These were the words of Grantland’s Zach Lowe, the man whose writing represents basketball’s best combination of analytics with the league beat. But what does Lowe mean by that exactly? And where can we see information relevant for the struggling Cleveland Cavaliers, owners of the NBA’s worst defense since Luol Deng’s debut last month?
Looking into those questions requires background information on the rise of basketball analytics and a look into what we actually do know about defensive analytics in the game today.
Overview of basketball analytics
In a sense, math is evident on offense in basketball just like on-base percentage is obvious in baseball. Those are the team statistics. They make most sense as a starting point. You can easily and directly control the final outcomes of an offensive possession. It’s perhaps understandable then that defensive analytics are just heating up of late.
For years, basketball minds have been aware of Dean Oliver’s Four Factors. The four essential statistics are regarded as the building blocks for basketball efficiency. The stats are Effective Field Goal Percentage, Free Throw Rate, Turnover Rate and Offensive Rebounding Rate.
Effective Field Goal Percentage (eFG): (Two-pointers made + Three-pointers made * 1.5) / Field goal attempts. This measures overall shooting efficiency by assigning appropriate additional weight to three-pointers. NBA average is .495.
Free Throw Rate (FTR): Free throw attempts / Field goal attempts. Shooting efficiency on field goal attempts is not the only way to score, however. This is the most common measure of free throw attempts. On defense, this is especially important considering free throw shooting percentage should be fairly constant. NBA average is about .284.
Turnover Rate (TOR): Team turnovers / Team possessions. This stat is fairly obvious, although a bit tricky to compute as a fan because possessions are often estimated. Just like baseball outs or football turnovers, basketball turnovers are crucial since teams always look to optimize offensive scoring opportunities. NBA average is about 15.3%.
Offensive Rebounding Rate (ORR): Team Offensive Rebounds / (Team Offensive Rebounds + Opponent Defensive Rebounds). Then, finally, Oliver’s final factor is this measure of a team’s percentage of offensive rebounds it corrals. This also leads to extra offensive scoring opportunities. NBA average is 25.6%.
As it relates to defensive analytics, these Four Factors are still being used regularly. They are key footprints for team and opponent data. Do all have equal weight in terms of importance? Certainly not and research is constantly ongoing to see various relationships with these statistics.
Continuing forward with other advances, Kirk Goldsberry, a visiting scholar at Harvard and a Grantland contributor, shared this chart last year within his introduction to a new offensive statistic called ShotScore. The chart hits home the value of offensive shooting attempts very succinctly: mid-range shots carry the least amount of value; restricted-area and three-point attempts are far more advantageous.
One can look no further than the evolution of Daryl Morey’s Houston Rockets offensive footprint to see the changing nature of NBA to an awareness of those numbers. Just 20 years ago, three-point shots were infrequent and not-as-revered. Today, they are the X-factor of offensive success. Especially long-distance attempts from the corners, where efficiency is even higher on average.
On the topic of basketball defensive analytics, Goldsberry’s research project for last year’s Sloan Sports Analytics Conference was titled “The Dwight Effect.” Goldsberry’s signature spatial analysis quantitatively showed how different players add value through defending and even preventing interior shot attempts. This is an obvious trait from defensive superstars like Dwight Howard and Roy Hibbert. The paper popularized the LARRY SANDERS meme within the NBA Twitter world last season.
Within the last year, we’ve also had the mass publication of SportVU player tracking data from Stats LLC. With special cameras now installed in every NBA arena, there is a litany of data never before considered. All fans can access this information in the supersized www.nba.com/stats website.
Some of the SportVU data is relatively meaningless fluff – such as miles per hour on the court – while others have long-term staying power. Those substantial stats include categories like interior defensive impact and defined shot attempts like pull-ups, catch-and-shoots and drives.
This year at Sloan, Goldsberry will be sharing his latest research on Expected Possession Value. The ambitious research project, empowered by the SportVU data, aims to quantify every single action step of a team’s offensive possession. This can then be broken down by player to share their value-add to the game and the team. Certainly, this could later be expanded to show defensive impact as well.
Zach Lowe is a great overall writer because he combines his beat writing background with the latest trends in NBA analytics. His article this week described “The Delicate Balance of an NBA Defense.” Lowe shares that analytics have made their impact on offense, but we’re still in the learning stages of what it means defensively.
In order to showcase some of the leading statistics, Lowe shares anecdotes from specific teams and their defensive strategies. He did not, however, focus on the Cavs. That’s unfortunate for our understanding of the Cavs defense but perhaps fortunate for shielding our eyes. Here were his comments on the teams he analyzed. In his reports, he used the Four Factors statistics:
Minnesota: Allow lowest opponent Free Throw Rate in the game. But head coach Rick Adelman appears to be furious in asking for more foul calls. Still good at forcing opponent turnovers. Terrible in interior defense.
San Antonio: “The most consistently foul-averse team of the last decade.”
Portland: Reconstructed defense for this season. Now play extremely conservatively leading to historically low opponent Turnover Rate. Also have low opponent Free Throw Rate.
Phoenix: Very high foul rate defensively with their aggressive style. They also lead NBA in fast-break points.
A question to ponder: Are opponent Turnover Rate and Free Throw Rate correlated? Could they leave the clues for a team’s defensive footprint? Lowe focused on the increasingly popular conservative defensive style in the NBA, as headlined by Minnesota and Portland. He shares that personnel can certainly still dictate the potential for a team’s strategy.
“Coaching philosophy and roster makeup still determine the look and feel of each individual team defense, and we’re still in the early stages of learning which sort of system is the ‘best’ — or whether such a judgment is even possible, given each team’s unique personnel.”
Minnesota’s personnel (specifically Ricky Rubio and Corey Brewer) enable them to have a higher opponent turnover rate, despite their conservative style of play that usually leads to low turnovers and low fouls. Then, inside, Nikola Pekovic and Kevin Love are regarded as very poor interior defenders. That’s perhaps why Adelman is itching for more foul calls down low to avoid easy baskets.
In Portland, Terry Stotts likely just has less talented defensive players. The team’s offense, however, is quite good this season behind the hot shooting of LaMarcus Aldridge, Damian Lillard and Wesley Matthews. With a stronger bench this season, Stotts’ players are perhaps slightly more likely to foul when needed, but the team’s defense is just not very good.
Is a conservative defensive style a bad thing? Lowe remarked that it’s nearly essential for conservative defensive teams to have an elite offense (like Portland) in order to still be competitive overall. And in order to still be competitive defensively, a team should have an elite rim protector (unlike Minnesota) that can contest those easy-to-ascertain inside shots.
Long-term trends that Lowe notes: Turnover Rate is on a “general long-term decline” and Free Throw Rate “has dropped more severely.” Perhaps the drop in free throws attempted is related to the Morey-ball evolution-based increase of three-pointers attempted, since such shots rarely draw fouls. It’s also a possible correlation-causation research project due to the rise of the conservative defensive style.
You’ve now read Zach Lowe’s reports from the on-court NBA battlefield. Conservative defensive styles are in vogue, leading perhaps to low opponent Turnover Rates and low opponent Free Throw Rates. How does the entire NBA stand up? Here’s a chart with overall team defensive efficiency and Four Factors data.
Conservative defensive teams likely have low numbers in Free Throw Rate and Turnover Rate. Teams with top 10 lowest opponent Free Throw Rates and bottom 10 lowest opponent Turnover Rates: San Antonio, Charlotte, Houston, Orlando, Portland and oh, Cleveland. The Los Angeles Lakers also are quite close. The Timberwolves do indeed force a large number of opponent turnovers.
On the opposite end, teams with aggressive defensive styles likely would have higher numbers in both statistics. Teams with bottom 10 highest opponent Free Throw Rate and top 10 highest opponent Turnover Rate: Brooklyn, New York and Dallas. Phoenix, Toronto and Miami also are right on the fringe, perhaps unsurprisingly so.
Do any of the statistics necessarily correlate that much with overall defensive efficiency? The correlation values are the following: .817 (Effective Field Goal Percentage), .469 (Free Throw Rate), -.142 (Turnover Rate) and .587 (Offensive Rebound Rate).
It’s quite obvious how opponent shooting efficiency can have the largest direct impact. From there? Fairly muted, with Offensive Rebound Rate and Free Throw Rate having some positive relationships. Actually, Turnover Rate is somewhat negatively correlated, counterintuitively but perhaps as a result of the strong conservative defenses in the NBA.
“Almost everyone I’ve talked to in the league believes fouling too much, above some unknown threshold probably a bit higher than league average, is a universally bad thing.”
Are Free Throw Rate and Turnover Rate correlated together as well? The two statistics have a very mild positive correlation level of 0.27. That indicates there is some slight mathematical truth of lower free-throw rates occurring with lower turnover rates. Is it correlation or causation? Further data mining could lead to the answer. At the least, we know the correlation value isn’t that strong.
To continue looking into NBA defensive data, I’ve expanded the shooting data into the usual shooting zones provided by www.nba.com/stats. The chart below looks at the frequency of shot attempts – with restricted area and three-point shots being undesirable for a defense – and efficiency of those shots.
Year-by-year NBA shooting zone averages can be viewed in this chart from Wednesday’s edition of The Diff. Most notably, over the past few years, three-point attempts are up and mid-range attempts are down. This is part of the evolution of the NBA.
Minnesota is indeed 30th in opponent restricted area shooting efficiency. Portland, which stays close to home to encourage more mid-range shots, allows the lowest attempts from both corner threes and above-the-break threes. Indiana quite obviously is best in limiting restricted area attempts and defending them.
Some of the most notable correlation values with overall defensive efficiency: -.665 (Mid-Range %/FGA), .639 (Restricted Area eFG), .523 (Corner Threes %/FGA), .493 (Restricted Area %/FGA). So yes, one can see how the shooting zone ratio breakdown can be an additionally effective way to look at NBA defenses, just like the Four Factors.
So this is all a lot of background on NBA defensive analytics and the statistics from the 2013-14 season. In Cleveland, however, fans are simply curious about how the team has dropped off so significantly in the past month.
In Byron Scott’s three years in Cleveland, the Cavs ranked 29th (109.1), 26th (106.0) and 27th (106.9), respectively, in defensive efficiency. Those were several years of absent defensive schemes with inadequate defensive talent. The struggle was obvious.
With the return of defensive-minded Mike Brown, however, the team got off to a hot start defensively this year. Through 35 games as of Jan. 7, 2014, the Cavs were 13th (102.5) in defensive efficiency. Since Luol Deng’s debut on Jan. 10, they are 30th (113.3) by a large margin. What has happened since?
In Lowe’s article, he noted that Brown has incorporated the principles of the conservative foul-averse defensive style from his background in San Antonio. In another article this week on the upheaval with the Cavs, Lowe also wrote that Brown has an “aggressive blitzing scheme” defensively. This is way different than Portland’s current approach, which is perhaps empowered by statistical weapon Ben Falk.
From a statistical perspective, opponent Free Throw Rate is up and opponent Turnover Rate is down for the Cavs since Deng’s debut. Neither change is that dramatic, however. The biggest difference is in the shooting data.
One can immediately see the overall opponent eFG to see how the Cavs have not been good defensively in the post-LeBron James years. The last 15 games have been as bad as ever. What has been worse? Well, the efficiency everywhere, pretty much, and by large margins. That obviously is an issue everywhere. But perhaps even more concerning is this trend: A gigantic rise in three-point shot attempts allowed.
In Scott’s years, the Cavs were merely average in three-point shot attempts permitted. They were never very good in defending them (or really, any other shot attempt), but teams never sought out that opportunity and/or the Cavs didn’t allow it with that much alarming regularity. This season? Mike Brown’s defensive-minded style has led Cleveland to allowing the NBA’s most regular three-point shot attempts. It’s been substantially worse with Luol Deng.
“Studies have shown that just limiting the number of corner 3-point attempts correlates more strongly with overall defensive ratings than a lot of traditional measures”
Earlier, I introduced Kirk Goldsberry’s 2013 Sloan research paper called “The Dwight Effect.” The paper defined an exact skill of not only defending but also preventing interior shot attempts, as exemplified by the star centers in the NBA like Dwight Howard and Roy Hibbert. Orlando’s Stan Van Gundy relied upon Howard to anchor his conservative defensive style. San Antonio relies upon Tim Duncan.
Cleveland obviously does not have an interior rim protector. That has been a commonly criticized issue of now-gone GM Chris Grant’s tenure. But the data shows that inside shot attempts against the Cavs are dropping steadily. The Cavs actually have allowed the fourth-fewest interior shot attempts.
Instead, there should be a “Dwight Effect” relationship with three-point shot attempts. The best teams at prohibiting long-distance attempts are Portland, San Antonio, Indiana and Chicago. The latter three are the NBA’s top defenses. Portland? They have an adapted conservative style whereas their lesser defensive players work strategically to encourage mid-range shots, not threes.
The issue has not been Luol Deng directly. In fact, the Cavs defense has an even more horrendous 116.6 defensive rating with him off the court the last month. The real issue appears to be the Cavs strategy (or lack thereof) with regards to opponent threes. Two weeks ago, Kirk’s Film Room shared these thoughts as well.
Mike Brown’s blitzing scheme likely succeeded in the past because of LeBron James’ extreme versatility on the wing. Now, it appears, Browns is failing to adapt to the changing NBA landscape in developing a way for the Cavs to prevent a very efficient shot from even occurring in the first place.
The Cleveland Cavaliers are 17-33 with 32 games left in this already terrible season. The defense is getting worse and worse, and their last two wins are against the NBA’s worst team (Milwaukee) and because of a miraculous offensive performance (Washington). If Mike Brown cannot solve this glaring defensive issue, it could be appropriate for the two sides to part ways again after this season.
Math can win out on defense in the NBA. Zach Lowe said the Cavs reportedly do have a “beefed up analytics department,” as created by Grant. It’s about time that math made its way into the schemes of the team to cure its obvious defensive woes.
Jacob Rosen is a long-time contributor to WaitingForNextYear. He's also a writer online at SportsAnalyticsBlog and Nylon Calculus . An Akron native, Jacob is a current MBA student at the University of Oregon's Warsaw Sports Marketing Center. You can follow him on Twitter @WFNYJacob or e-mail him at udjrosen(at)gmail(dot)com.