The NBA has been tracking all of its games since 2013, using an optical system via Second Spectrum that captures footage from cameras positioned in the catwalks of each arena. Those cameras transmit data into the league’s proprietary software, which uses algorithms to get intel on every single player in the league.
The end result is a complement to the traditional “eye test” of watching players to figure out who’s contributing during a game. Here’s a look at how the NBA is embracing these new metrics.
Did you ever try this as a kid: you were in your driveway or at the park, counting down the seconds and emulating a buzzer sound as the ball left your hands?
Well, with today’s countless shooting statistics, you can make sure that game-winning shot is a high-percentage one that ends in celebration.
While most fans are familiar with field goal percentage, you can get far more granular. The NBA tracks shots by distance and defender.
Remember when Damian Lillard made that incredible game-winning shot to win a playoff series?
Well, if Paul George looked at the stats for shots between 30 and 37 feet, he’d have seen that Lillard is quite adept at shooting from that distance, making about one out of every three attempts.
Of players that attempted at least 10 shots between 30 and 37 feet last year, Kyrie Irving and Kyle Lowry were the most accurate, each making 50 percent. The player most likely to shoot from that distance? Atlanta’s Trae Young, who attempted 67 such shots.
Among all players that logged at least 1,000 minutes last season, players shooting against Jamal Crawford hit their shots at a 91.7 pct clip. Meanwhile, Derrick Favors had the lowest DFG percentage at 50.1 pct. Even the “best” defender in the league still gets scored on every other possession.
Teams also measure players by the type of shot they prefer. Steph Curry, Eric Gordon, and Klay Thompson had the most catch and shoot three-point attempts of qualified NBA players in 2018-19. Each of those guys averaged at least six such shots per game – no surprise there.
But bigs like Lauri Markkanen, Nikola Mirotic and Brook Lopez are also primarily catch and shoot guys on threes with respectable percentages. Defenses that know those strategies can adjust accordingly by making them drive to the basket instead of camping at the arc.
Measuring Analytics by Percentiles
While the stats available on NBA.com are the “official” metrics for the league, coaches, scouts, and writers use other resources, too.
Basketball Reference is an encyclopedia of player data. You can search through a whole career, from total stats to per-36 minutes averages to advanced metrics.
Basketball Reference also measures specific shots, games and seasons. Let’s say you were curious about who made the first game-winner of the 2018-19 regular season – you could find out. Congrats to Miami Heat’s Kelly Olynyk, whose buzzer-beating layup on October 18 gave the Heat their first win of the season.
Cleaning the Glass, from former NBA executive Ben Falk, takes available stats and makes them more accurate and easier to interpret.
For example, it filters out garbage time and last-second heaves, and offers a breakdown of players based on whether they were in the half court or transition, position, or if they were on the court.
“It’s generating great stats out of play-by-play and assigning percentiles by position to most everything players do,” said Div Bhansali, NBA tweeter and blogger. “Want to know which guards drew shooting fouls at the highest rates, or which wings excelled at nabbing steals without fouling? CTG has you covered.”
The above stats are terrific for tracking individual events. But what if you want to look at a player more holistically?
That’s where all-in-one metrics come into play, which aim to show how a player affects the whole game.
“All-in-ones are better as a guidepost than as a really specific comparison tool,” Bhansali said.
For instance, tools like John Hollinger’s Player Efficiency Rating compiles a player’s actions during a game to give an overall performance score. As Hollinger puts it: “The PER sums up all a player’s positive accomplishments, subtracts the negative accomplishments, and returns a per-minute rating of a player’s performance.”
A player’s plus-minus (+/-) rating shows how many points his team scores while he’s on the court, subtracted by how many the defense scores. Player Impact Plus Minus (PIPM) takes that a step further by taking luck and additional box score value into consideration. And Real Plus Minus (RPM) looks at a player’s average impact based on net point differential per 100 offensive and defensive possessions.
In the future, we may move into areas that look at a player’s well-being. Though wearable tech isn’t currently legal, it provides insights into players’ vitals and how they’re feeling.
“Teams can analyze physiological and mechanical loads,” said Johnny Carver, who’s worked with two NBA teams and author of Ranketology: A New Way of Determining Basketball’s Greatest Player. “Wearable tech can track their heart rate, how fast they’re moving, and how their body reacts to certain movements.”
Where Things Can Improve
A lot can change in 10 years. Think about it: a decade ago, the NBA was just scratching the surface of tracking technology. Now, we can determine all sorts of wild things about players’ shots, dribbles, passes, and movements on the court. The league has come a long way.
But there’s still room for growth, particularly around defense.
“There’s not a reliable defensive advanced metric,” said Carver. “It doesn’t encompass the difference between being a defensive guard versus a big. There’s no in between. It’s very difficult to measure; finding a more reliable defensive metric would interest any team.”
Aside from counting stats like blocks and steals, there are metrics such as defensive rating. Created by statistician Dean Oliver, it looks at how well a player keeps the other team from scoring points. It calculates blocks and steals, as well as measurements like times blown by, deflections, and points scored.
Even in the defensive tracking stats, most of the value comes in looking at offense. It’s hard to say Jamal Crawford is the worst defender in the league simply because players make nearly all of their shots when he’s defending. The stat only considers shots attempted – opponents took 48 shots against Crawford last year, making 44 of them.
Crawford played in 64 games, though. That means, on average, opponents didn’t even attempt a shot on him in one out of every four games. You could argue that’s the mark of a good defender. It simply depends how you look at it.
It’s that lack of cohesive, all-in-one defensive statistics that leave those in the analytics community believing more can be done.
“You can understand that someone like Andre Iguodala has a huge impact, but do we have specific stats that isolate where that value’s coming from?” said Bhansali. “Can we account for anticipation, or communication with teammates? I think these might be some of the next frontiers over the next few years for fans. It’s going to be awesome to learn more in these areas.”