By: Anthony Zhao
Disclaimer
The following claims are based on observations we have made with data and other sources, and are not indicative of real relationships between NBA Players and Sean “Diddy” Combs.
An Overview of Sean Combs
Sean John Combs is an American rapper and producer, commonly known by his stage name, Diddy. Through music and his other ventures, Diddy accumulated a net worth of well over $500 million. Notably, he used this money to fund his infamous White Parties, colloquially known as “Freak-offs” or “Diddy Parties”. Many celebrities attended these lavish functions, and it has been reported that Justin Bieber, Mariah Carey, Paris Hilton, Jennifer Lopez, Jay-Z, and Beyoncé frequented Diddy Parties.
These celebrations were well-known even among NBA players. In 2020, Lebron James, sporting a lovely black bucket hat, phoned Diddy on an Instagram livestream. James, in his deep, charismatic voice, remarked, “Ay, everybody know, ain’t no party like a Diddy Party!”
In light of the allegations against Diddy, a clip of this livestream resurfaced on social media. Youtuber Rebound Rewind posted a video associating the two using NBA stats. He revealed that with Diddy in attendance, James averages 32.7 Points, 8.0 Rebounds, 7.7 Assists, 1.5 Steals, and 1.3 Blocks per game across six games—a notable increase from his career regular season averages.
Game Type | Points | Rebounds | Assists | Steals | Blocks |
Diddy | 32.7 | 8.0 | 7.7 | 1.5 | 1.3 |
Career | 27.1 | 7.5 | 7.4 | 1.5 | 0.7 |
With seemingly such a drastic difference between Diddy-games and games without Diddy in attendance, it begged the question: was this pattern in data exclusive to Lebron James or were other players in cahoots with the Puff Daddy?
Diddy’s Favorites
Using images found on X (f.k.a. Twitter), Alamy, WireImage, and GettyImages, we found Diddy at 33 Regular Season and Playoff games. Unsurprisingly, Combs was courtside mostly at games in Miami, New York, and Los Angeles as he owned property in those regions. He also attended home games of the New Jersey/Brooklyn Nets, LA Clippers, Houston Rockets, and Golden State Warriors. Intriguingly, although the Celtics were Diddy’s fourth most watched team, Diddy never attended a game in Boston. All but one of the Boston games were playoff games—2 in the Finals against the Lakers (2008, 2010), 2 against the Heat (2011, 2023), 1 against the Knicks, and 1 against the Nets.
Diddy’s appearances spanned three decades and overlapped with 362 players. Dwyane Wade, Kobe Bryant, Lamar Odom, and Pau Gasol had the most appearances at 8 games each. Lebron James and Derek Fisher both appeared in 7 games. Note that this statistic is different from Rebound Rewind’s claim that Diddy attended 6 of Lebron James’ games.
Investigation Individuals
Let’s further examine these top 20 players. For each player, we took each Diddy game and compared it to the player’s season averages, taking season type (Regular and Playoff) into account. To demonstrate this process, let’s analyze LeBron James’ stats.
The all-time leading scorer played in three playoff games (2011, 2017, 2023) and four regular season games (2008-09, 2010-11, 2011-12, 2012-13) with Diddy in attendance. So, we would compare Lebron’s playoff game on April 24, 2023 to his 2023 playoff stats and his December 6, 2012 game to his 2012-2013 regular season averages. Here’s what it would look like visually for James’ points, rebounds, and assists:
At a glance, the plots show that there isn’t an obvious trend across these seven games. Lebron’s most notable Diddy games are his 51 point, 9 rebound, 11 assist performance against the Knicks on February 4, 2009 and his OT playoff win against the Grizzlies on April 24, 2023, where he dropped 22 points, 20 rebounds, and 7 assists. These games are likely the reason why Lebron’s averages appear inflated in Rebound Rewind’s videos.
Looking at the big picture, we took the differences between the games and the season averages across all games, averaged them, and did this for each of the top twenty players. A positive average differential denotes that the player produced more with Diddy in attendance, while negative means they were less productive.
For points, Dwyane Wade’s playoff and Rajon Rondo’s regular season positive differentials stand out. However, both players’ differentials span across one game, so the observed spike isn’t really indicative of a repeated trend of better performance. Similarly, Chris Bosh’s abysmal playoff scoring differential is from a single game, so we can’t really say that Chris Bosh would always panic in the presence of Diddy during the playoffs.
On the rebounding side of things, LeBron and Pau Gasol are of note. Lebron’s big jump in playoff rebounds can be partially attributed to his aforementioned 20-20 game against Memphis. On the other hand, Pau Gasol consistently had great rebounding games across his 2 regular season and 6 playoff games. In an elimination game against the Spurs in the 2008 Playoffs, Gasol grabbed 9 offensive and 10 defensive rebounds, complementing Kobe Bryant’s 39 point effort. For all 8 games, Gasol grabbed 10 or more boards.
In the assists plot, Derek Fisher and Pau Gasol’s differentials are notable. While Derek Fisher’s spike is due to the fact that he only played one regular season game, the boost in Gasol’s play-making abilities seem to be slightly more nefarious. However, in the 2010 Finals against the Celtics, Gasol dropped a near triple-double with 17 points, 13 rebounds, and 9 assists in a blowout win, amplifying his assist differential quite a bit. In his seven other Diddy games, he averaged 4.7 assists and had a median of 4 assists and a high of 7 assists. For reference, Gasol averaged 3.2 assists in both the playoffs and the regular season across his career.
The Diddlers
Filtering for positive differentials for all Points, Rebounds, and Assists (PRA) and for games played over 1, only regular season Dwyane Wade and playoff Paul Pierce remain as players positively impacted by the Diddler. Here are their stats:
Dwyane Wade (Regular Season) | Paul Pierce (Playoffs) | |
Games Played | 7 | 5 |
Points Differential | 2.53 | 2.90 |
Rebounds Differential | 0.08 | 0.48 |
Assists Differential | 0.11 | 1.24 |
Blocks Differential | -0.02 | -0.13 |
Steals Differential | 0.34 | -0.27 |
Turnovers Differential | 1.19 | 0.68 |
Personal Fouls Differential | -0.53 | 0.12 |
Wade's and Pierce’s increase in PRA with Diddy also corresponded in a noteworthy increase in turnovers. Perhaps, the basketball wasn’t the only ball they were handling loosely that night. A more logical explanation for the increase in turnovers is that the players experienced an increase in offensive load in those games, and as such they were able achieve a higher PRA, sacrificing turnovers in the process. In other words, there exists a correlation between PRA and TO unrelated to Puffy.
However, this isn’t really the case. Looking at Dwyane Wade’s regular season and Paul Pierce’s playoff career statistics, there is only a slight increase in TOVs as PRA increased. Running a linear regression model on Paul Pierce’s statistics reveals a measly R-squared value of 0.172. Wade’s R-squared is even lower at 0.046, reinforcing the idea that PRA and TOV are unrelated.
To find further connections between these two players and Diddy, we looked at news articles and tweets relating the two. Paul Pierce posted two tweets about Diddy, one in clear support of Diddy, and the other in favor of Diddy’s musical adversary, Suge Knight, who had spoken out against Diddy’s behavior in the past. It is unclear whether or not Paul Pierce and Diddy are connected to each other using these sources, though his 2013 quote is suggestive as he does label Diddy as “my boy”.
Dwyane Wade’s connections with Diddy are much deeper. In the video, “Dwyane Wade & Gabrielle Union QUIET After Diddy Drama | Jaguar Wright WARNED Us”, Youtuber Culture Spill claims that not only did Dwyane Wade and his wife Gabrielle Union attend Diddy Parties, they hosted parties similar to the freak offs. Culture Spill’s claims are not without reason as the video shows a clip of Diddy and a child reading Dwyane Wade and Gabrielle Union’s names off a guest list in 2022. Other celebrities in the clip included known attendees like Usher and Justin Bieber.
Quoting the words of Dwyane Wade’s former business partner Richard von Hautman, Culture Spill described that, several times a week, Wade and his entourage would host parties, leaving bottles of alcohol, food, and other questionable objects lying around. As incriminating as this sounds, Hautman was sued for libel by Dwyane Wade in 2009, so his testimony may not be entirely credible.
The Diddled
As for players who consistently did worse with Diddy around, Derek Fisher, Luke Walton, Kevin Garnett, and Rajon Rondo all consistently did worse in the playoffs. Kobe Bryant also performed worse in his two regular season Diddy games.
Derek Fisher | Luke Walton | Kevin Garnett | Rajon Rondo | Kobe Bryant | |
Games Played | 6 | 4 | 4 | 4 | 2 |
Points Differential | -0.29 | -2.46 | -4.99 | -5.31 | -0.95 |
Rebounds Differential | -0.41 | -1.05 | -0.87 | -0.44 | -0.03 |
Assists Differential | -0.03 | -0.98 | -1.23 | -3.08 | -1.88 |
Blocks Differential | -0.08 | 0.09 | 0.00 | 0.09 | 0.03 |
Steals Differential | 0.47 | -0.19 | -0.30 | 0.22 | 1.25 |
Turnovers Differential | 0.38 | -0.78 | -0.44 | 0.11 | -2.03 |
Personal Fouls Differential | -0.30 | 0.65 | 0.78 | -0.60 | -0.37 |
There are a few things to note from this data. Luke Walton was a bench player and saw his minutes drop significantly in the postseason. In his four games, he played a bit more than four minutes less with Diddy in attendance; percentage-wise, this is, on average, 28% less than his playing time in those years during the playoffs. Kevin Garnett’s and Rajon Rondo’s significant negative points and assists differentials may partially be attributed to the fact that all four of games were away games in a hostile playoff environment. Another explanation may be that Paul Pierce’s elevated performances in those games allowed Garnett and Rondo to play a lesser role, though this is likely not the case because the Celtics lost all four Diddy playoff games and needed KG and Rondo to play their normal roles.
On the internet, there is not much known about the relationships between Derek Fisher and Luke Walton with Diddy. Kevin Garnett briefly mentions Diddy once in a podcast as he recounts meeting Biggie and Diddy at an All Star game in 1997. Not much is said about Diddy in KG’s anecdote.
On the other hand, there are videos linking Rajon Rondo and Kobe Bryant with Diddy. In 2019, a retired Kobe was spotted next to Kanye West at a gathering hosted by Diddy. Combs posted an image of Kobe Bryant at this party on his instagram page, confirming his attendance. A video titled “Rajon Rondo at Harrah’s Atlantic City Pool for P Diddy’s 4th of July party” was posted on Youtube on July 5, 2010 and features Rondo walking in a hotel. The person who posted the video alleges that Rondo was leaving Diddy’s party at the hotel, Harrah’s Resort Atlantic City. Diddy could not be observed in the video, as he was shut out of his own party. Interestingly, this event happened less than three weeks after Diddy attended Rondo’s game on June 15, 2010 against the Lakers in the Finals.
Diddy Simulator
In the grand scheme of things, we can view these games as an arbitrary set of games that the players happened to perform better or worse at. Perhaps, in each of the games that Diddy went to, Kevin Garnett happened to be in a shooting slump or Savannah James happened to have promised Lebron extra Madden time if he put up extra boards. To address this, we can take the following approach.
Let’s assume that we don’t know which games Diddy went to. Removing this label, the preceding games look like regular season and playoff games randomly chosen from a set of seasons. Using this idea, we can run Monte Carlo simulations and randomly sample games from the players’ seasons. Simulating provides us a distribution of sets of games that we can use to gauge how different the set of Diddy games is.
How it works
Here’s a brief overview of how the simulation works.
To create a comparable set of games, we randomly selected one game from each season that Diddy attended (Diddy games were not excluded from the set of games). This sampling would be conducted 1,000 times, allowing for repeated games in sampling.
After obtaining our large population of sample games, we compared the set of Diddy games to the distribution of randomly sampled games using the same differential calculation as before. Using PCA dimension reduction, we plotted the sets of games to visually inspect whether or not Diddy’s set of games appeared significantly different.
Finally, we calculated the distance (Mahalanobis Distance) of the simulated and Diddy differentials to the average set of simulated games. If the set of Diddy games was significantly different from the average set of games, then its distance would be high. For example, if the average differential from the simulations looked like this:
Points | Rebounds | Assists | Blocks | Steals | TOV | Fouls |
0.1 | -0.1 | 0.0 | 0.0 | 0.1 | -0.1 |
And if the player’s Diddy differential looked like the following table, then the distance would be pretty high. To test for a significant difference, we calculated the p-values using the chi-squared test.
Points | Rebounds | Assists | Blocks | Steals | TOV | Fouls |
5.1 | 4.1 | 2.2 | 0.8 | 1 | 0 | 0.5 |
Considering what we have seen already, we performed these simulations on Dwyane Wade, Paul Pierce, Kevin Garnett, Rajon Rondo, and Pau Gasol. Lebron James was also tested in the simulator.
Results
This is how Dwyane Wade’s Diddy games look in comparison to the simulated games.
Wade’s Diddy games appear on the outskirts of the distribution, though not enough to be considered an outlier. Numerically, we can conclude the same thing with an insignificant p-value of 0.38. Comparing the statistics individually, only Dwyane Wade’s TOV differential stands out when compared to the simulation. Out of the 1000 samples, Wade’s TOV differential was higher than 954 of them.
What about the three C’s on the list: Paul Pierce, Kevin Garnett, and Rajon Rondo?
The Truth’s p-value was even higher than Wade’s at 0.61. The only notable stat was his assist differential which was higher than 95.7% of the simulated sets of games.
Now, Rajon Rondo’s simulation results were more compelling. Although we can’t say that his set of Diddy games is significantly different from the sets generated by simulation (p=0.23), there may be something fishy going on with Rondo’s points and assists differentials.
Across 134 playoff games in his career, Rondo averaged 12.5 points and 8.5 assists per game. From our initial analysis of Rondo’s differentials, we found that on average Rondo scored 5.31 points and 3.08 assists less than his averages for the season. This is a 42% decrease in points and 36% decrease in assists from his career averages. Compared to the average points and assists from the simulation, Rondo’s Diddy points and assist differentials are worse than 984 and 963 sets of simulated games respectively. What perhaps is “saving” Rondo from being an outlier is his relatively ordinary blocks, steals, and turnovers differentials.
Similarly, Kevin Garnett’s set of Diddy games is not significant (p = 0.195). However, KG’s point, assist, and personal foul differentials with Diddy attending are quite outstanding. The Big Ticket’s points differential was in the bottom 1.4% of simulated sets, his assists differential in the bottom 6.7%, and his personal foul differential greater than 92.8% of simulated averages.
We saved the best for last: Lakers Legends Pau Gasol and Bronny James Sr. Recall for the previous players we looked at either their regular season games or their playoff games, not both. For Gasol and James, we analyzed both types of games separately and collectively.
Using a significance value of 0.05 (α=0.05), we found that Pau Gasol’s set of Diddy games was significant. His two regular season games clocked in at p = 0.02, his six regular season games at p = 0.0007, and his eight games combined at p = 0.012. Broadly, what this means is that, since the set of Diddy games deviates significantly from the simulation mean set, we can view the set as an outlier. In everyday vernacular, these results suggest that it is likely that Pau Gasol was getting freaky with Diddy after games.
The following plot displays the combined games. The Diddy games don’t appear significantly different from the rest of the simulated games; however, that is an artifact of PCA dimension reduction and that these two dimensions only capture 38.4% of the variance.
In contrast, Lebron’s games combined and in the regular season were not significantly different (p_combined=0.11, p_regular=0.22), though his three playoff games were significantly different than average simulated playoff set (p=0.04). We found this to be quite fascinating, since Lebron’s regular season points (5.98), assists (1.31), and turnover (-1.21) differential were higher than his playoff differentials (pts = 1.33, ast = -0.06, tov = -0.75). Out of the seven statistics, only Lebron’s playoff rebound differential (3.88) was greater relative to his regular season differential (-0.50). This signifies that the variance in Lebron’s performance in the regular season was much greater than his variance in the playoffs.
What about Drizzy?
In the spirit of creating comparable sets, we can use other celebrities to determine whether or not the Diddy effect is distinct to him or a more widespread effect for all celebrities. Though not as renowned as Comb’s, Aubrey Drake Graham’s parties are no laughing matter. Moreover, Drake is an ambassador for the Toronto Raptor and is regularly seen courtside in Scotiabank Arena. In fact, Drake was seen at Raptor’s game as recently as December 5th, 2024, where Shai Gilgeous-Alexander and the Thunder soundly defeated Scottie Barnes and the Raptors.
From what we could find on GettyImages, Drake attended 118 NBA games. Out of these games, 85 of them involved the Raptors in some capacity. Outside of Toronto, the cities that Drake frequented the most were Los Angeles at a combined 13 games watching the Lakers and Clippers, and 9 games in Miami for the Heat. Although he won over $800,000 betting on the Nuggets in the 2023 Finals, there is no evidence that Drizzy has ever attended a Nuggets game. The Utah Jazz were similarly never blessed to have Drake watch one of their games in person.
As for Drake’s most watched players, former Raptors dominated the top 20. Kyle Lowry played in over half the NBA games Drake went to with 62 games. Shaun Livingston (20 Games) and Toronto’s namesake Lebronto (24 Games) were the two only non-Raptors on the list. Livingston was a Golden State Warrior in 15 of the 20 Drake games and a Brooklyn Net in the remaining 5 games.
Pau Gasol played in one Drake game across his 1362-game career. On November 11, 2009, Gasol dropped 15 points, 8 rebounds, 5 assists, and one block in 28 minutes against the young OKC Thunder. Though less than his season averages, Gasol ended up with a plus-minus of 23 en route to a 16 point victory, indicating that he played exceptionally well in the time he was on the court and didn’t need to play his season average 37 minutes per game.
Objectively, across 24 Drake games, Lebron performed better than his career averages. In fact, for all of the following statistics except assists (which is marginally worse), Lebron’s Drizzy averages were better.
Game Type | Points | Rebounds | Assists | Steals | Blocks |
Diddy | 32.7 | 8.0 | 7.7 | 1.5 | 1.3 |
Drake | 30.0 | 8.8 | 7.3 | 1.7 | 1.2 |
Career | 27.1 | 7.5 | 7.4 | 1.5 | 0.7 |
Taking the season and season type into account, as we did for Diddy, here is how Lebron’s Drake games compare to his season averages:
In 14 playoff games, Lebron appears to have scored significantly more than his season averages. Lebron’s top three games (as determined by Game Score) were all playoff performances: a 43-point, 8-rebound, 14-assist game against the Raptors in the 2018 playoffs; a 41-point, 16-rebound, 7-assist performance in the 2016 Finals (the same game where Kyrie also scored 41 points); and a 37-point, 12-rebound, 4-assist effort to close out the Spurs in Game 7 of the 2013 Finals.
Now, for his stat differentials, we see that the numbers corroborate our visual analysis of Lebron’s games. In the regular season, none of Lebron’s averages stand out, except that he typically commits 0.9 fewer fouls than his season average. In both Diddy and Drake games, Lebron tends to handle the ball more punctiliously, especially in the playoffs.
Differential Type | Drake (Regular) | Diddy (Regular) | Drake (Playoffs) | Diddy (Playoffs) |
Points | 0 | 6.0 | 4.6 | 1.3 |
Rebounds | 0.5 | -0.5 | 0.6 | 3.9 |
Assists | 0.1 | 1.3 | 0 | -0.1 |
Steals | 0.4 | 0.4 | -0.2 | -0.2 |
Blocks | 0.1 | -0.4 | 0.4 | 0.5 |
TOV | -0.1 | -0.7 | -1.3 | -1.2 |
Personal Fouls | -0.9 | -0.4 | -0.2 | 0.4 |
Besides simulation, another approach to testing for significant differences in multivariate data is Hotelling’s T-squared test.
We tried to conduct the Hotelling’s test on Diddy’s data, but 7 games were not enough data to accurately determine a significant difference from the average. For Drake, we found that the set of Drake games (disregarding playoffs/regular season categorization) was significantly different from average with p=0.026.
Comparing the set of Drake and Diddy games to each other with a two-sample Hotelling’s test, we got a T2=0.26 and a p=0.963, indicating no significant difference between the two sets of games. A T2 statistic of 0.26 indicates that the differentials of the two sets of games are quite similar. In Lebron’s eyes, it doesn’t matter whether or not Drake or Diddy attends his games. He’ll perform roughly the same when either are in attendance.
Conclusion
So, who did Diddy do it with? Along with the frequently mentioned Lebron James, we found that Pau Gasol was a highly suspicious individual. Across eight games, Pau Gasol averaged 2.9 rebounds and 2.0 assists more than his season average. Moreover, Pau Gasol was one of Diddy’s four most watched players, giving Gasol ample opportunity to collaborate with Combs throughout his career.
Game Type | Games Played | Points | Rebounds | Assists | Blocks | Steals | TOV | PF |
Career | 1458 | 16.7 | 9.1 | 3.2 | 1.6 | 0.5 | 2.1 | 2.2 |
Diddy | 8 | 17.4 | 13.1 | 5.3 | 1.4 | 1.1 | 2.8 | 2.9 |
Other notable figures investigated were Dwyane Wade, Paul Pierce, Rajon Rondo, and Kevin Garnett. Wade and the Celtics trio’s performances with Diddy in attendance were not significantly different from their season averages. However, we were able to find some alleged dishonorable off-court behavior linking Dwyane Wade to Diddy.
Our most shocking discovery was that Lebron not only plays superb with Diddy in attendance, but his high productivity is quite similar to when Drake is in attendance. This begs the question, what other dirt does Kendrick Lamar have on the Champagne Papi, and does it have to do with Bron?
Next Steps
In this article, we found the NBA games that Sean Diddy Combs attended and investigated several players who played in those games. A massive limitation in our analysis is that there isn’t much data to work with, causing our findings to be quite limited in power. Moreover, we don’t know if the list of games are comprehensive and capture all games that Diddy attended, though we have proof that Diddy did attend every game in the dataset. We hope readers use the data we collected for their own analyses.
One idea we wanted to investigate was how well a player was expected to play in the game that Diddy attended. For example, what if Lebron happened to be playing well during the week that Diddy went to or what if the team Pau Gasol was playing against happened to be injury ridden?
We attempted to create a time-series model to predict how we would expect the players would perform given their previous performances for each Diddy game. Unfortunately, our model performed quite poorly. An alternative to the machine learning approach we took would be to extract an expected value from betting odds and lines set for the players by sportsbooks. Sportsbooks create their lines based on the betting market, indicating how people expect for a player or a team to perform. However, we were unable to find readily available historical data that fit our needs, so we scrapped this idea.
Sources:
_________________
Note that Mahalanobis Distance takes into account the Covariance Matrix (kind of like a multivariate standard deviation) in its formula, so this isn’t entirely true, i.e. there is more to the distance than just the difference of values.
No power calculations/analysis was done to determine whether this is a viable significance value, but ChatGPT agrees with our assessment and says it’s pretty standard.
コメント