Introduction
Today I will be comparing statistics between players in the NBA seasons from 2009-2019. I want to see what statistics are necessary for an MVP candidate, and ultimately figure out what stats a player would need to achieve to get MVP.
Details of the dataset
The data for this analysis is derived from the NBA seasons 2009-2019. The data contains the 22 variables described and consists of 4,847 rows. Each row of the data represents a player’s statistics for a season in the NBA, and contains information like points per game, games played, what season it was, etc. A summary of the data follows.
| Variable | Description |
|---|---|
| player_name | The name of the player |
| team_abbreviation | The abbreviation of the team the player played for in their respective season |
| age | The age of the player during that season |
| player_height | The player’s height (in centimeters) |
| player_weight | The player’s weight (in kilograms) |
| college | The college the player went to |
| country | The country the player is from |
| draft_year | The year the player was drafted |
| draft_round | The round the player was drafted |
| draft_number | The number in which the player was picked in the round |
| gp | Number of games played throughout the season |
| pts | Average number of points scored per game |
| reb | Average number of rebounds per game |
| ast | Average number of assists per game |
| net_rating | Team’s point differential per 100 possessions while the player is on the court |
| oreb_pct | Percentage of available offensive rebounds the player grabbed while he was on the floor |
| dreb_pct | Percentage of available defensive rebounds the player grabbed while he was on the floor |
| usg_pct | Percentage of team plays used by the player while he was on the floor (FGA + Possession Ending FTA + TO) / POSS) |
| ts_pct | Measure of the player’s shooting efficiency that takes into account free throws, 2 and 3 point shots |
| ast_pct | Percentage of teammate field goals the player assisted while he was on the floor |
| season | The NBA season the statistic relate to |
| mvp | Indicates whether the player won MVP. 1 = won MVP, 0 = did not win MVP |
Available data for the NBA Dataset
This searchable table contains all the data available for the NBA players from 2009-2019
To summarize what an MVP looks like, here are the MVP winners for the last 10 seasons
| player_name | team_abbreviation | age | gp | pts | reb | ast | net_rating | usg_pct | ts_pct | season | mvp |
|---|---|---|---|---|---|---|---|---|---|---|---|
| LeBron James | CLE | 25 | 76 | 29.7 | 7.3 | 8.6 | 10.8 | 0.333 | 0.604 | 2009-10 | 1 |
| Derrick Rose | CHI | 22 | 81 | 25.0 | 4.1 | 7.7 | 8.3 | 0.319 | 0.550 | 2010-11 | 1 |
| LeBron James | MIA | 27 | 62 | 27.1 | 7.9 | 6.2 | 10.7 | 0.317 | 0.605 | 2011-12 | 1 |
| LeBron James | MIA | 28 | 76 | 26.8 | 8.0 | 7.3 | 14.1 | 0.298 | 0.640 | 2012-13 | 1 |
| Kevin Durant | OKC | 25 | 81 | 32.0 | 7.4 | 5.5 | 8.0 | 0.327 | 0.635 | 2013-14 | 1 |
| Stephen Curry | GSW | 27 | 80 | 23.8 | 4.3 | 7.7 | 17.0 | 0.283 | 0.638 | 2014-15 | 1 |
| Stephen Curry | GSW | 28 | 79 | 30.1 | 5.4 | 6.7 | 18.3 | 0.320 | 0.669 | 2015-16 | 1 |
| Russell Westbrook | OKC | 28 | 81 | 31.6 | 10.7 | 10.4 | 3.3 | 0.408 | 0.554 | 2016-17 | 1 |
| James Harden | HOU | 28 | 72 | 30.4 | 5.4 | 8.8 | 10.0 | 0.353 | 0.619 | 2017-18 | 1 |
| Giannis Antetokounmpo | MIL | 24 | 72 | 27.7 | 12.5 | 5.9 | 12.5 | 0.314 | 0.644 | 2018-19 | 1 |
The average statistics for MVPs
To get a better idea instead of multiple instances, I decided to average all the statistics to show on average what the MVP numbers put up every year
| Average Age | Average Games Played | Average Points per Game | Average Rebounds per Game | Average Assists per Game | Average Net Rating | Average Usage Percent | Average True Shooting Percentage | MVP |
|---|---|---|---|---|---|---|---|---|
| 26.2 | 76 | 28.42 | 7.3 | 7.48 | 11.3 | 0.3272 | 0.6158 | 1 |
How the main statistics look over the years
It is safe to say that the statistics people mainly look at are points per game, rebounds per game, and assists per game. I thought it would be useful to show how these statistics have varied over the past 10 years.
Points per Game over the years
Rebounds per Game over the years
Assists per Game over the years
It is interesting to see the variability of these stats, but with these graphs, you get an idea of what it takes to be in MVP in the NBA. The most variable is rebounds per game. I would assume because guards don’t get as many rebounds as forwards or centers, so the low rebound per game years were most likely mvps won by guards (like Stephen Curry).
Seeing the public opinion on the Top 3 NBA MVP prospects of 2022
The top 3 prospects of the NBA 2022 MVP race are Nikola Jokic, Joel Embiid, and Giannis Antetokounmpo. I wanted to see what twitter was saying about them:
Here are common words in tweets for the top 3 NBA MVP prospects
All the common words among them make sense. They are all top player in the playoffs right now, so you see some similarities in their common words other than the teams they are playing on and against.
Sentinments for the Top 3 MVP Prospects
I want to use sentiments to see how the public opinion of twitter varies from player to player
Looking at these sentiments, I would say Nikola Jokic is favored the most. It is important to know that Embiid recently suffered an injury, so many of the negative sentiments could come in to play there. It is also worth noting that Embiid had the most tweets about him. Embiid had 1000 observations, while Jokic had 847 and Giannis had 978 respectively. Even though taking in all those factors, in my opinion, Nikola Jokic is the most favored by the public opinion of Twitter.
Prescriptive Analysis
Here, I want to see what are the most important statistics in terms of winning the MVP award as an NBA player. I have a general idea based on the stats and graphs above, but I want to see what a regression model would describe about it.
The formula I will use for this regression is: \[MVP = \alpha_i + age_i + gp_i + pts_i + reb_i + ast_i + netrating_i + usgpct_i + tspct_i \]
##
## Call:
## lm(formula = mvp ~ I(age) + I(gp) + I(pts) + I(reb) + I(ast) +
## I(net_rating) + I(usg_pct) + I(ts_pct), data = nba_mvp_analysis)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.04182 -0.00555 0.00019 0.00359 0.97069
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -6.415e-05 6.463e-03 -0.010 0.992080
## I(age) -1.590e-04 1.538e-04 -1.034 0.301226
## I(gp) -1.189e-04 3.376e-05 -3.523 0.000431 ***
## I(pts) 1.147e-03 2.548e-04 4.503 6.85e-06 ***
## I(reb) -1.991e-04 3.633e-04 -0.548 0.583676
## I(ast) 1.815e-03 5.086e-04 3.569 0.000362 ***
## I(net_rating) 9.646e-05 5.785e-05 1.667 0.095495 .
## I(usg_pct) 3.057e-03 1.790e-02 0.171 0.864414
## I(ts_pct) 3.851e-04 7.366e-03 0.052 0.958311
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.04469 on 4838 degrees of freedom
## Multiple R-squared: 0.03179, Adjusted R-squared: 0.03019
## F-statistic: 19.86 on 8 and 4838 DF, p-value: < 2.2e-16Takeaways
With the results, it shows 3 significant variables: gp (games played), pts (points per game), and ast (assist per game). These all make sense, and here is why: For games played, you need to play most of a season as an NBA player to put together a respectable season to be considered as an MVP. The least amount of games ever played by an NBA MVP is 49, but usually you need to play at least 70 games (and that is on the low end). If you haven’t played many games, you won’t be considered for an NBA MVP. For points per game in the last 10 seasons of this dataset, the least amount of PPG by an NBA player who won MVP was 25. It is probably one of the highest volume and most important statistics in the NBA, shown by it’s INSANELY small p value. For assists per game, this is a statistic that shows you are contributing to your team making baskets. In the last 10 seasons of the dataset, the smallest assists per game was about 6, which is high in terms of NBA standards. To put that in perspective, in the most recent NBA season, there were only 16 players who average over 6 assists per game. Summing up these statistics, to be an NBA MVP, you need to be around a top 10 scorer, while also being a top 15 contributor in the league. Not to mention that you also have to play almost the whole season without getting injured (which is difficult when the NBA season is 82 games long).
Some things about the analysis that I’m surprised about
Age wasn’t as significant as I thought it would be. The range of MVP winners is around 23-29; a pretty small window to go for an MVP. an MVP player has only a couple seasons in their prime to achieve this goal. The P value is lower than some of the other non-significant values, but it’s not significant. I am also surprised that the P value for usage percent is so high, usually every player who wins MVP is heavily used on their team, but I guess if you think about it, there is also probably 29 other players in the NBA who were highely used but did not win MVP.
Some things about the analysis that are about where I expected them to be
Many might be surprised that rebounds isn’t significant when it is one of the main statlines in the NBA. For example if someone said LeBron had 27/7/7, that means he had 27 points, 7 rebounds, and 7 assists. This didn’t turn out significant because the rebounding stat is generally more positional. Forwards or Centers that win the MVP will usually have a high amount of rebounds (10+), while guards, like Stephen Curry, won’t have a large rebound count. No one is really talking about how many rebounds Steph Curry got. Lastly, Net rating having a low p value but not quite being significant doesn’t suprise me either. It is almost significant because you will not be selected for the MVP if you are not on a winning team, but it is a statistic that can highly vary among players, as the best player on the best team doesn’t always win MVP.
For Fun
Since I did sentiment analysis on the top 3 MVP prospects this year, I thought I could try and use the knowledge I got from this assignment to predict this year’s MVP
| Player | G | MP | FG | FGA | FG. | X3P | X3PA | X3P. | X2P | X2PA | X2P. | eFG. | FT | FTA | FT. | ORB | DRB | TRB | AST | STL | BLK | TOV | PF | PTS | X |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Nikola Jokic | 74 | 33.5 | 10.3 | 17.7 | 0.583 | 1.3 | 3.9 | 0.337 | 9.0 | 13.8 | 0.652 | 0.620 | 5.1 | 6.3 | 0.810 | 2.8 | 11.0 | 13.8 | 7.9 | 1.5 | 0.9 | 3.8 | 2.6 | 27.1 | NA |
| Giannis Antetokounmpo | 67 | 32.9 | 10.3 | 18.6 | 0.553 | 1.1 | 3.6 | 0.293 | 9.2 | 15.0 | 0.616 | 0.582 | 8.3 | 11.4 | 0.722 | 2.0 | 9.6 | 11.6 | 5.8 | 1.1 | 1.4 | 3.3 | 3.2 | 29.9 | NA |
| Joel Embiid | 68 | 33.8 | 9.8 | 19.6 | 0.499 | 1.4 | 3.7 | 0.371 | 8.4 | 15.9 | 0.529 | 0.534 | 9.6 | 11.8 | 0.814 | 2.1 | 9.6 | 11.7 | 4.2 | 1.1 | 1.5 | 3.1 | 2.7 | 30.6 | NA |
| Chris Paul | 65 | 32.9 | 5.6 | 11.3 | 0.493 | 1.0 | 3.1 | 0.317 | 4.6 | 8.3 | 0.559 | 0.536 | 2.6 | 3.1 | 0.837 | 0.3 | 4.0 | 4.4 | 10.8 | 1.9 | 0.3 | 2.4 | 2.1 | 14.7 | NA |
| Luka Doncic | 65 | 35.4 | 9.9 | 21.6 | 0.457 | 3.1 | 8.8 | 0.353 | 6.8 | 12.8 | 0.528 | 0.529 | 5.6 | 7.5 | 0.744 | 0.9 | 8.3 | 9.1 | 8.7 | 1.2 | 0.6 | 4.5 | 2.2 | 28.4 | NA |
| James Harden | 65 | 37.2 | 6.3 | 15.3 | 0.410 | 2.3 | 6.9 | 0.330 | 4.0 | 8.4 | 0.476 | 0.485 | 7.2 | 8.2 | 0.877 | 0.8 | 6.8 | 7.7 | 10.3 | 1.3 | 0.6 | 4.4 | 2.4 | 22.0 | NA |
| Devin Booker | 68 | 34.5 | 9.7 | 20.9 | 0.466 | 2.7 | 7.0 | 0.383 | 7.0 | 13.9 | 0.508 | 0.530 | 4.6 | 5.3 | 0.868 | 0.7 | 4.4 | 5.0 | 4.8 | 1.1 | 0.4 | 2.4 | 2.6 | 26.8 | NA |
| Rudy Gobert | 66 | 32.1 | 5.5 | 7.7 | 0.713 | 0.0 | 0.1 | 0.000 | 5.5 | 7.6 | 0.718 | 0.713 | 4.6 | 6.7 | 0.690 | 3.7 | 11.0 | 14.7 | 1.1 | 0.7 | 2.1 | 1.8 | 2.7 | 15.6 | NA |
| Trae Young | 76 | 34.9 | 9.4 | 20.3 | 0.460 | 3.1 | 8.0 | 0.382 | 6.3 | 12.3 | 0.512 | 0.536 | 6.6 | 7.3 | 0.904 | 0.7 | 3.1 | 3.7 | 9.7 | 0.9 | 0.1 | 4.0 | 1.7 | 28.4 | NA |
| Jayson Tatum | 76 | 35.9 | 9.3 | 20.6 | 0.453 | 3.0 | 8.6 | 0.353 | 6.3 | 12.0 | 0.524 | 0.526 | 5.3 | 6.2 | 0.853 | 1.1 | 6.9 | 8.0 | 4.4 | 1.0 | 0.6 | 2.9 | 2.3 | 26.9 | NA |
Here are the top 10 MVP candidates for the most recent season, but really I am just looking at the top 3. I got this data from Basketball Reference. To be specific, here is the exact link: https://www.basketball-reference.com/friv/mvp.html
Looking at the top 3, my pick is Nikola Jokic. He is behind bot Giannis and Embiid in points per game which was the most significant variable from the analysis, but he almost doubles the other 2 candidates in assits per game. He is scoring and contributing to the whole team. I think here it is also fair to take into account rebounds per game, because the top 3 candidates are all bigs. Jokic and Embiid are centers, while Giannis is a power forward. With that taken into account, in my opinion, Nikola Jokic blows the others out of the water. Not to mention he is having one of the, if not the most, efficient seasons of all time (Wasn’t covered as much in this project but thought it was still worth mentioning).
Conclusion
I thought this project was really enjoyable, and I got exactly what I wanted from it. I asked myself what are the key statistics to winning an NBA MVP. The analysis told me that points per game, assists per game, and games played were the most significant; along with honorable mentions net rating and age. I found some surprises and answered my question. I even used what I had learned with this project to pick who I thought was best fit for this years MVP, Nikola Jokic. I also got to take a trip down memory lane and see what past MVP statistics were, and what an average MVP’s statistics may look like.