. Therefore, high regular season offensive efficiency acts as a buffer. Higher offensive efficiency in the regular season has a stronger correlation with postseason wins because the teams need it to survive the postseason.

4. Conclusion

This project set out to empirically test the validity of the basketball phrase, "defense wins championships." By analyzing twenty-five years of NBA data, we have found that the reality is far more complex than the phrase suggests.

Our visual analysis of past champions clashes directly against the phrase. All but one champion in the last 25 years resides in the top-right quadrant of performance, possessing both elite offense and defense. Teams have been able to win with incredible offenses and defenses. Although the 2000s saw mainly defensive juggernauts win the title, this pattern flipped in the 2010s, and recent champions have won by dominating either side of the floor. This answers our first question; it is shown that offensive-skewed teams can definitely win championships.

Furthermore, we found that in the current NBA, regular season offensive efficiency is a slightly stronger predictor of playoff wins than defensive efficiency, even when controlling for injuries and seeding. Statistically, a potent offense is at least just as valuable as a lockdown defense for winning playoff games. This answers the second question: offense is just as valuable as defense.

However, the "Playoff Grind" analysis offers the critical context needed to reconcile these findings with the persistence of the old saying. We discovered that offensive efficiency suffers a drastic drop-off (nearly 4 points per 100 possessions) in the postseason, while defensive performance remains relatively stable. One possible explanation linking this finding to the regression results is that, because of this offensive decline, teams with higher regular-season offensive ratings are better able to withstand it than teams with weaker offenses. 

Ultimately, does defense win championships? The answer is no, not by itself. All but one champion in the past 25 years were elite on both offense and defense, and recent champions have been able to win by dominating either side of the floor. In the current NBA, it is essential for a team to be elite on both offense and defense.

Bibliography (MLA)

“Can you also create a graph showing the change_ortg and change_drtg for the 80 team sample size csv. As well, give me the code for it.” prompt. ChatGPT, OpenAI, 9 Dec. 2025, https://chat.openai.com/chat.

“Create a trend in relative rating as well, with the past 20 champions.” prompt. ChatGPT, OpenAI, 9 Dec. 2025, https://chat.openai.com/chat.

“Run a linear regression of playoff wins on relative ortg, relative drtg, injury dummy, and seed dummy for the 80 sample size in the past 5 years.” prompt. ChatGPT, OpenAI, 9 Dec. 2025, https://chat.openai.com/chat.

“Run a linear regression of playoff wins on relative ortg and relative drtg, for the 80 sample size in the past 5 years.” prompt. ChatGPT, OpenAI, 9 Dec. 2025, https://chat.openai.com/chat.

“Using the attached csv files, create a scatter plot of relative ortg vs relative drtg for the past 20 champions.” prompt. ChatGPT, OpenAI, 9 Dec. 2025, https://chat.openai.com/chat.

Basketball-Reference. Basketball-Reference.com, 2025, https://www.basketball-reference.com/teams/BOS/2025.html.