Data Visualization Link:
Showing the correlation of key team statistics during the 2017-18 NBA regular season and regular season wins that were deemed significant by our statistical analysis
NEW ORLEANS, LA – Teams in the National Basketball Association with older, higher salaried, veteran players had the most regular season success during the 2017-18 season according to our recent statistical analysis. This analysis reviewed over 70 variables included in NBA team statistics, from traditional stats like points, rebounds, and assists per game, to shooting percentage stats, advanced metrics that cover efficiency rates, and miscellaneous stats like market population and attendance.
It turns out that free throw attempts, percentage of shots coming from three point range, and percentage of possessions that end in turnovers had very little effect on winning. Age, steals and blocks, three point percentage, and team payroll had a significant effect our analysis found.
The National Basketball Association (NBA) is more popular and profitable than ever. The 2018 Forbes team valuations valued each team worth at least $1 billion, while the league average among the 30 teams was valued at $1.6 million. The New York Knicks led the league at a valuation of $3.6 million. Just four years ago, the Knicks were valued at $1.4 billion, while the league average was $634 million. Those numbers show a 257% increase in value for the Knicks, and 252% increase for the league average.
Statistics recording in the NBA has advanced exponentially over the years. FiveThirtyEight, a statistical analysis website owned by ESPN, published an article in 2014 detailing the story of Dick Pfander, who has been collecting NBA box scores since the league’s inception in 1946.
“It was a hobby for me,” Pfander was quoted saying. “It was a fun thing for me to do. I had always been interested in statistics. I kind of liked doing statistical-type things.”
Basketball Reference, a leading basketball statistical website, credited Pfander in helping the website add box scores from 1946-1986 to its database in 2012.
“I would be remiss in not thanking Dick Pfander, who did the lion’s share of the work for this project. At first it sounded like an urban legend. I mean seriously, someone had made a scan of every box score for every game in NBA history? To our surprise, Dick had in fact done just that. Many thanks to Dick for his outstanding work.”
Dick Pfander is certainly the extreme, but it’s obvious that people really enjoy reading and analyzing NBA statistics, on a game-by-game basis, or as part of a larger overview. From scanning box scores out of the newspaper to the many innovative technologies of today, the NBA has truly become futuristic.
During the 2013-14 season, the NBA added player tracking cameras to all 29 arenas. The SportVU tracking cameras capture extremely specific movements of a player, including miles ran, speed in miles per hour, and the number of dribbles, passes, and touches a player has during the game. The NBA publishes each detail of these games to its website, allowing fans, journalists, and team statisticians the ability to play and manipulate the data as they like. Basketball Reference and ESPN Stats & Info also compile the stats into their own databases as well.
Marc Spears, Senior NBA Writer for ESPN Undefeated, talked more about the importance of having statistical databases, and specifically ESPN Stats & Info.
“Perhaps I’m a little biased, but I love the guys at ESPN Stats & Info,” Spears said. “They do a great job of breaking down stats I need, and were very helpful during the (NBA) Finals. When writing a story, you want to give your story a little extra punch to really make it sound like you know what you’re talking about.”
Traditionally, joining the coaching or basketball operations staff for an NBA team required previous experience playing at the professional level. Now, analysts from companies like PricewaterhouseCoopers and Ernst & Young, have joined NBA teams to bring expertise at analyzing all the data that the NBA offers, and helping organizations find new ways to build a winning model.
Ben Mehic, a contributor to Bullets Forever, an SB Nation blog covering the Washington Wizards, spoke on how the rise of analytics in the sport has helped teams and the league.
“It’s given rise to the nerds. Once upon a time, sports were dominated by freaks of nature – the societal arena was occupied by athletes. Now it’s led by mathematicians and number crunchers. And you know what? That’s a good thing. The nerds can take what we see happening on the court or field and tell us why it is or isn’t working.”
Mehic also noted how the rise in analytics has led to the emergence of the three point shot as the most important shot in basketball.
“Take the long two-point shot in basketball, for instance,” Mehic continued. “It’s a shot that was often taken before analytics discovered how inefficient it actually is. It changed the game forever. Teams are now perimeter focused, building rosters entirely based on 3-point shooting. It’s helped teams and the league evolve at a greater pace – the game, as a whole, has become more fun. Big men – like, freakishly tall 7-foot players who were locked under the basket not too long ago – are chucking shots from behind the 3-point line. The rise of analytics has been a positive for viewership. Without it, I don’t think we get a Stephen Curry.”
In historical terms, the world is studied in terms of BC (Before Christ) and AC (After Christ). In the NBA, there’s BAA (Before Advanced Analytics) and AAA (After Advanced Analytics). Not only did the success of analytics lead to the rampant increase in three point attempts Mehic mentioned, it also led to the decreased importance of a traditional position on the court; the center position.
Spears, who has covered the NBA as a journalist for 18 years, has seen firsthand how analytics have changed the game and the thirty organizations in the league.
“They are a great tool, so NBA teams are looking for any edge they can get to take their teams to the next level, or make their team more prepared, and that’s why over the last decade analytics has been growingly important. It’s just another way to make a team or player get better prepared for a game. But if you have great talent, and great preparation, analytics can be a part of. I know Chris Paul is a big analytics video guy. You’re basically giving yourself a big chance of success. Success that is combined with your overall abilities.”
Vander Pruna, a NBA Data Scout in Statistics & Analytics for Sportradar US, regularly attends Los Angeles Lakers games on behalf of the company, and records “extensive, live in-game data and statistics, tracks ball movement, creates shot charts, and records offensive/defensive tendencies and full game statistics.” Pruna gave his input on the rise of analytics as well.
“Analytics has become a great equalizer for all teams,” Pruna remarked. “Rise of big data was naturally the next evolutionary step for smaller market teams who have historically struggled to financially keep up with larger market teams. Larger market teams enjoy larger budgets and are more economically appealing to free agents. Because they have had more resources at their disposal, larger market teams have historically attracted more top-shelf talent much more frequently than smaller market teams have. With analytics, however, teams were able to squeeze more from each possession and thus earn more wins. Big data allows each team to increase the efficiency of each possession by as little as tenths or even hundredths of percentage points, and thus narrow the talent gap without necessarily spending that much more. This is good for the league. Analytics raise the level of competition between each ball club. Fans get to watch a better product on the floor. Analytics is the reason the NBA has become a faster, 3-point shooting, and more versatile league.”
Becca Winkert, also known on Twitter as BeccaMVP, is the cohost of a podcast covering the Washington Wizards, and spoke on the evolution of the league as well.
“Analytics in sports has created an evolution in team success. You can now isolate each player and look at their strengths and problem points in greater detail. By tracking every player’s move, teams can now use them to their full potentials. On the other side, analytics also make it clear what a player needs to improve.”
Not every aspect of the rise in analytics has been positive, however.
“There’s been good and bad to it,” Spears added. “Because in a lot of ways it’s had some counter effect on former players and people of color. ‘Oh you’re not strong in analytics so we can’t use you.’ So I think it has had some negative effect, because no matter what stat you have or be able to conjure up somebody that has spent time playing. I’ve heard stories of someone in stats talking down to someone who played in the NBA for decades, because they’re just so confident in the analytics.”
So what data are these statisticians looking at? While the league is a business and teams want to make money, the number one goal is always to win as many games a possible, and in turn win a championship.
“I really enjoy examining true shooting percentages because I know one of my blind spots in analysis is getting caught up in how a player scores, rather than whether or not it’s effective,” said Jake Whitacre, editor of Bullets Forever. “Looking at the true shooting numbers helps me cut through the noise and figure out if how they’re scoring is actually beneficial, or potentially holding the team back.”
Winkert, disclosed some of her favorite analytical measurements as well.
“It’s very important to analyze defensive efforts. By tracking each individual player’s defensive efforts and what they are good at, it will help the team stop their opponent. If a player is best at contesting three point shots, that’s what they should be implementing on defense. I also value the sleep tracking greatly. Sleep tracking can allow teams to enhance a player’s recovery and health. A good sleep schedule results in better performance on the court.”
There are many different categories of statistics. Traditional stats measure a player’s points, rebounds, assists, steals, blocks, and turnovers amongst other numbers. Shooting statistics measure the number of shots a player makes and attempts, that percentage, and also detailed percentages based on where the player is on the court (two vs three pointer, inside of five feet, etc.).
Advanced stats measure player efficiency and other stats on a per minute basis, in an effort to keep comparisons between players effective. Still, each situation a player is in on a given night is different, making it hard to compare advanced stats between players.
Ben Alamar, the Director of Sports Analytics at ESPN, is known across the industry for his expertise in sports analytics, and specifically the NBA, where he spent seven seasons consulting with teams on “how to build analytics into their decision-making.” Alamar also published “Sports Analytics: A Guide for Coaches, Managers, and Other Decision Makers” in 2013, and is considered to have “founded the first journal dedicated to sports statistics, the Journal of Quantitative Analysis in Sports.”
“There’s a new set of metrics that I think are really important, and they basically separate out your shooting ability into your ability to take quality shots, and make shots,” Alamar said. “So there’s a shot making and a shot taking metric. Separating shooting into those two categories really gives you a sense of who this player is and how well they fit into what you want to do. And it also lets you know where they need to grow as a player, and what’s possible in terms of their growth. If a player is taking terrible shots that they’re making, like DeRozan, he’s still got huge upside, he just needs to train to take slightly better shots. Whereas if someone is taking great shots, but is sort of an average shooter, well that’s probably as much as you’re going to get out of him.”
Outside of on court statistics, teams also analyze things like age, sleep as Winkert mentioned, and nutrition. More than ever there are players on vegan diets and those sleeping in hyperbaric chambers. It’s no wonder that players are playing longer careers in recent years as medicine and analytics have converged. This has led to a league struggle, as teams have begun “resting” players during the regular season, essentially choosing to not play a star player because the analytics say he needs more rest, or that the next game is more important to the team’s long term success.
Just this season, LeBron James, at age 33 and in his 15th season, played all 82 games during the regular season, and another 22 games in the playoffs, leading his team to the NBA Finals. Cleveland had a disappointing 4th seed finish during the regular season, winning 50 games, but with LeBron, their eye was always on the postseason performance.
Players like LeBron make it hard to analyze how a team can improve its number of wins during the regular season, because the ultimate goal is not necessarily to win every game in a long 82 game season. If a team is down 20 in the third quarter, it might make more sense to rest the starters for the last 15 minutes of the game so that they are better suited to compete in the next game. Some teams will claw back at that deficit. Some will move on to the next one.
LeBron, and other All-Stars also hamper analytics measures because of how greatly the NBA has historically been built around superstar players. As Spears put it, “If you don’t have any talent you’re not going to win, I don’t care how great the coaching is, how great the scouting reports are, talent cures a lot of ills. Talent is certainly the most important thing to winning.”
Talent truly is king in the NBA. With the best players playing during 75% of the game, and playing both offense and defense, they can make huge impacts in the right situation. LeBron is the literal “King” at showcasing how much talent wins out, after taking the Cleveland Cavaliers to the NBA Finals this postseason, despite a rather underwhelming supporting cast. The fact that this was LeBron’s eighth straight appearance in the NBA Finals doesn’t hurt either.
Winkert had a few additional points to add to the discussion about what factors go into making up a team’s regular season record.
“I think there are several factors that can affect a team’s record,” Winkert said. “There are several components of the schedule that can predict wins/losses. There are things such as long road trips, time zones, and certain competition that can factor into the record. For example, if an east coast team is on a week road trip out to play highly competitive teams, that can have a negative impact on the record. On the other hand, if any given month has a lot of home games with less challenging teams on the schedule… it can produce a positive outcome.”
When analyzing statistics that relate to wins, a key note is that almost all stats related to scoring have a positive relationship with wins. The team that scores more points than its opponent wins, thus unless a team is historically poor on defense, offensive stats relating to points, assists, field goals made, and field goal percentage will usually be indicators of team success.
One of the statistics I was most interested in looking at was age, because of the approximately twenty year age difference between the oldest and youngest players in the league, and how experience seems to usually trump physical talent. Across the thirty teams, the mean age of the fifteen players on the roster was 26.
Rookies usually join the NBA between ages 19 and 23, with the youngest players going to worse teams because they are better prospects, and thus drafted higher (draft order is somewhat inversely related to the number of wins a team had the year before.) The data from the 2017-18 season saw a significant positive correlation between the average age of the players on the team, and the number of wins.
The jump from college to professional basketball is steep, and even the best young players usually require a minimum of three to four years before they are able to lead a team to sustained success across 82 games.
While the use of analytics in 2018 has become advanced, basketball analytics five years ago were pretty simple. The key finding amongst league executives? Three points are more than two points. Thus, if a team could shoot a comparable rate on three point shots as mid range shots (in between the three point line and FT line), the three point shot wins out every time. For example, if a team shoots 50% on 30 two pointers, that’s 30 points (15×2). If a team shoots 33% on 30 three pointers, that’s 30 points. Considering the league average on three point shooting is 36% it makes sense to limit two point attempts that are not layups and dunks (high percentage) and shoot more threes.
Looking at the data, there were only three teams that had over 38% of their field goal attempts as three pointers. The Brooklyn Nets (41.1%), Dallas Mavericks (38.2), and Houston Rockets (50.2%). The Nets and Mavericks were amongst the worst ten teams in the league in wins on the season, while the Rockets won a league high 65 games.
Sometimes the willingness to attempt three pointers outweighs the team’s ability to make them, but in this case all three teams shot within 1% of that league average, but Houston still saw more success in part because they attempted 42 threes per game, compared the Brooklyn’s 36 and Dallas’ 32.
Houston, who ended up with the most regular season wins, is routinely regarded as being at the forefront of the analytics movement. GM Daryl Morey has led the Rockets since 2007, and was a cofounder of the annual MIT Sloan Sports Analytics Conference, the leading conference within the sports analytics industry.
Spears noted that the evolution of analytics across the NBA has in fact been league wide.
“I think they all are. I don’t know that anyone is (at the forefront) in the league,” Spears said. “Every team is always looking for the newest, latest. The one thing that is kind of interesting is I’ll meet guys who have different analytical programs that they’re trying to sell to teams. It’s amazing what they can do visually with video, with graphics, with information that you can get as the game is going on… (It’s) changing by the day, but I know now there’s no team that sleeps on it. There’s just too much money, too much at stake, too much invested in it.”
Alamar noted that although every team uses analytics, not each team uses them to their full capabilities.
“On the larger part of where analytics is headed, I recently did a story on draft analytics and how we project players in the draft into the league and which teams are really good at it and which teams the analytics even has an impact on their work. What was really interesting to me, I was not too surprised by who the people I talked to thought the best were, because it’s Houston, it’s Boston, it’s San Antonio. But, I was very surprised that the people I talked to, most of them said ‘Maybe 30-50% of the league use analytics in the draft.’ Which is shocking to me. Even though we see and hear talk about analytics all the time, there is still room for huge growth there.”
The last key statistic to cover while looking back at the 2017-18 season is the effect payroll had on wins this past season. The relationship was significant at below the .001 level, and had one of the highest recorded correlations at .635. It certainly makes sense. As with most jobs, the more expensive the person’s salary, usually the better he/she is at the job.
Additionally, the NBA has a tax penalty, that charges teams who exceed the tax level of salaries on the roster. Teams with that high of a payroll, are usually willing to go over luxury cap only if they have a team they believe can compete in the league, and win more games.
The relationship between age and payroll is pretty strong as well, since older players can receive larger contracts as part of the NBA’s Collective Bargaining Agreement. Thus teams with a lot of young talent routinely have smaller payrolls, and less experience, leading to less wins.
Another statistic that was interesting to look at was man games missed. Mangamesmissed.com created a graph at the end of the season showing how many games combined players on the team missed to injuries. Though having a healthy team is important to team success, there are so many factors that go into how injuries affect a team, that the correlation being found not significant was not surprising.
The teams widely regarded as the two best, the Houston Rockets and the Golden State Warriors, had players miss 220 and 152 games respectively. Because of the amount of star players on the rosters (six across both teams), each could afford more leeway when a star player like James Harden, Chris Paul, or Steph Curry got hurt.
Meanwhile, three of the most five injured teams were the Phoenix Suns, Dallas Mavericks, and Memphis Grizzlies, who turned their injury filled seasons into three of the worst records in the league.
As great as the options are across the various statistics, there are still plenty of areas where statistics haven’t been able to bridge the gap. Alamar gave his insight on some of the measures the NBA can’t track yet.
“To me one of the great things that you can try to measure with the data now that we haven’t been able to, teammate fit, and how well teammates actually play together,” explained Alamar. “Do they make each other better and how? The other area is basketball intelligence. I think there are ways to measure basketball intelligence, with spatial data now that exists, and we haven’t gotten there yet. We haven’t really done a good job of trying to understand how well players make decisions on the court. So those are two areas where I think there is significant growth possible, that hasn’t happened yet.”
Mehic added on with his thoughts on where statistical analysis can’t get the job done just yet.
“Chemistry is hard to measure. Teams have put together solid rosters on paper that didn’t pan out. The Oklahoma City Thunder are an example. Russell Westbrook is an MVP talent, Paul George is one of the best two-way players in the league and Carmelo Anthony is built to play the modern four position. They’re all popular players – because, well, they’re really good. They got bounced out of the first round by the Utah Jazz who, on paper, looked like a lottery-bound team. The team just didn’t click – they looked like they hated playing with each other. Anthony declined – and he declined quickly. George played uninspired. And Westbrook looked like he was playing more for stats than victories. No one really expected it. Their personalities clashed on the court. You can’t quite measure that yet, which makes basketball the most exciting game we have.” – Ben Mehic
Pruna agreed with Mehic’s statement that chemistry is important to the game, but not yet attainable via analytics.
“It can never replace the psychological profiling of a player,” Pruna said. “Numbers can only explain so much but understanding a player’s mood or the team’s chemistry, for example, can never be fully quantified. Analytics can figure a player’s hustle statistics, such as how likely he is to dive for a loose ball, but it can never fully capture what goes through a player’s head. This still puts a premium on coaches to connect with and understand their players.”
Hustle stats really are interesting, because they give the impression that a player has great heart and hustle, but it really more showcases good instincts of a player, through recording his deflections, loose balls recovered, and charges drawn.
“Analytics can’t account for heart,” Spears remarked. “Analytics can’t (account) for confidence. Can’t account for injuries. It’s not a foolproof thing.”
With so many areas of opportunity the future of analytics is very bright. Experts are flocking to the industry to revolutionize it and get in on all the available money in the league.
This analysis from the 2017-18 regular season, helped to breakdown the many statistics available to sports statisticians today, and it will be very interesting to see how the NBA innovates in the future.
“I think we’re going to see even more focus on how to analyze draft prospects in the future,” said Whitacre. “Generally speaking, we’re at the point where most teams have a good analytical perspective of the 30-team league and what to value within it, so teams are more open about speaking to what the data says. Interestingly, we don’t hear nearly as much about how teams try to decode the draft analytically, even though it’s the best way for contending teams to stay a step ahead of the field.”
With the amount of money in the league, and the pressure building amongst top level teams to unseat the Warriors, teams will do anything they can to have an upperhand. For some teams that will mean tampering (what’s a $500,000 fine to a team that lands LeBron James…) For other teams it means trying to change the league like the Warriors did four years ago.
The NBA landscape goes in waves, and it’s almost time for the next innovative change. A team that can discover a new way to predict team chemistry before trading for or signing a player, will certainly have an advantage in building a roster.
And if you are someone who finds enjoyment in statistical analysis, just as Dick Pfander all those years ago, be sure to make some connections, do your own analysis, and who knows, you may be the next person responsible for causing a seismic shift in the fastest growing professional sports league in the world.
About the Data:
Data for this project was pulled from leading basketball statistics websites, including BasketballReference.com, NBA.com, ESPN.com, and ManGamesLost.com. Additional data on city population amount came from US and Canadian census data.
The majority of the data was available for direct download into excel, but the data on Man Games Missed needed to be estimated from a graph, as direct data was not available for free.
Map of NBA teams’ locations and wins. The red line separates the Western Conference from the Eastern Conference.
Showing the correlation of key team statistics during the 2017-18 NBA regular season and regular season wins that were deemed significant by our statistical analysis
Cover Photo from www.casino.org/news/nba-playoffs-finally-have-early-round-excitement by Yahoo! Sports/Getty
About the Sources
Follow Marc Spears, @MarcJSpearsESPN
Follow Ben Alamar, @BenAlamarESPN
Follow BeccaMVP, @BeccaMVP
Follow Ben Mehic, @BenMehicNBA
Vander Pruna, @VanderPruna
Special thanks to Aaron Dodson, @aardodson for arranging the interviews with the ESPN talent.
Alamar, B. (2015, February 19). The Inside Man: NBA Analytics. Retrieved July 2, 2018, from http://www.espn.com/nba/story/_/id/12338383/nba-analytics-man-ben-alamar
Box Score For Every Game in NBA History. (n.d.). Retrieved July 1, 2018, from https://www.sports-reference.com/blog/2012/01/box-score-for-every-game-in-nba-history/
C. (2016, March 07). Meet the Man Who Preserved Decades of NBA History. Retrieved June 30, 2018, from https://fivethirtyeight.com/features/meet-the-man-who-preserved-decades-of-nba-history/