It's getting loud inside Cameron Indoor Stadium.
Thousands of Duke's "Cameron Crazies," alongside hall-of-fame coaches Mike Krzyzewski and Roy Williams, NBA scouts, announcers and reporters, have their eyes glued to the action as the pivotal Duke vs. North Carolina basketball game unfolds in front of them on Feb. 18, 2015.
Unbeknownst to most of the arena, six additional sets of eyes are watching the rivalry's 239th game far more precisely than anyone -- or anything -- before them.
Six tracking cameras, which lurk unobtrusively in the stadium's rafters, are part of a system called SportVU. The cameras capture ball and player movement 25 times per second throughout games and then break down the data into advanced performance analytics and scouting reports within 90 seconds of a play.
The technology of SportVU, a company founded in Israel in 2005, first made a splash when TV networks adopted its yellow first-down line for NFL and college football games. Illinois-based statistical firm STATS LLC acquired SportVU in 2008 and introduced its innovative technology to the basketball world during the 2010-11 NBA season, when four teams (the Dallas Mavericks, Golden State Warriors, Houston Rockets and San Antonio Spurs) installed the system.
Over the next two seasons, the company brought 11 more teams on board and then received the ultimate validation when the NBA mandated that all 29 league arenas be equipped with the cameras for the 2013-14 season. The league also picked up the hefty tab -- the software had cost each team about $100,000 a year for installation and season-long use. SportVU's rise in the professional ranks spilled over to the NCAA in 2013, when Duke became the first college team to install the technology that is anticipated to grow into a must-have for programs nationwide.
STATS provides a data pool that includes obscure statistics like rebounding surface area, field goal percentage based on a defender's distance from the ball and shot arc on both makes and misses. The analytics serve to assist users like Duke with scouting, game preparation and personnel decisions.
"You get a whole lot of basketball information," said Kevin Cullen, Duke basketball's director of information technology. "You can look at yourself in a much more detailed way without wasting hundreds of man hours. All the statistics they provide, from rebound chances to potential assists to tracking shot types, is information that would take many, many hours to log manually."
Cullen, who was introduced to the technology at the 2013 Elite Eight in Indianapolis, added that colleges pay "significantly less" for the system than NBA teams but declined to give an exact cost figure.
Shortly after Duke announced its adoption of the system, Louisville and Marquette's basketball programs both revealed that they, too, would use SportVU during the 2013-14 season. Doug Davenport, Louisville's video coordinator, recalled the process that brought the cameras to the Cardinals' KFC Yum! Center.
"The NBA game has turned so much to it," he said. "Coach [Rick] Pitino had spoken to some guys in the NBA who were hyping it up. We didn't want everyone to get on it and then, all of a sudden, we're behind."
For Marquette, the decision was more straightforward because the Golden Eagles share the Bradley Center with the NBA's Milwaukee Bucks, so SportVU already was installed. The Golden Eagles still pay a fee to use the system, but the NBA already has covered installation costs. Marquette's video coordinator, Jake Presutti, noted that SportVU can be used as a recruiting tool.
"To say that you have the best technology is something," Presutti said. "It's not something that comes up for every prospect, but it shows that your program is cutting-edge, that you're open to new ideas and that you're forward-thinking."
In the eyes of Los Angeles Lakers reporter Mike Trudell, the race for SportVU has begun.
"If Duke has it, then Kentucky's going to want it and North Carolina's going to want it," he said. "Nobody's going to want to be left out of the movement."
Despite Trudell's notion that no team can afford to fall behind, some simply cannot afford to keep up. According to The New York Times, Duke basketball generated $25.7 million in revenue during the 2012-13 season, but small mid-major schools like Santa Clara University inevitably struggle to implement costly cameras.
"We'd absolutely do it if they weren't hundreds of thousands of dollars to install," said Brandon Rosenthal, Santa Clara's director of basketball operations. "It is the future of basketball."
Just as Marquette shares the Bradley Center with the Bucks, the University of Memphis plays in the FedEx Forum, home of the NBA's Memphis Grizzlies, which could provide the Tigers with an attractive opportunity to take advantage of the technology. However, Eric Sebastian, Memphis' director of basketball operations, remains skeptical about using the already-installed SportVU cameras because college teams play no more than 40 games in a season, compared to the NBA's 82.
"We were approached by SportVU last year," Sebastian said, "but we have such a small sample size that, by the time we'd get enough useful information, the season would be over."
The same concern exists even among programs such as Duke's that already have embraced SportVU.
"There certainly are some sample-size issues," Cullen said. "We're only using it at our home games, so if we go play against Virginia, we don't have access to the SportVU data from that game. So you're looking at maybe 15 to 18 games we're able to use the data for."
Ryan Warkins, STATS' director of basketball products, laid out the company's plan to expand SportVU to entire conferences, which he feels will help eliminate the sample-size problem by having more games played in SportVU-equipped arenas.
"It's definitely a valid concern, which is why we'll go at full conferences," Warkins said. "So we would get data parity and depth across entire leagues."
But Duke's Cullen noted that SportVU's benefits are not limited to actual games.
"We're able to run the data during practice," he said. "That data is even more reliable than game data because the quality of the opponent stays constant, and we run it for almost 100 practices per season. That was our answer."
J.A. Adande, who has covered the NBA for ESPN since 2007, sounds a note of caution about relying too much on technology.
"I'm still a bigger believer in going to games, observing, talking to people," Adande said. "I was at a game during the [NBA] Summer League, and I was standing next to a coach. He was pointing out things like player interaction. These are things you can only observe by watching. If I see a guy who doesn't move without the ball and just stands there holding his hands up and slumping his shoulders and shaking his head, I would think that he's not a guy I want on my team. There's no way to figure that out if you're just looking at a database."
Despite its limitations, the system has garnered the approval of statistical experts around the country.
"The only shame is that it wasn't around 10 years ago," said Philip Maymin, co-editor-in-chief of the Journal of Sports Analytics. "It's obviously the way of the future."
Kirk Goldsberry, a Grantland.com staff writer and visiting scholar at Harvard University, echoed the praise for SportVU.
"The cameras allow us to create more categories than conventional statistics," Goldsberry said. "There's no question that they have the potential to help us understand performances a lot better."
But based on his experience at Santa Clara, Rosenthal has concluded that a coaching staff that does nothing more than pore over pages of advanced analytics is unlikely to accomplish much in team meetings. The true value of statistics, in his mind, is when coaches interpret the numbers' significance and then communicate to their players specifically how they can improve.
"If coaches go up to the whiteboard and put up all these different stats, but the players don't understand what they mean, then those stats are useless," Rosenthal said. "It really comes down to how well the coaching staff is able to relate those statistics to the players."
Cullen pointed to an example of how that approach is applied at Duke.
"Maybe we shot 3-for-11 on jump shots off-the-dribble," Cullen said. "Coaches can bring those numbers into a meeting and say, ‘Guys, you're not going to be successful doing that.' It's not rocket science."
And just one telling analytical stat from SportVU can make all the difference for NBA prospects. For example, Louisville graduate Stephan Van Treese grabbed 51 offensive rebounds for the Cardinals at home games, with SportVU cameras watching. Teammate Montrezl Harrell grabbed 54. A look at just those traditional statistics might suggest that the two are equals.
But before he decided to return to school, Harrell was set to be taken in the first round of the NBA Draft -- in part because of his offensive rebounding skills. Van Treese went undrafted and will play in Japan next year.
Why the discrepancy? A look into SportVU's analytics tell us the answer: 76 percent of Harrell's offensive rebounds were contested versus only 61 percent for Van Treese. Harrell scratched and clawed for more of his rebounds, making him attractive to NBA scouts. Ten years ago, before SportVU, that analysis would not have been possible without a considerable time commitment.
Within five years, Warkins hopes to see SportVU running in international basketball and the "big six" NCAA conferences. And Cullen notes that, as time passes, the system will become more useful to his Blue Devils.
"Let's say Jabari Parker had 20 rebound chances in a game and got 15 of them," Cullen said. "But what did Mason Plumlee do the year before? What did Luol Deng do when he played for Duke? What did Elton Brand do before that? The answer is simply that we don't know. We've never tracked it before, so we don't have anyone to compare Jabari to other than opponents and teammates. We stepped into this knowing it would be a couple of years before the data would truly become valuable and referenceable."
Today, SportVU analytics are used mostly off the court, in front offices and team meetings, but Adande sees its ultimate future in the huddle.
"The future is going to be in-game utilization, using this stuff in the heat of the battle," Adande said. "If there's a fourth-quarter timeout, can you get the data in real time to figure out which play to run based on how the game is going? That would be the real triumph of numbers."