How Big Data Is Changing Basketball
With March Madness in full swing we take a look at the latest way coaches are turning to tech and stats to improve play.
Sports statistics were once the realm of the guys who couldn't make the team: note-takers, pencil-pushers and number nerds who watched and recorded every swing, miss and pass of every player throughout the season. Today, high-definition cameras and high-powered data analytics are replacing human statisticians and scouts to give coaches -- and many fantasy league fans -- an edge when it comes to predicting the score of the game and the performance of athletes.
Pro and college teams are using so-called big data and machine learning algorithms that analyze a game by reviewing images of each player's activity that are taken 25 times per second.
"We can extrapolate a ton of information on what they are doing on the floor ... through machine learning processes," said Ryan Warkins, associate vice president at Stats, a Chicago-based firm that owns the SportVU system of cameras in all 29 NBA arenas (LA's Staples Center shares two teams), as well as five college venues.
Six cameras are installed around the perimeter of the court, high up in the arena catwalks where they record picks, post-ups, perimeter shots and layups by each player.
"In the past you would need a human to log in this information themselves," Warkins said. "You can find out efficiencies of teams and players and it's all based on storing that raw information at 25 frames per second."
SportVU's algorithms use the bits of data to help coaches determine the probability of what a certain player will do at the beginning or end of the shot clock, for example, or how players perform when matched up against certain individual opponents. In addition to all 30 pro NBA teams, five college squads -- including Duke and Louisville who are still alive in the NCAA March Madness tournament -- are using SportVU, Warkins said.
Machine learning and predictive algorithms are also at work in a March Madness contest run by San Francisco data analytics startup Kaggle. The idea is to figure out not just who will win at the end, but the probability of how all 64 teams will fare against each of the other teams, said Kaggle data scientist Will Cukierski.
After the weekend of upsets, Cukierski has posted the odds of each of the other teams of beating the number one ranked and tourney favorite Kentucky.
"We are scoring something called the log loss, which is a way to measure the probability vs. the actual outcomes," Cukierski said. "We are asking people for the probabilities of Kentucky beating Duke, not just who is going to win."
The Kaggle contest has drawn 613 entries from 405 people this year. Last year's winners -- Skidmore College math professor Michael Lopez and Loyola University's Gregory Matthews -- used a simpler recipe for success. Instead of crunching mind-numbing amounts of data, they built upon the work of others: combining the Las Vegas betting odds plus the team's performance per possession.
"Sometimes using lots of data won't provide as good predictions as using the best data," Lopez said.
The Las Vegas line is a type of probability that incorporates injuries, the strength of the team's coach, the distance traveled, the relative home field advantage, and other off-court intangibles.
"The first thing we thought to use was the line set by sports books in Las Vegas," Lopez added. "They have large towers in the middle of the desert for a reason. They are very good at setting (betting) lines and can provide a proxy for team strengths."
Playing probability games is fun for sports fans, but for coaches and team executives, the numbers are important for building a winning team, according to SportVU's Warkins.
"You can break down a potential draftee or free agent by their performance," Warkins said. "Say the Chicago Bulls are looking for someone to replace Tony Snell. (They) want a decent perimeter defender and shoots a corner three-pointer, as well as someone who can dribble-penetrate. You can measure this analytically."
As for the future, Warkins sees a time when high-powered data algorithms and miniature cameras will allow fans to watch each game from an individual player's perspective -- like the inside of a NASCAR cockpit.
Nowadays, it's hard to tell the jocks from the geeks. Athletes capitalize on advances from engineering, material science, biomechanics, communication and information technologies to maximize training and performance. And brainiacs develop technologies that are transforming every aspect of sport, including coaching, judging, even the design of sports arenas and spectator experience. To date, most advances in sports technology have been in material science and design. Aerospace engineer Kim B. Blair, founder of the Sports Innovation Group LLC, an affiliate of the Massachusetts Institute of Technology, thinks the playbook is changing. "We've hit a plateau," he says. "The next big thing is the information revolution." These advances will move all developments into the computer era, where everything will be tracked, monitored, optimized, refined and disseminated in ways that athletes can't possibly imagine. Here, in alphabetical order, is our list of 10 technologies that are changing the way sports are practiced, played, scored and watched:
1. Ingestible Computers
Heat exhaustion is the second-leading cause of death in athletes. Until now, core body temperature has been monitored through observation, but athletes can ignore signs of heat exhaustion and trainers may be too far away to make accurate observations. A "thermometer pill" may save lives. Initially developed by NASA and Johns Hopkins University to monitor astronauts from space, the pill contains a quartz crystal sensor and micro-battery wrapped in silicon. Once swallowed, a sensor transmits temperature and heart rate data to the trainer as it travels through the gastrointestinal tract. Athletes in field and track, auto racing, football, hockey, cycling and soccer have used a commercial spin-off based on NASA's version.
2. Wearable Computers
The best sports innovation, ever? Synthetic fibers that wick moisture, dry fast, and are anti-microbial and water- or wind- resistant: sweat-soaked cotton is so '70s. Perhaps the next best thing? "Smart" clothing that uses embedded microscopic sensors and wireless networks to monitor athletes' heart rate, body temperature, hydration and more. Applications extend far beyond the sports arena. Medical and military technicians are developing patient and soldier models to record and transmit real-time biometrics from blood pressure to a bullet wound, from any location.
Science is helping athletes set new records, but we lag far behind the quickest and strongest species on the planet. Increasingly, engineers are turning to nature for inspiration, an approach known as biomimicry. Textured fabrics inspired by dermal denticles or "toothlike" projections found on sharkskin were one of the many innovations that may have helped Michael Phelps and others dress for gold at the 2008 Summer Olympics. In another example, scientists have developed materials that increase in adhesive strength while in motion -- just like the feet of geckos. Coming soon, gecko-inspired nonskid grips and climbing shoes.
4. Carbon Nanotechnology
The secret to a material's strength lies in the properties of the atomic bonds connecting one atom to another. Carbon atoms have extremely strong bonds. Using nanotechnology, scientists manipulate carbon's atomic structure to form hollow, carbon-based tubes that are super small (approximately 100,000 times thinner than a human hair), super light and stronger than steel. Researchers at the University of Texas' Nanotech Institute have developed artificial muscles from carbon nanotubes that contract 30,000 percent per second (human muscles contract around 20 percent per second). They can operate at extreme temperatures, which makes them especially attractive for space applications and is one reason why the Air Force Office of Scientific Research is has teamed up in this area. So far, there are no human applications, but a "smart skin," on an aircraft would have the ability to change appearance in situations of danger.
5. Computational Fluid Dynamics
Thanks to supercomputers, the subfield of physics that focuses on the movement of air, water or gasses called computational fluid dynamics is indispensable to the design of anything that moves -- including cars, oars, bicycles, helmets and swimsuits -- even human athletes. Using 3-D body scanners, computers, visualization and fluid dynamics software, engineers can analyze skin friction. "In the last five years aerodynamic technology has become very prevalent in the development of equipment and clothing for speed-based sports," says Blair. "In competitive cycling, bikers use 90 percent of their power to overcome wind. Even a 5 percent improvement in drag can be the difference between a podium and no podium." Speedo's AQUALAB used computerized scans of hundreds of athletes to pin-point areas of high friction on the athlete's body. With this information, swimsuit designers were able to position low-friction fabric in the right locations to reduce drag.
6. Digital Imaging and Video
It's impossible to imagine a multibillion-dollar global sports industry without television to play (and replay) stunning moments, show us legends in the making -- and generate advertising revenue. And media technology has, in turn, shaped sports. Phil Orlins, senior coordinating producer of ESPN's X Games and Winter X Games, says miniature, wireless and handheld digital cameras that "give viewers unbelievable proximity, put them in the action and take them just about any place so that they can see just about anything" have transformed the sports-viewing experience. Others might argue that these technologies have had a more profound impact, creating a fan base and popularizing new sports and new superstars, seemingly overnight. The United States Olympic Committee and Comcast are partnering to launch an Olympic sports cable channel, while NBC, which won the bid to broadcast the Olympics, has its own 24-hour sports channel, Universal Sports.
7. Information Technologies
Across industries, the trend is toward mobile, rich and instant data. Sports are no exception. Mix radio frequency identification tags, global positioning system devices, remote cameras and broadband networks, and then synchronize and display and what do you get? More information than you'll want to know, guaranteed. Sounds big league, but information technology is infusing sports at all levels. "The sports world is on the cusp of changing into a whole new paradigm because of information technologies," says 94 Fifty's CEO and founder, Mike Crowley. 94Fifty's system captures up to 6,000 pieces of information a second and is designed to be embedded in a moving object like a basketball, soccer ball or hockey puck. Once collected, the data is uploaded to an off-site server and analyzed in seconds. Want to know how your top player is performing compared to last month or year? That's possible. So is comparing one athlete to other athletes her age, across the country. When scores are posted, kids treat it like a video game and become more competitive, reports Crowley. He thinks his system will motivate kids to work on skills they'll need to become great athletes.
8. Reactive Materials
High-speed sports put athletes at risk. Until recently, protective clothing that could absorb impact was often bulky and restrictive. That's changing with the development of materials such as U.K.-based d30 and Dow Corning's Active Protection System materials. Both are made from materials that flex and move with a body in action but immediately harden upon impact. Researchers at the University of Delaware have embedded materials with nanoparticles that become instantly rigid as soon as a kinetic energy threshold is crossed. That can come in handy for athletes, who could seriously hurt themselves if they fell and for people in the military or in law enforcement, who come into bodily harm unexpectedly on the job. And as a benefit, many of these products are washable. Recent applications include gear for downhill skiers and dirt bike racers, as well as ballet shoes, soccer balls, shorts for equestrians, and protective gear for soldiers and law enforcement agents.
Robots offer scientists many benefits. They don't complain, get sick, charge overtime or take vacation. And they can be programmed to repeat the same motion over and over. Commonly used in automobile and other manufacturing settings, robots programmed to simulate sports movements such as tennis or golf swings can help engineers test equipment and surfaces. Robots can even be programmed to sweat. Using robots, researchers can do more tests in less time, under highly controlled settings. A team of researchers from Kanazawa University in Japan has developed an experimental system using a skiing robot to investigate the effects of joint motions on ski turns. Such a system could ultimately serve as a model to help skiers improve their own movements. And a team from the the University of Tokyo has developed two robots, one that can pitch and one that can bat, to study the physics of baseball.
10. Tool-less Manufacturing
Henry Ford's early customers could have any color car, so long as it was black. In the marketplace, the greatest barrier to choice is cost. Elite athletes can drop thousands of dollars on custom-fit equipment, but for most players, it's just a dream. Now, affordable, in-store diagnostics, including 3-D body scanners that analyze body geometry and kinematics, coupled with "tool-less" or direct digital manufacturing in place of molded dies or templates, are making custom-fit a true possibility. University research labs are helping to make the technology a reality. Scientists at Cornell University's College of Human Ecology are using 3D body scanners that image about 300,000 points on the body to develop virtual try-on systems and clothes that can be custom-made on the spot. Caine is guardedly optimistic: "This will happen, but I'm not sure when."